Bail Laws

“Model” Bail Laws:
Re-Drawing the Line Between
Pretrial Release and Detention
Timothy R. Schnacke
4/18/2017
2
Table of Contents
Acknowledgments……………………………………………………………………………………….7
Executive Summary ……………………………………………………………………………………8
Preface ……………………………………………………………………………………………………….9
Introduction ……………………………………………………………………………………………. 10
Part I – General Questions and Answers …………………………………………………. 16
What is Bail? ……………………………………………………………………………………………16
What is the Right to Bail? …………………………………………………………………………18
Why Do We Even Have A Right to Bail? …………………………………………………..20
Do We Have to Take Risks and Expect Failures? ………………………………………24
What Does the History of Bail Tell Us About Re-Drawing the
Line Between Release and Detention? ………………………………………………26
Interference With Release/Detention Leads to Reform …………….26
Secured Money Bonds Interfere With Release/Detention …………28
What Does the Law Tell Us About re-Drawing the Line Between
Release and Detention? ……………………………………………………………………30
How to Do Bail/Release …………………………………………………………..31
How to Do No Bail/Detention ………………………………………………….33
A Particularly Acute Problem ………………………………………..34
Limited By Charge ………………………………………………………..35
Procedural Due Process …………………………………………………38
3
The Purpose of Bail and No Bail ………………………………………………39
What Do Ratios Have to Do With This? …………………………………………..41
What Else Do the Law and History of Bail Tell Us? ………………………….44
The Big Rule …………………………………………………………………………..44
The Big Change (The American Overlay) ……………………………….46
America’s Struggle With Unintentional Detention …………………..48
America’s Struggle With Intentional Detention ……………………….51
Lessons From the Detention Cases ………………………………………….61
The Big Fix …………………………………………………………………………….66
The D.C. Court Reform and Criminal Procedure Act …….68
A Detention Eligibility Net and Further Limiting
Process ………………………………………………………………..72
The Bail Reform Act of 1984 …………………………………………74
A Detention Eligibility Net and Further Limiting
Process ………………………………………………………………..78
United States v. Salerno ………………………………………………….81
What Does the Pretrial Risk Research Tell Us About Re-Drawing
the Line Between Pretrial Release and Detention? ……………………………87
What Do Actuarial Risk Assessment Instruments Tell Us? ……………….89
How Actuarial Pretrial Risk Assessment Instruments
Are Created ……………………………………………………………………………91
4
How Do the Risk Research and Actuarial Pretrial Risk
Assessment Instruments Illuminate Flaws in the Current System? ……96
What Do Actuarial Pretrial Risk Assessment Instruments
Not Tell Us? …………………………………………………………………………………..103
Individual Risk …………………………………………………………………… 104
Detail Concerning “Risk of What” ………………………………………..110
Protective Factors and What to Do With Risk ……………………….114
How Does Risk Research Interact With the Law When it Comes
to Re-Drawing the Line Between Pretrial Release and Detention? …..115
Initial Balancing Issues …………………………………………………………115
Excessive Bail Generally ……………………………………………………….117
Due Process and Equal Protection Generally …………………………118
Fair Notice ……………………………………………………………………………119
Can We Use An Actuarial Pretrial Risk Assessment Instrument
Solely As Our Eligibility Net When We Re-Draw the Line
Between Pretrial Release and Detention? ……………………………………….124
Will Future Actuarial Pretrial Risk Assessment Instruments Be
Sufficient to Function as a Detention Eligibility Net? ………………………126
What Do the National Standards Tell Us About Re-Drawing the
Line Between Pretrial Release and Detention? ……………………………….127
The Detention Eligibility Net …………………………………………………129
The Further Limiting Process ……………………………………………….137
What Have the States Done Up Until Now to Re-Draw the Line
Between Pretrial Release and Detention? ……………………………………….139
5
Do We Have to Eliminate Money at Bail Before We Re-Draw
The Line Between Pretrial Release and Detention? …………………………144
Will We Need To Make Sure We Have Some Resources – Like
Pretrial Services Functions – To Make Everything Work? ……………..145
Will We Have to Change Our Constitutions? …………………………………146
Will We Have to Change Our Statutes/Court Rules? ………………………148
Will We Have to Change Our New Bail Guidelines/
Praxes/ Matrices? ………………………………………………………………………….149
Part II – If We Change, To What Do We Change? …………………………………..153
Analysis Based on General Narrowing Principles ……………………………154
Analysis Based on Current Law ……………………………………………………..157
Excessive Bail ………………………………………………………………………159
Equal Protection ………………………………………………………………….161
Due Process ………………………………………………………………………….162
Test for Punishment ……………………………………………………163
Test for General Substantive Due Process ……………………165
Analysis Based on Threshold Requirements for Predictive Models ….168
Precise Definitions ……………………………………………………………….169
Prediction Validation ……………………………………………………………170
Procedural Safeguards …………………………………………………………171
Part III – A “Model” Release and Detention Process ……………………………….172
6
Articulating Generally Whom to Release and Whom to Detain ……….172
Articulating the Detention Eligibility Net ……………………………………….172
Articulating the Secondary Net ………………………………………………………177
Can We Create a Different Net? …………………………………………..181
Articulating the “Further Narrowing Process” for Detention ………….182
Part IV – Holding Up the Model to the Three Analyses ……………………………185
The Model Held Up to General Narrowing Principles …………………….185
The Model Held Up to the Law ………………………………………………………186
The Model Held Up to the Threshold Requirements ……………………….187
The Model Applied to Difficult Hypotheticals …………………………………187
Part V – The Language of the Model ……………………………………………………….190
Relatively Brief Provision ………………………………………………………………191
More Detailed Provision …………………………………………………………………193
Under This Model, What Will Be Our Ratio of Released to Detained
Defendants? …………………………………………………………………………………..196
What Will My “Failures” Be Under This Model? …………………………..197
Part VI – Essential Elements of Bail Statutes or Court Rules …………………..198
Conclusion ……………………………………………………………………………………………..200
Appendix One …………………………………………………………………………………………201
Appendix Two ………………………………………………………………………………………..202
7
Acknowledgments
As always, I am grateful to far more people than I can list, including all
those invaluable persons to whom I cite throughout the document. Anyone
who has spoken to me over the last two years or who has answered one of
my many questions (I’m thinking about you, John Clark, Joel Bishop, Bo
Zeerip, Spurgeon Kennedy, Tara Blair, and Cherise Burdeen) has formed
part of the analysis found in this paper. In particular, I would like to thank
Lori Eville (NIC), who gave me the time and resources to do the research,
Dr. Michael R. Jones (PJI), who gave me both the idea for the paper as well
as the ultimate answer, Claire Brooker (independent pretrial consultant),
who kept me rooted in pretrial reality, Judge Truman Morrison, III, who
knows more about the theoretical and practical application of fundamental
legal foundations of pretrial justice than anyone in America, and Alec
Karakatsanis (Civil Rights Corps) and Larry Schwartztol (Harvard Law
School Criminal Justice Policy Program), who, by lending their considerable
brains to the cause, are changing the lives of millions of Americans.
This topic is so complex that reasonable persons can (and do) disagree on
many aspects, but each of the people listed above helped me enormously in
coming to my own conclusions about the various issues. Moreover, despite
the significant complexity involved in coming to any set of workable
answers, each of these people also helped me greatly by constantly
reminding me that all release and detention laws must reflect deeply held
American notions of liberty and freedom in addition to natural concerns for
safety and judicial administration. Any errors are my own.
8
Executive Summary
This paper is designed to help persons craft and justify language articulating
who should be released and who should be eligible for detention in a
purposeful in-or-out pretrial system through a study of the history of bail, the
fundamental legal principles, the pretrial research, and the national standards
on pretrial release and detention. It does so, in Part I, by providing the
answers to a series of questions that every jurisdiction should be asking
before embarking on the task of re-drawing the line between pretrial release
and detention. These questions, based on the fundamentals of bail, range
from elementary (i.e., “What is bail?”) to somewhat complicated (i.e., “How
has America traditionally defined ‘flight’ and how did it struggle with both
unintentional and intentional detention for noncapital defendants?”) to very
practical (i.e., “Can we use the results of actuarial pretrial risk assessment
instruments when determining our detention eligibility net?”).
In Part II, the paper begins to answer the question, “If we change, to what do
we change?” It then introduces three analyses that should be used to assess
any proposed model for re-drawing the line between release and detention.
In Part III, the paper proposes a “model” process – this author’s attempt at
purposefully re-drawing the line between release and detention – based on
the history, the law, the pretrial research, and the national standards on
release and detention, and then, in Part IV, the paper holds the proposed
model up to the three analyses. In Part V, the paper operationalizes the
concepts from the proposed model into sample templates designed to
illustrate how a jurisdiction might phrase certain crucial elements contained
in the model. And finally, once this re-drawing of the line between release
and detention is done, Part VI of the paper articulates notions that should be
a part of any state bail legal scheme in order to make the model provision
work. The proposed model can be accepted or rejected by American
jurisdictions. Nevertheless, any different model should be subjected to either
the same or a more rigorous justification process as is provided in this paper.
This paper is likely useful to all persons seeking answers to questions
surrounding pretrial justice today. But it should be especially useful to those
persons who are taking pen to paper to re-write their laws to determine
whom to release and whom to potentially detain pretrial – essentially, to redraw the line between purposeful release and detention.
9
“Model” Bail Laws: Re-Drawing the Line Between
Pretrial Release and Detention
“‘Preventive detention’ is a phrase that could have been coined on
Madison Avenue, and raises associations of rationality and
impartiality and common sense and science.”
Caleb Foote, Comments on Preventive Detention
Preface
For some time now, people in the national pretrial justice movement have
been discussing the idea of writing a document instructing jurisdictions on
changes to their laws during this generation of bail reform. During those
discussions, I had thought that such a document would need to be incredibly
detailed and comprehensive, full of extensive word-choice examples from
current state laws to provide various alternatives in phrasing to reach some
objective. I thought that America could wait for such a document, which
would meticulously describe those choices, and parse and reflect on the
various available alternatives. I was wrong for two reasons.
First, I was wrong because the bail reform movement is moving far too
quickly to wait any longer for guidance. States are changing fundamental
aspects of their release and detention laws, including their constitutional
right to bail provisions. Additionally, states are making comprehensive
changes to their statutes and court rules, and they are doing so rapidly, and,
in many ways, somewhat in the dark.
Second, I was wrong for thinking that most of the current laws from which I
would craft a “model” were legally defensible. The more I worked on this
paper, the more I came to believe that our country’s current bail laws suffer
from numerous and often fatal flaws. Most release provisions fail to ensure
release; most detention provisions are currently unjustified, do not
necessarily reflect those whom society desires to detain, and fail when held
up to the federal model reviewed by the United States Supreme Court in
1987. Indeed, the federal model itself is now likely unjustifiable and has
largely failed to limit detention to constitutional levels. More importantly,
most of the assumptions underlying our current bail laws have now been
shown by the research to be faulty assumptions. Accordingly, this paper is
10
written to provide at least some initial guidance for jurisdictions seeking to
make changes to their bail laws on a clean slate.
For several years, people like me have focused primarily on helping
jurisdictions to change their policies and practices concerning pretrial release
and detention. As jurisdictions began making those changes, however, they
quickly learned that many of their existing laws actually hindered creating
rational, fair, and transparent release and detention systems based on legal
and evidence-based practices. Thus, not so very long ago, we added bail
laws to the list of things that jurisdictions might need to change. Now that
they are making those changes, this paper attempts to answer the inevitable
question, “If we change, to what do we change?”
In no way should this paper be used as justification to slow down the
national bail reform movement. Even though some of these concepts are
complicated, they do not diminish the need for extensive pretrial justice
reform work to appropriately release more defendants pretrial, to infuse
research into the bail system, and to reduce or eliminate secured money
bonds. States can do significant work long before that work begins to require
answers to the questions posed in this paper. Bail reform exists on a
continuum, which is advanced incrementally as states become more
comfortable with more complicated concepts. At some point, however,
states will be forced into the weeds, so to speak, of determining the details of
a rational, fair, and transparent release and detention system. This paper is
designed to cut through those weeds, and, hopefully, to accelerate
completion of this generation of bail reform.
Introduction
In the movie, The Big Short, Michael Burry (who famously shorted the
American subprime mortgage market in 2007 when he predicted that a
housing bubble would burst), is talking to Lawrence Fields (his boss at the
hedge fund), and tells Fields that he sees the housing bubble and its eventual
crash. Fields says, “No one can see a bubble. That’s what makes it a
bubble.” And Burry replies, “That’s dumb, Lawrence. There’s always
markers.” The same is true in bail reform. There are always markers
showing the need for bail reform, and when these markers exist, bail reform
becomes inevitable.
11
We are currently witnessing the inevitability of the so-called “third
generation of bail reform” in America. And if you look at the previous two
generations – indeed, if you look at all instances of bail reform over the
centuries in both America and England – you will see that the same markers
leading to changes in those eras are leading to changes we see today. Those
markers, which include game-changing pretrial research, a meeting of minds
over the need for reform, and, most importantly, interference with our
underlying notions of both release and detention, tell us that bail reform is
not merely some fleeting and quickly dissipating trend among just a few
states. No, bail reform is unavoidable and will happen in every state in
America. If it does not come from states desiring to change on their own, it
will come from states being forced to change by the courts, which are
increasingly requiring jurisdictions to follow fundamental legal principles
and to justify their release and detention schemes.
The certainty of bail reform is made more consequential when one learns
that the only way to move through this generation of reform (and, indeed,
the only way to avoid future generations of reform) is to re-think and rearticulate answers to the three foundational questions concerning pretrial
systems across America, which are “whom should we release,” “whom
should we detain,” and “how should we do it?” Right now, most
jurisdictions are releasing the wrong persons, and the essence of bail reform
in this generation is a move to rational, fair, and transparent systems that
purposefully put the right persons in the right places pretrial. To do that,
jurisdictions must examine the very cornerstones of their bail laws – the
foundational pillars that express who is eligible for release and who is
eligible for detention. All states have such expressions, yet virtually all states
have been ignoring those expressions for decades. Among other things, this
generation of reform is forcing states to move from systems in which money
determines release and detention randomly to systems in which judges make
intentional and un-hindered release and detention decisions. To do this, the
states must, in the first instance, set out clearly who should be released and
who should be eligible for detention based on American notions of freedom
and liberty.
So-called “model” bail laws are somewhat simple. Once a jurisdiction has
decided whom to release and whom to detain, model laws simply make this
happen by using legal and evidence-based practices to achieve the
constitutionally valid purposes of bail (release) and no bail (detention). But
re-drawing that initial line between release and detention – or, more
12
importantly, designing a fair and rational process that ultimately leads to
particular lines being drawn – is deceptively complex. The process requires
knowing something about the history of bail in both America and England,
its fundamental legal principles, and the pretrial research, especially the
research on risk and risk assessment. It requires understanding why we even
have a thing called pretrial release to begin with, and it requires knowing
both the positive and potentially negative consequences of attempting to
replace our current charge-based systems with systems that identify
defendants in varying degrees of risk based on actuarial pretrial risk
assessment instruments. In this generation of bail reform, jurisdictions are
tempted to completely replace their charge-based schemes with so-called
risk based ones – to dump risk into their charge-based constitutions and
statutes – but this paper explains why that temptation is overly simplistic.
For example, a positive consequence of moving toward a risk-based or riskinformed release and detention process is that it can bring us ever closer to
knowing the answers to the two essential questions for determining whom to
release and whom to detain: (1) “How risky is this person?”; and (2) “Risky
for what?” These questions have been the two primary questions asked ever
since a thing called bail or pretrial release was created. If you know that a
person is high risk to commit any crime while on pretrial release, you can
begin to purposefully respond to that risk. But that is only half of the
inquiry, because that same person may be high risk for committing only a
relatively low level offense, such as trespassing or loitering. Likewise, there
is value in knowing that, although a person is labeled as “low” risk, his or
her failure might lead (albeit perhaps rarely) to a violent crime. Actuarial
pretrial risk assessment instruments are getting very good at telling us the
answer to question number one, and are only now beginning to tell us the
answer to question number two. In the future, the answers to both questions
will only become clearer.
Nevertheless, a potentially negative consequence of using actuarial pretrial
risk assessment instruments is that we can actually make things worse by
essentially labeling all defendants as “risky,” something that did not happen
in the past without jumping over certain mental hurdles. In the past,
defendants were charged with crimes, and American jurisdictions viewed
those crimes as proxies for risk and based entire release and detention
systems on assigning money to those charged crimes, which did not always
result in the intended in-or-out placement. In a traditional money bail system
based on charge, it was not so much the defendant who was risky, it was the
13
charge, and the charge determined the price to gain freedom. The system
was based on certain flawed assumptions (such as that high levels of crimes
equaled high levels of risk to commit the same or a similar offense while on
bail), but those assumptions broadly kept judicial officials from thinking that
all defendants are risky. Today, empirical risk assessment – one of the
hallmarks of this generation of reform – places the emphasis on the
defendant, and no defendant is immune from its label of risk. In this new
world of risk assessment, one can literally go out on the streets of any city
and show that 100 random pedestrians are all risky at varying levels – from
extremely low to extremely high – before they have even been accused of
committing any offense. This begs the question: since everybody is already
risky, at what point do we begin assessing it? Where do we draw the line?
Labeling citizens as risky reminds us of Harvard Law Professor Laurence
Tribe’s articulated fear of allowing the government to incarcerate people for
something they may or may not do in the future. Tribe wrote: “Throughout
history, governments have been tempted to establish order by identifying
and imprisoning in advance all likely troublemakers.”1 Quite simply,
empirical risk assessment helps with identifying and smoothing the resulting
detention of persons once they have caught the attention of the government
for any reason. Moreover, according to Tribe, the potential harm generated
by a system of detaining so-called risky individuals is exacerbated by the
built-in bias toward ever more detention. “The pretrial misconduct of
[released] persons will seem to validate, and will indeed augment, the fear
and insecurity that the system is calculated to appease,” he wrote. “But when
the system detains persons who could safely have been released, its errors
will be invisible.”2
In other words, “the degree to which judges wrongfully
detain defendants is unknowable because their decisions are
‘unfalsifiable.’”3Accordingly, despite America’s gradual shift toward a more
preventive justice system over the last several decades, fundamental
American legal principles must be upheld as constant reminders to limit

1 Laurence H. Tribe, An Ounce of Prevention: Preventive Justice in the World of John Mitchell, 56 Va. L.
Rev. 371, 376 (1970) [hereinafter Tribe].
2
Id. at 375.
3
Jeffrey Fagan & Martin Guggenheim, Preventive Detention and the Judicial Prediction of Dangerousness
for Juveniles: A Natural Experiment, 86 J. of Crim. L. and Criminology 415, 428 (1996) [hereinafter Fagan
& Guggenheim] (quoting John S. Goldkamp, Danger and Detention: A Second Generation of Bail Reform,
76 J. Crim. L. & Criminology 1, 28 (1985)) [hereinafter Goldkamp]; see also Harvard Law School
Criminal Justice Policy Program, Moving Beyond Money: A Primer on Bail Reform, (2016) at 19 (“[I]f . . .
almost all ‘high risk’ defendants are detained, it becomes impossible to test whether individuals who
receive that designation actually have high rates of pretrial failure.”) [hereinafter Harvard Law School
Primer].
14
pretrial detention and to embrace the risk inherent in pretrial release, just as
we have chosen to accept some risk of crime in the first instance by relying
on the moral deterrence of clearly articulated laws to govern human
behavior.
The move from a charge and money-based system to one informed by risk
and using little or no money thus requires extensive safeguards designed to
assure Americans that things will not be made worse through more
purposeful practices that nonetheless lead to over-detention. These
safeguards are best placed in our state constitutions – documents that, unlike
statutes, “ought to state rules not for the passing hour, but principles for an
expanding future.”4 But wherever they exist, it is imperative that they do,
indeed, exist. At the very least, these safeguards, which are rooted in
fundamental notions of American liberty, require any changes to our bail
laws to have justification in the history, law, or research. They require us to
create limits or floors, crimes for which pretrial detention is never allowed,
and for which risk assessment is neither sought nor used. And they caution
us to ensure that infusing the concept of actuarial risk does not diminish
traditional legal principles creating and molding the right to release and
requiring that pretrial liberty be “the norm.”
5
This document is designed to help jurisdictions craft language to determine
who should be released and who should be eligible for detention so that
traditional legal principles at bail are not eroded or erased. It does so in Part
I by providing the answers to a series of questions that every jurisdiction
should be asking before embarking on the task of re-drawing the line
between release and detention. These questions, based on the history of bail,
the law, the pretrial research, and the national standards on release and
detention range from elementary (i.e., “What is bail?”) to somewhat
complicated (i.e., “How has America traditionally defined ‘flight’ and how
did it struggle with both unintentional and intentional detention for
noncapital defendants?”) to very practical (i.e., “Can we use the results of
actuarial pretrial risk assessment instruments when determining our
detention eligibility net?”).

4 Benjamin N. Cardozo, The Nature of the Judicial Process, at 83 (Yale Univ. Press 1949). Forty-one states
have right to bail provisions in their constitutions. States without such constitutional provisions still attempt
to articulate who should be released and who should be detained in their statutes, and these states should
similarly understand a need to guard certain fundamental American principles of freedom and liberty from
slow but significant erosion.
5
See United States v. Salerno, 481 U.S. 739, 755 (1987) (“In our society, liberty is the norm, and detention
prior to trial or without trial is the carefully limited exception.”).
15
In Part II, the paper begins to answer the question, “If we change, to what do
we change?” It then introduces three analyses that will be used to assess any
proposed model for re-drawing the line between release and detention. In
Part III, the paper proposes a “model” process – this author’s attempt at
purposefully re-drawing the line between release and detention – based on
the history, the law, the pretrial research, and the national standards on
release and detention, and then, in Part IV, the paper holds that proposed
model up to the three analyses. In Part V, the paper operationalizes the
concepts from the proposed model into sample templates designed to
illustrate how a jurisdiction might phrase certain crucial elements contained
in the model. And finally, once this re-drawing of the line between release
and detention is done, Part VI of the paper articulates notions that should be
a part of any state bail legal scheme in order to make the model provision
work. The proposed model can be accepted or rejected by American
jurisdictions. Nevertheless, any different model should be subjected to the
same or a more rigorous justification process as is provided in this paper.
States reading this document will likely arrive at a few basic conclusions.
First, the idea of simply moving from a charge-based to a risk-based system
of release and detention is deceptively complex. Second, the history, the law,
the research, and the national standards nonetheless point to an answer for
how charge and risk can co-exist. Third, and most importantly, this answer is
one that can make this the first generation of bail reform in America that
ultimately succeeds. The solution is not easy, and requires certain prerequisites (such as the elimination of money’s ability to detain and some
likely extremely minimal level of re-allocated resources). Nevertheless, it is
a solution with the potential not only to complete this generation of reform,
but also to largely eliminate the need for reform in the future.
16
Part I – General Questions and Answers Concerning
Pretrial Release and Detention
What is Bail?
The principle motive for writing the National Institute of Corrections’
document titled, Fundamentals of Bail,
6 was to arrive at an accurate
definition of “bail” and to articulate a universally true purpose of bail.
Knowing the proper definitions of terms and phrases at bail is fundamental
to American bail reform, and yet, across America, states have struggled with
varying definitions leading to improper statements of purpose. These
variations and improper statements, in turn, have caused confusion within
the national pretrial justice movement. This confusion can be erased,
however, simply by studying the other fundamentals of bail – the history, the
legal foundations, the pretrial research, and the national standards. Indeed,
when one studies those other fundamentals, one quickly learns bail’s true
definition and purpose: bail is a process of conditional release and the
purpose of bail is to provide a mechanism for release, just as “no bail” is a
process of detention with a purpose to provide a mechanism for potential
pretrial detention. The Fundamentals paper sums up its justification for
defining bail as a process of release as follows:
Legally, bail as a process of release is the only definition that:
(1) effectuates American notions of liberty from even colonial
times; (2) acknowledges the rationales for state deviations from
more stringent English laws in crafting their constitutions (and
the federal government in crafting the Northwest Territory
Ordinance of 1787); and (3) naturally follows from various
statements equating bail with release from the United States
Supreme Court from United States v. Barber and Hudson v.
Parker, to Stack v. Boyle and United States v. Salerno.
Bail as a process of release accords not only with history and
the law, but also with scholar’s definitions (in 1793, Anthony
Highmore defined bail as ‘the means of giving liberty to a
prisoner,’ and in 1927, Arthur Beeley defined bail as the release
of a person from custody), the federal government’s usage

6
See Timothy R. Schnacke, Fundamentals of Bail: A Resource Guide for Pretrial Practitioners and a
Framework for American Pretrial Reform (NIC, 2014) [hereinafter NIC Fundamentals].
17
(calling bail a process in at least one document), and use by
organizations such as the American Bar Association, which has
quoted Black’s Law Dictionary’s definition of bail as a ‘process
by which a person is released from custody.’ States with older
(and likely outdated) bail statutes often still equate bail with
money, but many states with newer provisions, such as Virginia
(which defines bail as ‘the pretrial release of a person from
custody upon those terms and conditions specified by order of
an appropriate judicial officer’), Colorado (which defines bail
as security like a pledge or a promise, which can include release
without money), and Florida (which defines bail to include ‘any
and all forms of pretrial release’) have enacted statutory
definitions to recognize bail as something more than simply
money. Moreover, some states, such as Alaska, Florida,
Connecticut, and Wisconsin, have constitutions explicitly
incorporating the word ‘release’ into their right to bail
provisions.7
Most relevant to this paper, however, is that bail defined as a process of
release is the only definition that allows jurisdictions to re-articulate their
release and detention processes without confusion. As noted above, most of
that confusion comes from the fact that many people (indeed, many courts
and legislatures) define bail by one of its conditions – money. And although
defining bail as money is understandable based on the fact that promising to
pay money was the sole means of effectuating release with a goal of court
appearance for nearly all of bail’s 1,500 year history, bail is not money.
Quite simply, money is a condition of bail with a different purpose.
Defining bail as money causes confusion especially when jurisdictions are
confronted with bail’s history (which tends to define it as release), the law
(with multiple court cases, including U.S. Supreme Court cases, describing
bail as a process or procedure of release), and even a state’s own right to bail
provision (when a state defines bail as money but nonetheless has a “right to
bail” in its laws, it is inherently confusing to speak of reducing or
eliminating money). Because understanding the history, the law, and the
research is crucial to both understanding current law as well as to re-drawing
the line between pretrial release and detention, it is important that
jurisdictions understand that these three areas of knowledge point to a

7
Id. at 113.
18
universally true definition of bail: once again, bail is a process of conditional
release, and the purpose of bail – the reason we have it – is to provide a
mechanism for conditional release. When people speak of eliminating
money at bail, it in no way erodes one’s right to bail; indeed, eliminating
money at bail would actually give meaning to the right to release.
Jurisdictions should not be alarmed when their own laws define bail as
money, but knowing how their definition differs from the legal and historical
definition will only help them when they seek to improve their laws.8
Accordingly, when re-drawing the line between pretrial release and
detention, jurisdictions must remember that bail means release, and that
pretrial release is embedded in our American system of justice.
What is the Right to Bail?
It follows, then, that the right to bail, whenever it is articulated, should be
read to mean a right to release. This is true from a study of the history, as
historical documents repeatedly refer to the right to bail as a right to release,
and of the law, as even the United States Supreme Court has equated the
“right to bail” with “the right to release before trial,” and “the right to
freedom before conviction.”9
This is difficult to understand today only
because the right to actual release in America has been eroded to the point
where people simply do not think it exists. Nevertheless, ever since the
Norman Invasion, in both England and America, calling persons bailable
always meant that they were to be released. Indeed, keeping a bailable
defendant in jail interferes with one of our underlying notions of release,
which, as alluded to previously, is a marker of bail reform.10
In the 1960s, the federal government (and possibly some states) began
slowly to recognize the need for more precise terminology. The Act of 1789
spoke of “admitting” all defendants to bail except for those charged with
capital offenses.11 Later articulations of the right, as note by the Supreme

8
For example, after an approximately 18 month comprehensive study of “bail” and “no bail” in Colorado,
legislators changed the statutory definition of bail, which had defined bail as an amount of money, to better
reflect legal and historical principles of release and freedom.
9
Stack v. Boyle, 342 U.S. 1, 4 (1951). Likewise, the Court in United States v. Salerno has noted that
“liberty” – a state obtained only through release – is the essence of the right. See 481 U.S. 739, 755 (1987).
10 See generally, NIC Fundamentals, supra note 6, passim.
11 See The Judiciary Act of 1789 (“An Act to Establish the Judicial Courts of the United States”), 1 Stat. 73.
The Judiciary Act provided a detailed organization of the federal judiciary that the constitution had
sketched only in general terms. Section 33 of the Act read: “And upon all arrests in criminal cases, bail
shall be admitted, except where the punishment may be death, in which cases it shall not be admitted but by
19
Court in Stack v. Boyle, still spoke of a noncapital defendant’s right to be
“admitted to bail” prior to conviction.12 In 1966, however, the Bail Reform
Act began a trend toward gradually using the word “release” when
discussing the right involved,13 and by 1984 Congress explained that it had
replaced the word “bail” with “release” throughout the Bail Reform Act of
that year to avoid what had become apparent confusion over the use of the
term “bail” in a system of various modes of release, including release on
recognizance or non-financial conditions.14
America’s erosion of its understanding that bail is or should be release
means that people often incorrectly state that the right to bail is a right only
to have one’s conditions set, a statement that merely reflects the poor state of
bail practice in America today and that runs counter to the law and the
history of bail. To articulate that only a certain group is eligible for
detention, but then to allow for a sizeable number of defendants outside of
that group to be detained in fact – whether intentionally or unintentionally
because they cannot meet the conditions of release – is likely unlawful, and
should at least be considered an egregious aberration to legal and historical
notions surrounding pretrial release and detention.
For purposes of this paper, however, the reader should remember that we are
discussing jurisdictions purposefully and justifiably re-articulating which
defendants should be released and which should be detained. We are
discussing jurisdictions reformulating that determination in a generation of
reform interested in individual risk, and then enacting provisions to ensure
the immediate effectuation of the purposeful decision primarily by using
legal and evidence-based practices and by eliminating barriers to either
release or lawful detention. Thus, even though it is important to equate bail
with release when doing “bail reform” – and especially when consulting
legal or historical documents to guide any reform efforts – it is best for
jurisdictions to refrain from using the somewhat confusing terms “bail” and
“no bail” and to simply refer to release and detention. As will be shown
later, once a state has articulated its detention eligibility net, it is safe to then

the supreme or a circuit court, or by a justice of the supreme court, or a judge of a district court, who shall
exercise their discretion therein, regarding the nature and circumstances of the offense, and of the evidence,
and the usages of law.”
12 See Stack v. Boyle, 342 U.S. 1, 16-17 (1951).
13See Bail Reform Act of 1966, Pub. L. 89-465, 80 Stat, 214 (1966) [hereinafter 1966 Act]. For example, in
Section 5(a) the Act read, “The first sentence of section 3041 of title 18, United States Code, is amended by
striking out “or bailed” and inserting in lieu thereof “or released as provided in chapter 207 of this title.”
14 S. Rep. No. 98-225, at 4 (1984).
20
say that the remainder of defendants (as well as many within the net) should
enjoy a right to actual release. If done properly, the right to release pretrial
can be made meaningful in America once again.
Accordingly, when re-drawing the line between pretrial release and
detention, jurisdictions must remember that the right to bail in America has
always been intended to mean a right to actual release.
Why Do We Even Have A Right to Bail, or Release? What
Keeps Us From Simply Detaining Everyone Prior to Trial?
The answer to this question comes partly from history and tradition, and
partly from the law. Historically, even before the Statute of Westminster in
1275, persons facing criminal charges were separated out as either “bailable”
or “unbailable” based on custom.15 The Statute of Westminster codified that
tradition, and expressly articulated that those defendants deemed “bailable”
had to be released, just as those defendants deemed “unbailable” had to be
detained. The reasons for release in those times were not necessarily the
reasons we cite today. For example, release to personal sureties was often
desirable in thirteenth century England due to the lack of adequate jails, and
the process of suretyship was designed to continue to exert control over a
defendant beyond incarceration. It was later in America that the right to
release began finding its foundation on concepts of liberty and freedom.
In the centuries between 1275 and the 1700s, any efforts on the part of
government officials to detain otherwise bailable defendants led to reform.
For example, a stated purpose for the creation of habeas corpus in 1679 –
often called the “Great Writ” in America to reflect its importance – was to
provide a remedy to defendants who were “detained in Prison, in such cases
where by Law they were bailable.”16 The Excessive Bail Clause, when
enacted in England, was in response to judicial officials setting the financial
condition in amounts leading to the de facto denial of bail, or release, as a
way of avoiding the provisions of habeas corpus.17

15 See generally NIC Fundamentals, supra note 6, at 33-56.
16 See June Carbone, Seeing Through the Emperor’s New Clothes: Rediscovery of Basic Principles in the
Administration of Bail, 34 Syr. L. Rev. 517, 528 n. 53 [hereinafter Carbone] (quoting The Habeas Corpus
Act, 31 Car. 2, c. 2 (1679)).
17 See id. at 528-28; William F. Duker, The Right to Bail: A Historical Inquiry, 42 Alb. L. Rev. 33, 66
(1977-78) [hereinafter Duker]; W. Holdsworth, A History of English Law, Vol. 9, 118-19 (1965)
[hereinafter Holdsworth], found at https://archive.org/details/historyofenglish09holduoft.
21
The tradition of calling persons either “bailable” or “unbailable,” and then
making sure that bailable defendants obtained actual release, followed into
the Colonies, and quickly became Americanized in three ways: (1) the
purposes for release became more associated with liberty and freedom; (2)
the right to release was gradually expanded to virtually all defendants; and
(3) likely most importantly and discussed in greater detail later in this paper,
the right was bestowed upon defendants before looking at any of the
traditional English factors – such as the weight of the evidence or criminal
history – used to determine bailablility in that country.18
In sum, having a mechanism for pretrial release in a justice system
resembling our own was a part of English tradition for centuries. It was built
upon and molded by other monumental jurisprudential phenomena and
documents, such as the Magna Carta, habeas corpus, and due process, and it
was adopted by America but expanded to better reflect purely American
notions of criminal justice. Indeed, in a comprehensive article on the right to
bail, author Matthew J. Hegreness uses both tradition (including America’s
long tradition, until only recently, of upholding the right to release) and the
law to, among other things, argue the existence of a broad “consensus right
to bail” that exists in both federal and state law even when unarticulated, and
despite even significant recent erosion.19
The law, too, has grown to foster this tradition of pretrial release, to
articulate it, and, until very recently, to protect it. Citing descriptions of the
bail process from William Blackstone, whose Commentaries on the Laws of
England influenced our Founding Fathers as well as the entire judicial
system and legal community, author F.E. Devine wrote that denying the
release of bailable defendants during the American colonial period was itself
considered to be a crime.20 Moreover, maintaining the process of bail as a
mechanism of release was mentioned in Supreme Court opinions of the
nineteenth and twentieth centuries, with perhaps the following as the best
known expression of our continued protection of a right to release:
The practice of admission to bail, as it has evolved in AngloAmerican law, is not a device for keeping persons in jail upon

18 See Carbone, supra note 16, at 529-548.
19 See Matthew J. Hegreness, America’s Fundamental and Vanishing Right to Bail, 55 Ariz. L. Rev. 909
(2013).
20 F.E. Devine, Commercial Bail Bonding: A Comparison of Common Law Alternatives, at 4 [hereinafter
Devine] (citing William Blackstone, Commentaries on the Laws of England, at pp. 291, 295-97, Chitty ed.
(Philadelphia: J.P. Lippincott, 1857) (Praeger Publishers, 1991)).
22
mere accusation until it is found convenient to give them a trial.
On the contrary, the spirit of the procedure is to enable them to
stay out of jail until a trial has found them guilty. Without this
conditional privilege, even those wrongly accused are punished
by a period of imprisonment while awaiting trial, and are
handicapped in consulting counsel, searching for evidence and
witnesses, and preparing a defense. To open a way of escape
from this handicap and possible injustice, Congress commands
allowance of bail for one under charge of any offense not
punishable by death . . . providing: ‘A person arrested for an
offense not punishable by death shall be admitted to
bail’ . . . before conviction.21
In the 1895 case of Coffin v. United States, the United States Supreme Court
wrote that the presumption of innocence – the principle that says, for the
most part, that defendants do not have to prove their own innocence – is
“axiomatic and elementary” to the administration of our criminal laws.22
This language might be mere surplusage to the current discussion were it not
for the fact that the Supreme Court has cited the presumption of innocence
as a primary reason that we have a right to release to begin with. In Stack v.
Boyle, the Court wrote:
From the passage of the Judiciary Act of 1789, to the present
Federal Rules of Criminal Procedure, federal law has
unequivocally provided that a person arrested for a noncapital
offense shall be admitted to bail. This traditional right to
freedom before conviction permits the unhampered preparation
of a defense, and serves to prevent the infliction of punishment
prior to conviction. Unless this right to bail before trial is
preserved, the presumption of innocence, secured only after
centuries of struggle, would lose its meaning.23
Although there is unwarranted confusion over the presumption of innocence
at bail,24 one concept should be clear: when it comes to the right to bail –
i.e., the right to freedom before conviction – one reason that we even have it

21 Stack v. Boyle, 342 U.S. 1, 7-8 (1951) (Jackson, J. concurring).
22 Coffin v. United States, 156 U.S. 432, 453 (1895).
23 Stack, 342 U.S. at 4 (internal citations omitted).
24 See The Presumption of Innocence at Bail, found at http://www.clebp.org/images/10-19-
2016_presumption_of_innocence_and_bail.pdf.
23
is to maintain the presumption of innocence, a principle axiomatic and
elementary to our system of justice. This notion is evident in Justice
Rehnquist’s opinion in United States v. Salerno, in which even while
upholding a federal bail scheme allowing for increased detention of criminal
defendants for the purpose of public safety in addition to court appearance,
the Justice nonetheless wrote that “liberty” – a state necessarily obtained
from actual release – is the American “norm.”25
In Coffin, the Supreme Court wrote that it was precisely the fact that the
presumption of innocence was so elementary and universal that instances of
neglecting it in particular cases were rare.
26 The same is true of the right to
bail or release, where we simply do not see states attempting intentionally to
eliminate or limit the right to only a few cases (instead, the right has eroded
primarily through “unintentional detention,” discussed infra). It may be the
law or tradition in other countries not to provide some mechanism for
pretrial release, but not in America. Thus, the issue today is not whether we
have the right, because clearly we do. The issues, instead, are twofold: (1)
determining who, exactly, can and should be given a right to release within
constitutional boundaries; and (2) once given, how to make sure that the
right remains meaningful.
Jurisdictions should look at the right to bail or release as an offshoot of the
American principle of allowing clearly articulated laws to govern free
persons in a criminal justice system. Articulating those laws means that we
leave it to persons to reject those laws at their peril, and yet to remain free
from government interference so long as they are followed. Because persons
charged with crimes maintain many, if not virtually all of their constitutional
rights, the vast majority of these persons were (and are) meant to remain free
until the law has formally adjudged their guilt.
Accordingly, when re-drawing the line between pretrial release and
detention, jurisdictions must remember that we have a right to bail or release
because it is fundamental to other important American notions concerning
liberty and freedom. We have the right due to tradition and history, but also
due to the law. Unlike post-conviction release into the community through
legal mechanisms such as probation, which states could eliminate altogether,
a right to pretrial release in America can never be abolished.

25 United States v. Salerno, 481 U.S. 739, 755 (1987) (“In our society, liberty is the norm, and detention
prior to trial or without trial is the carefully limited exception.”).
26 See Coffin, 156 U.S. at 457-58.
24
Does Having A Right To Release Pretrial Mean That We Have
to Take Risks and Expect Some Failures?
Though not easily understood by many, to be an American enjoying the
presumption of innocence and the right to release before trial means that we
must absorb some amount of risk and thus to expect some failure pretrial.
The entire American criminal justice system is based, in large part, on taking
risks. In America, rather than giving the government unlimited powers to
protect the public, we have, instead, only allowed limited government
intervention by using “the moral and deterrent effect of laws which define
particular acts as criminal and which punish all who violate their
proscriptions.”27 Allowing only limited government intervention inevitably
means that American society takes risks each day and that we can most
certainly expect failure. Nevertheless, we choose this system to maintain our
basic American freedoms as articulated in our founding documents. This is
important to remember: our fundamental notions of what it means to be
Americans mean that in the substantive criminal law we have to take risks
on people that can lead sometimes to even catastrophic failure. To enact a
system designed to eliminate that risk would undoubtedly alter and erode our
American identity.
The same is true in bail or pretrial release. While tracing the roots to the
presumption of innocence from Greek and Roman times, the Supreme Court
connected that principle to what is known as Blackstone’s ratio, which is the
maxim that “the law holds that it is better that ten guilty persons escape than
that one innocent suffer.”28 This articulation of a ratio is a quintessential
statement about accepting risk, and in the context of the criminal law it says
that we must embrace the risk inherent in freedom, and that the government
may not do things to people “just to make sure” that we capture all
wrongdoers. And thus it is fitting that in his concurring opinion in Stack v.
Boyle – the case articulating a right to release in order to uphold the
presumption of innocence – Justice Douglas wrote:
Admission to bail always involves a risk that the accused will
take flight. That is a calculated risk which the law takes as the

27 Tribe, supra note 1, at 376.
28 Coffin v. United States, 156 U.S. 432, 456 (1895) (quoting 2 Bl. Com. c. 27, margin p. 358); see also
What Do Ratios Have to Do With This?, infra.
25
price of our system of justice. We know that Congress
anticipated that bail would enable some escapes, because it
provided a procedure for dealing with them.29
When he wrote this language, the only constitutionally valid purpose for
limiting pretrial freedom was court appearance. Today, Justice Douglas
would doubtless write that risk of new criminal activity, too, is a calculated
risk taken at bail to protect our system of justice. Too often in this generation
of bail reform we have focused on assessing and managing risk of flight and
danger versus embracing the risk of release, and we have seen entities
claiming that “risk assessment” will be the salvation to our pretrial crisis.
But the fact is that we have had risk assessment ever since 400 A.D.; today’s
methods of assessing certain risks are simply superior to anything done
previously.
Instead, the true solution to our pretrial crisis is to understand that risk is
inherent in bail, and thus that we cannot, consistent with fundamental
American principles, be risk averse. Risk assessment and management are
important, but less so than the need to embrace the risk inherent in releasing
people pretrial to begin with. For the same reasons that government agents
do not roam the streets seeking to assess and detain those whom we think
might violate some substantive criminal law, we also do not maintain a
similar system within bail, but instead rely mostly on deterrence and the
threat of sanctions so as not to unduly interfere with pretrial freedom.
Because pretrial release comes after an arrest, we may certainly limit
freedom more than if no arrest occurred, but never to the point of complete
assurance of public safety or court appearance. Thus, when re-drawing the
line between pretrial release and detention, jurisdictions must remember to
accept some level of risk as they craft any language articulating the right to
release.

29 Stack v. Boyle, 342 U.S. 1, 8 (1951).
26
What Does the History of Bail Tell Us About Re-Drawing the
Line Between Pretrial Release and Detention?
Interference With Release or Detention Leads to Bail Reform
The history of bail contains many lessons for those seeking to re-draw the
line between pretrial release and detention in America. But perhaps the most
important lesson is the historical principle, mentioned previously, that
whenever something interferes with our notions of release or detention, bail
reform happens. Thus, if we see “bailable” or releasable defendants – or
even people who we think should be bailable or releasable – who are in jail,
bail reform happens. Likewise, anytime we see unbailable or detainable
defendants – or even people who we think should be unbailable or
detainable – who are not in jail, bail reform happens. In sum, whenever we
think that the wrong people are either in or out of jail pretrial, bail reform
happens. Today, many Americans believe that there are people in jail who
should not be in jail, and that there are at least some people outside of jail
who, perhaps, should be inside. This is a classic recipe for bail reform.
Indeed, it is arguably the first time since 1274 that both “bail” and “no bail”
are viewed as simultaneously needing reform.
By looking at English bail in 1274, we can see the parallels.30 In that year,
King Edward I obtained information showing that persons who were
customarily bailable were being held in jail while people who were
customarily unbailable were being released (in both cases, in return for
money to the custodian of the jail). That information led to the enactment of
the Statute of Westminster of 1275, which delineated bailable and unbailable
offenses, and which made it a crime for sheriffs to detain bailable defendants
or to release unbailable ones. Over the centuries, there would occasionally
be reforms designed to keep unbailable defendants in jail (or to add certain
classes of defendants to the “no bail” process), but historically most reforms
dealt with attempting to get bailable defendants out of jail. This notion, that
bail reform happens whenever we see bailable defendants in jail, has led to
such monumental jurisprudential improvements such as a right to release
from unlawful confinement through habeas corpus, the prohibition of
detention without a formal charge, and a right to non-excessive bail. And the

30 See generally, NIC Fundamentals, supra note 6 and resources cited therein; see also, Timothy R.
Schnacke, Money as a Criminal Justice Stakeholder: The Judge’s Decision to Release or Detain a
Defendant Pretrial (NIC, 2014) [hereinafter NIC Money].
27
notion continues today, with court opinions, legislative changes, and
executive actions all designed to foster the release of bailable defendants.
Nevertheless, bail reform today is happening to both bail and no bail, release
and detention. Bailable defendants are currently being detained, and certain
unbailable defendants whom we believe should be detained pretrial are
being released. Moreover, the cause of this overall pretrial dilemma is the
use of secured money bonds. In short, creating a layer requiring the payment
of money up front as a prerequisite to release, without more, will
automatically lead to the wrong people both staying in and getting out of
jail. The answer to this dilemma, though, is not as simple as making sure
bailable defendants are released and unbailable ones are detained, because in
many cases our labels are wrong. Moreover, as will be shown later, the
answer is also not as simple as using actuarial pretrial risk assessment
instruments alone to release all “low risk” defendants and to detain all “high
risk” ones. In fact, the answer lies somewhere in the middle.
Overall, many Americans currently believe that we have the wrong
defendants in the wrong places, and this problem is made most evident
through empirical risk assessment. Today in America, defendants who are
assessed to be “low” and “medium” risk (and many who are called “high”
risk) and who could safely be managed pretrial outside of secure detention
are being held in jail, while some extremely high risk persons (including
defendants who are actually high risk to commit a serious or violent crime
but who nonetheless have scored low on an actuarial assessment), are being
released. Bail reform in this instance, then, means delving deeply into our
definitions and foundational principles to answer those three overriding
questions – whom do we release, whom do we detain, and how do we do it?
– so that we can, once again, make purposeful release and detention
decisions based on the law and the research.
In sum, the history tells us that “bail” equals release and that “no bail”
equals detention, and that if anything interferes with these two concepts, bail
reform happens. In addition, the history tells us that the cause of this
interference today is our use and misuse of secured money bail. This is
monumentally important to remember, and so it requires a brief additional
explanation.31

31 See generally id, and sources and resources cited therein.
28
Secured Money Bonds Interfere With Release and Detention
Pretrial detention in England and America has always been done the same
way – by keeping a defendant in jail. But pretrial release is dramatically
different today, and especially in America. To make sure that bail equaled
release in England, that country used the so-called “personal surety system,”
which relied on people to be willing to watch over a released defendant for
no money and no promise of indemnification even in the event of a default.
Indeed, money in the bail process was only found in what we call today the
defendant’s financial condition of release. That condition (in the form of
property and then later, money) had been a part of English bail since roughly
400 A.D. Importantly, however, since 400 A.D., and until roughly the 1800s,
that financial condition of release had only ever been what we call today an
unsecured financial condition.
32
An unsecured financial condition is like a debenture – secured by the general
credit and not specific assets of the surety – that is promised to be paid only
if the defendant fails to appear for court after release. Thus, after the Norman
Invasion (when England’s first jails were built) nobody was required to pay
anything up front to obtain release from incarceration; one only had to
promise to pay the money (or property) in the event of default. Accordingly,
in England, once a defendant was determined to be bailable, personal
sureties administering mostly unsecured bonds upheld the “bail equaling
release” tradition by assuring that nothing hindered the defendant’s release
from jail.
Colonial America adopted the English bail system, and to make sure that
bail equaled release it also adopted the use of personal sureties administering
unsecured conditions on bonds. Thus, as in England, courts determined
surety “sufficiency” by requiring sureties (i.e., persons) to “perfect” or
“justify” themselves as to their ability to pay the amount set, but they were
not required to post an amount prior to release of the accused. Instead, the

32 F.E. Devine, supra note 20, at 4; Paul Lermack, The Law of Recognizances in Colonial
Pennsylvania, 50 Temp. L.Q. 475, 497, 504-505 (1977) [hereinafter Lermack]. When reviewing old bail
cases, one must remember to note fundamental differences between English and American cases, older and
newer cases, and, importantly, civil and criminal cases. Civil bail cases often reviewed amounts relative to
the underlying debt, which might be inflated in the cause of action and lead to a claim for maliciously
“holding” the defendant to bail. Criminal bail cases follow the explanation of bail as explained in the body
of the NIC Fundamentals paper, supra note 6, using personal sureties and unsecured bonds, and with
amounts reviewed in terms of their effect on release. See generally Charles Petersdorff, A Practical
Treatise on the Law of Bail in Civil and Criminal Cases, passim (Jos. Butterworth & Son, 1824).
29
sureties were held to a debt that would become due and payable only upon
their inability to produce the accused. Because sureties were not allowed to
profit or be indemnified against potential loss in America as well, bonding
fees and collateral also did not stand in the way of release.
This model remained the primary means of effectuating the release of
bailable defendants in both England and America until the mid-1800s, when
both countries began gradually running out of personal sureties who were
willing to take responsibility over defendants for no money. Historically
speaking, this meant interference with bail as release, and thus some period
of bail reform seeking a solution was inevitable. But while England (and
other countries facing the same basic issue) found solutions that assured
release without infusing profit and indemnification into the criminal pretrial
process, America was unique in its decision to replace personal sureties with
commercial ones. Worldwide, America and the Philippines stand alone
among like countries in their decision to introduce profit into pretrial release.
As author F.E. Devine observed, except for those two countries, “the rest of
the common law heritage countries not only reject [bail for profit], but many
take steps to defend against its emergence. Whether they employ criminal or
only civil remedies to obstruct its development, the underlying view is the
same. Bail that is compensated in whole or in part is seen as perverting the
course of justice.”33
The change from personal to commercial sureties was designed to help get
bailable defendants out of jail, but the new model also had one important
unintended consequence, which was that it forever changed the essential
nature of the financial condition of release. As noted previously, for
centuries in England and America until the 1800s – that financial condition
of release had always been what we call today an “unsecured” financial
condition. Under this new model, however, the financial condition of release
would mostly be what we call today a “secured” financial condition. A
secured financial condition requires someone (typically a defendant or the
defendant’s family) to pay something as a condition precedent to release.34

33 Devine, supra note 20, at 201; See also Adam Liptak, Illegal Globally, Bail for Profit Remains in U.S.,
New York Times (January 29, 2008), found at
http://www.nytimes.com/2008/01/29/us/29bail.html?pagewanted=all&_r=0.
34 Secured bonds themselves add a barrier to release, but because commercial sureties can refuse to help a
defendant “for good reasons, bad reasons, or no reasons,” those sureties often add another deleterious layer
to that barrier. Ronald Goldfarb, Ransom: A Critique of the American Bail System at 115 (NY Harper &
Row, 1965).
30
And for the last 180 years, the interference caused by secured money bonds
has been at the heart of virtually every problem experienced in American
pretrial release and detention. Indeed, within only twenty years of the switch
to commercial sureties, influential legal scholars began writing documents
describing the deficiencies of the new model and calling for its reform.35
Secured money bonds interfere with bail as release by keeping lower,
medium, and even some higher risk persons in jail for lack of money even
though those persons could be safely managed outside of secure detention. It
interferes with detention by allowing extremely high risk persons to buy
their way out of jail when they are better suited for secure detention.
Moreover, secured money bonds interfere with release and detention even
when states have attempted to enact fair and transparent detention provisions
based on risk. In many states, leaving money in the system allows a
convenient (albeit unlawful) means of efficiently detaining defendants
without the bother of a due process hearing.36
Accordingly, when re-drawing the line between pretrial release and
detention, jurisdictions must remember that the root cause of this and
previous generations of reform – secured financial conditions – will
undoubtedly interfere with even the best drawn line. The key to avoiding
bail reform in the future is to create a transparent, legally justifiable, and
purposeful in-or-out bail system, with nothing hindering the decision to
release or detain. Accordingly, jurisdictions must consider eliminating the
use of secured financial conditions along with re-articulating a purposeful
pretrial release and detention process.
What Does the Law Tell Us About Re-Drawing the Line
Between Pretrial Release and Detention?
Although we will look later at various elements of the law in a more detailed
fashion, a broad look at American law can also tell us important things about
re-drawing the line between pretrial release and detention. For the most part,

35 See, e.g., Roscoe Pound & Felix Frankfurter, Criminal Justice in Cleveland, at 290 (The Cleveland
Found., 1922) (“The real evil in the situation [is] the disreputable professional bondsmen, who make a
business of exploiting the misfortunes of the poor.”). Most reformers in that generation focused on
commercial bail bondsmen as the main gatekeepers in the secured money bond model.
36 The author knows firsthand that Colorado’s preventive detention process is rarely, if ever used, due to the
ease in which money detains. Collaborative meetings in Wisconsin (a state with no commercial bail
bondsmen) have also revealed that, despite having a robust preventive detention process, money is still
frequently used to detain.
31
America borrowed English bail law verbatim, using the Statute of
Westminster, the English Bill of Rights, and even the Magna Carta in
applying bail to the Colonies. As in England, calling someone “bailable”
meant that he or she was expected to be released, just as calling that person
“unbailable” meant that he or she was eligible to be detained. As noted
previously, to make sure that bail equaled release, America also borrowed
the idea of using personal sureties administering mostly unsecured bonds.
The court cases in America reflected this “bail as release” notion until well
into the 1800s.37
Indeed, the concept of “bail” as release and “no bail” as detention was
articulated in the law throughout America’s long struggle with both
intentional and unintentional detention in the nineteenth and twentieth
centuries (discussed in detail, infra).
38 Even today, state supreme courts
asked to interpret certain constitutional bail provisions will occasionally
equate a right to bail with a right to release before trial.39 Accordingly, and
very broadly, the law tells us two important things: (1) how to do “bail” and
“no bail” so that bail equals release and that “no bail” is justified, narrowly
limited, and fair, which, in turn, points states generally toward the line
between release and detention; and (2) the very definition and purpose of
bail in America, which only reinforces the historical definition, as discussed
above.
How To Do “Bail” or Release
How to do bail (release) and no bail (detention) is somewhat simplified by
the fact that we currently really only have two Supreme Court opinions to

37 See generally, NIC Fundamentals, supra note 6, and NIC Money, supra note 30, and sources cited
therein. As will be discussed later, the idea that bail might not necessarily mean release began when
America started running out of personal sureties, causing the unintentional detention of bailable defendants.
Later, the Excessive Bail Clause was read by courts to allow unintentional detention, which then led to
judges using the financial condition to detain intentionally, albeit without saying so.
38 See id; see also notes 81-152, infra, and accompanying text.
39 See, e.g., State v. Brown, 338 P.3d 1276, 1277 (2014) (“The Bill of Rights of the New Mexico
Constitution guarantees that ‘[a]ll persons . . . before conviction’ are entitled to be released from custody
pending trial without being required to post excessive bail, subject to limited exceptions in which release
may be denied in certain capital cases and for narrow categories of repeat offenders.”); State v. Briggs, 666
N.W. 2d 573, 583 (Iowa 2003) (“However, if the accused shows that the bail determination absolutely bars
his or her utilization of a surety of some form, a court is constitutionally bound to accommodate the
accused’s predicament.”); State v Brooks, 604 N.W. 2d 345, 353 (Minn. 2000) (“If judges have unlimited
discretion to specify the form of acceptable bail, they would, for example, be able to set bail payable only
by real property. If the accused in such a case does not own any real property, he is in essence being denied
bail when he may be able to provide adequate assurance by some other means. As a result, the accused’s
constitutional right is violated.”).
32
guide us – one for release, and one for detention. The opinion in Stack v.
Boyle40 guides us through the release side of the equation. It does this by: (1)
equating the right to bail with the “right to release before trial” and the “right
to freedom before conviction;”
41 (2) telling us that this release is nonetheless
conditional upon having “reasonable” and “adequate” assurance to further
the legitimate purposes of bail (currently court appearance and, in virtually
every state, public safety);
42 (3) warning of the need for standards to avoid
arbitrary government action; 43 (4) requiring those standards to be applied to
every individual being assessed through the bail process and not allowing
those standards to be replaced with blanket conditions based on charge alone
(a warning that throws considerable doubt on the use of traditional money
bail schedules);44 (5) expressly articulating that the “spirit of the procedure”
of bail is to release people;45 and (6) further noting that setting a financial
condition of release with a purpose of detaining a defendant is “contrary to
the whole policy and philosophy of bail.”46
This last concept – the concept that it is improper to set a condition of bail to
purposefully detain an otherwise bailable defendant – is important to
highlight because it is misunderstood and often ignored today. The
American bail system was set up to allow the federal government and the
states to determine for themselves who is bailable and who is not, with
certain fundamental legal principles providing boundaries so that the right to
release is not unconstitutionally eroded. And, following this broad
allowance, the federal government and the states have declared certain
persons to be bailable and others to be potentially unbailable. Once that
declaration is in place, it would be clearly unlawful for a judge to essentially
skip that declaration and instead to purposefully detain a different set of
defendants based on that judge’s personal opinion of who should remain in
jail.
The notion that “bail” may not be used to purposefully detain was
commonly understood in America until well into the 1960s,47 and that

40 342 U.S. 1 (1951).
41 Id. at 4.
42 Id. at 4-5.
43 Id. at 5.
44 Id.
45 Id. at 7-8
46 Id. at 10.
47 See Daniel J. Freed & Patricia M. Wald, Bail in the United States: 1964, at 8 (DOJ/Vera Found., 1964)
(“In sum, bail in America has developed for a single lawful purpose: to release the accused with assurance
he will return at trial. It may not be used to detain, and its continuing validity when the accused is a pauper
33
common understanding – its sheer obviousness at the time – likely kept
courts from the need to declare it in opinions. Nevertheless, courts have
occasionally come to that conclusion when parsing the necessary elements
of excessive bail analysis.48 And, occasionally, a state supreme court will
announce (expressly or impliedly) that the unattainable amount alone is
sufficient to show a purpose to detain.49 Nevertheless, the practice remains
because judicial officers and other bail setters have become wise to the idea
that articulating no purpose for any particular condition – that is, setting bail
without making a record as to exactly why it is being set – can essentially
insulate those officials from appellate court findings of error.50 The
fundamental point is that setting bail to detain is unlawful, and it is only a
matter of time before the appellate courts will correct what has become an
unfortunate but common practice.
How To Do “No Bail” or Detention
If the United States Supreme Court’s opinion in Stack v. Boyle guides us in
matters of release, its opinion in United States v. Salerno51 guides us in
matters of detention. It does this by: (1) settling, at least for the time being,
the debate as to whether the Eighth Amendment to the United States
Constitution confers some federal right to bail thus affecting the states – it
appears not to, even though the language of Salerno could be read to provide
a basis for future decisions going either way;52 (2) settling, once and for all,

is now questionable.”). Freed and Wald’s document was used as the primary guiding source for the first
generation of bail reform’s National Symposium on Bail and Criminal Justice sponsored by Attorney
General Robert Kennedy.
48 See Galen v. County of Los Angeles, 477 F.3d 652 (9th Cir. 2007) (“The court may not set bail to achieve
invalid interests.”) (citing Wagenmann v. Adams, 829 F.2d 196, 213 (1st Cir.1987) (affirming a finding of
excessive bail where the facts established the state had no legitimate interest in setting bail at a level
designed to prevent an arrestee from posting bail)).
49 See, e.g., State v. Brown 338 P.3d 1276, 1293 (2014) (“Neither the New Mexico Constitution nor our
rules of criminal procedure permit a judge to set high bail for the purpose of preventing a defendant’s
pretrial release.”).
50 In State v. Anderson, 127 A.3d 100, 125 (2015), the dissent argued, “The undeniable purpose and effect
of the court’s imposition of a high monetary bond was to ensure that the defendant would be detained.” The
majority, however, disagreed based on the lack of any record showing purposeful detention and found,
instead, that “the defendant was not actually denied bail but, rather, was unable to post the bail that the trial
court, in its discretion, properly set.” Id. at 113. So long as judges do not make a record expressly
articulating a purpose to detain, the law has evolved to make it relatively simple to do just that.
51 United States v. Salerno, 481 U.S. 739 (1987).
52 On the one hand, the Court quoted from Carlson v. Landon, 342 U.S. 524 (1952), which cited historical
notions to provide support for Congress’s ability to extend pretrial detention to noncapital cases. On the
other hand, the Court said, “Carlson v. Landon was a civil case, and we need not decide today whether the
Excessive Bail Clause speaks at all to Congress’ power to define the classes of criminal arrestees who shall
be admitted to bail.” Id. at 754. For the various arguments, see Wayne R. LaFave, Jerold H. Israel, Nancy J.
34
whether flight alone is the only permissible purpose for limiting pretrial
freedom – it is not, and public safety is now equal to flight as a valid reason
for conditioning or denying release; (3) articulating liberty as a fundamental
interest, which should lead future court opinions applying Salerno to use
strict or at least heightened scrutiny in pretrial detention cases;
53 (4) allowing
pretrial detention despite substantive due process concerns that it imposes
punishment before trial or is always excessive;
54 (5) allowing pretrial
detention despite concerns that it is based on a prediction of risk of
something a defendant may or may not do in the future.
55
This last notion – that pretrial detention may be based on a prediction of
something someone may or may not do in the future – was vigorously
debated in America’s second generation of bail reform. While bail has been
in the business of prediction ever since something even resembling bail was
created in 400 A.D., those debates in the 1970s and 1980s illustrated that
detention based on prediction was something seen by many as practically
un-American. And because the idea of detaining someone pretrial for
something he or she may or may not actually do in the future is practically
un-American, the Court in Salerno went out of its way to express the idea
that liberty is the norm and detention must be “carefully limited.”
56 To make
sure that detention is carefully limited, the Court, in turn, emphasized three
considerations that are arguably necessary in any detention scheme: (1) the
need to articulate a particularly acute problem or justification for detention;
(2) the need to limit detention by charge; and (3) the need for procedural due
process.
A Particularly Acute Problem
First, pretrial detention should narrowly focus on some “particularly acute
problem in which the government interests are overwhelming.”
57 In the Bail
Reform Act of 1984, it was the “alarming problem of crimes committed by

King, & Orin S. Kerr, Criminal Procedure (4th ed., West Pub. Co. 2015) [hereinafter LaFave, et al.].
LaFave, in turn, points to Hunt v. Roth, 648 F.2d 1148 (8th Cir. 1981), which, though later vacated for
mootness, noted that, “If a $1,000,000 bond set arbitrarily by legislative fiat [for defendants all facing the
same charge] is excessive there is little logic to support the proposition that Congress could arbitrarily deny
bail for any or all criminal charges whatsoever.” Id. at 1160-61.
53 See. e.g., Lopez-Valenzuela v. Arpaio, 770 F.3d 772 (9th Cir. 2014).
54 Salerno, 481 U.S. at 746-51.
55 Id. at 751.
56 Id. at 755.
57 Id. at 750.
35
persons on release,”58 backed up by various research documents as noted in
the legislative history to the Act. Compare this to a case in 2014, however, in
which the Ninth Circuit Court of Appeals struck an Arizona “no bail”
provision by holding the law up to Salerno and concluding that it was not
“carefully limited,” as Salerno instructs, in part because it did not address a
particularly acute problem.59
Today, the problems we seek to address may be different, but they must be
real. Thus, when re-drawing the line between release and detention,
jurisdictions must clearly identify the issues that they seek to address, and
they must remember that simply adding certain classes of defendants to the
detention eligibility net without some research or findings showing those
classes to warrant detention would likely run afoul of Salerno’s requirement
of some narrow but identifiable problem. Thus, for example, if risk of flight
is not the problem in America that it once was, jurisdictions may have no
legal basis for detaining persons for that purpose. Likewise, if social science
research shows that “sex offenders” simply do not pose high risks for flight
or danger during pretrial release, addressing some perceived problem by
declaring all sex offenders potentially detainable might run afoul of Salerno.
Even more generally, if pretrial risk is simply not the problem jurisdictions
once thought it was before they had any empirical evidence (and used
criminal charge as a proxy for risk), those jurisdictions must be honest in
their determinations as to whether certain defendants need be detained at all.
Limited By Charge
Second, pretrial detention should be limited to some “specific category of
extremely serious offenses,” which includes persons found “far more likely
to be responsible for dangerous acts in the community after arrest.”60 This
goes to the detention eligibility net, which is discussed at length below, and
which would now potentially include – if possible to demonstrate – charges
justified through a showing of extreme risk of flight.
61 This standard, which

58 Id. at 742.
59 Lopez-Valenzuela, 770 F. 3d 772, at 782-84 (2014).
60 Salerno, 481 U.S. at 750.
61 There is a tendency to think of “preventive detention” as only a response to public safety, but, in fact,
preventive detention may be based on flight as well as public safety. Salerno spoke to the ability to detain
based on danger because that was the issue on appeal and it was the novel question facing America. The
Bail Reform Act of 1984, however, provided for preventive detention based on extreme risk of flight in
addition to danger. Risk of flight was the historic reason for denying bail to capital defendants, and was
gradually adopted in the second half of the twentieth century for noncapital offenses. In Lopez-Valenzuela
v. Arpaio, Judge Fisher responded to the dissent’s attempt to distinguish preventive detention based on
36
is based on criminal charge, may arguably be violated through provisions
like those found in various state constitutions that allow detention for “any
other crime,”62 all felonies,
63 or even “violent offenses” 64 if those categories
do not have some justification through empirical evidence or other
legislative findings showing that defendants in those categories present
higher risk.
In this generation of bail reform it is tempting to think that simply adding
“risk-based” language into a detention eligibility net will solve this problem,
but doing so raises its own issues. Again, this concept is discussed in more
detail later, but for purposes of example, consider Missouri, which through
its crime victim’s rights provision, added the following language to its
constitution: “Notwithstanding section 20 of article I of this Constitution
[providing a right to bail for all except defendants charged with capital
offenses], upon a showing that the defendant poses a danger to a crime
victim, the community, or any other person, the court may deny bail or may
impose special conditions which the defendant and surety must guarantee.”
65
On its face, this provision raises excessive bail concerns due to the fact that
it can be applied to literally every offense (using the blunt hammer of
detention for, say, traffic or low level misdemeanor offenses) as well as
concerns with vagueness and other aspects of due process. Moreover, this
particular risk-based provision – like provisions basing the ability to detain
on whether a person is “unmanageable” in that “no condition or combination
of conditions will suffice” to manage the risk (a resource-based provision in
addition to a risk-based one) is highly subjective, and the right to pretrial
release is far too important to allow erosion through such subjective
standards.
In a field gradually incorporating risk research and statistical assessment of
risk into the bail determination, the justification jurisdictions use for
determining future detention eligibility nets becomes crucial. Certainly,
using actuarial pretrial risk assessment instruments is a more rational way to

public safety versus flight (and thus to shift the analysis to one of excessive bail) and wrote, “[t]he Supreme
Court has never recognized – or even suggested – that distinction.” 770 F. 3d at 792, n. 16 (2014) (citing
supporting cases). While the history illustrates a distinction between flight and public safety, and within
public safety, danger to persons inside and outside of the criminal process, those distinctions have been
largely erased under Salerno.
62 See, e.g., Utah Const. art. I, § 8.
63 See, e.g., Ohio Const. art. I, § 9.
64 See, e.g., S.C. Const. art. I, § 15.
65 Mo. Const. art. I, § 32.
37
glean whether a defendant poses some extreme risk than by merely assuming
high risk for serious charges. As mentioned previously, statistical risk
assessment comes ever closer toward allowing us to answer two basic
questions at bail: (1) “How risky is this defendant?” and (2) “Risky for
what?” Nevertheless, before jurisdictions drop the concept of “risk”
wholesale into their right to bail provisions, the law requires that they take a
closer look at the risk research, which, in turn, adds some element of
complexity.
This paper strives to reduce this complexity, but for sake of illustration
before a more thorough examination of the issues, consider this example.
Salerno approved of the Bail Reform Act’s detention process, in part
because it included detention provisions that were limited only to “extremely
serious charges.” Through risk research, however, we see that there might
exist relatively “high” risk persons alleged to have committed all crimes, not
just “extremely serious” ones. And because Salerno does not dictate absolute
prerequisites to detention, jurisdictions might instead apply broad concepts
from Salerno, such as the need to limit detention to some narrow eligibility
net that is justified as relevant to addressing an acute problem, and apply
those concepts to a risk-informed field. And yet, because we live in a free
society and because risk inheres to the individual and not the crime, we must
build in some safeguard through a floor below which no risk assessment is
done and no detention is available. Right now, across America, people are
walking the streets who would be deemed “low,” “medium,” and “high”
risk, if only they were assessed. Accordingly, when one of those persons is
arrested for a felony, should they be assessed for that risk? What about a
misdemeanor? What about a traffic infraction? Should jurisdictions allow
courts to use detention under such a subjective or resource-driven notion of
whether conditions exist to manage risk, or should they provide meaningful
limits in our laws to simply remove certain classes of defendants from
detention eligibility? Understanding these complex scenarios means
understanding that it is Salerno’s broader concepts concerning “no bail” that
must be followed when fashioning detention provisions that do not offend
fundamental American notions of liberty and freedom.
Today we are seeing states change their constitutional right to bail or release
provisions to account for risk. So far, they have replaced extremely narrow
detention eligibility nets based on charge to extremely broad eligibility nets
to better account for risk. Often, these nets are coupled with a further
limiting process that expresses the idea that detention can be used only when
38
no condition or combination of conditions suffice to provide reasonable
assurance of public safety and court appearance, a subjective determination
that limits the right to bail based on existing jurisdictional resources. As we
will see later in this paper, crafting purely risk-based detention eligibility
nets raises many legal issues, especially when those nets have seemingly no
meaningful limits enacted through their implementing statutes or rules. The
fundamental point is that Salerno broadly instructs that detention must be
extremely limited in some justifiable way. In the 1980s, the limit was
charge-based and justified through certain underlying assumptions
concerning criminal charge, which are now being rightfully questioned.
Today, justifications might be different, but they must nonetheless exist.
Procedural Due Process
Third, detention may only be used after the government provides certain
fundamental procedural due process protections. In Salerno, the Court noted
that the government must first demonstrate probable cause that the arrestee
committed the crime.66 Next, the Court emphasized the Bail Reform Act’s
inclusion of a “full blown adversary hearing,” at which the government must
prove by clear and convincing evidence that no conditions of release can
manage the documented risk.67 This individualized determination of risk
served as an additional limitation on pretrial detention, as some persons
falling within the eligibility net would no doubt be deemed to pose
manageable risks after assessing their individual characteristics and
situations. The hearing itself allowed defendants to have counsel, crossexamine witnesses, testify and proffer evidence, and rely upon clearly
articulated standards used to guide judicial officers in the bail decision
(including standards designed to direct the judicial officer’s focus toward
“the nature and seriousness of the danger posed” by release) as well as a
requirement for written findings of fact and immediate appellate review.68
Taken together, Salerno’s prerequisites for a proper detention process
include a narrow eligibility net – justified by the law or the research – and a
process designed to further narrow the class of defendants held without bail.

66 Salerno, 481 U.S. at 751.
67 Id. at 750. Well known criminal law and procedure scholar, Wayne LaFave, considers both the finding of
probable cause and the necessary proof of the lack of sufficient conditions to be part of the substantive (as
opposed to procedural) due process analysis. LaFave et al., supra note 52, § 12.3(f), at 82. LaFave suggests
that many of the state preventive detention provisions today would not pass muster when held up to Salerno
for a variety of reasons. See id. at 55-68, 82.
68 Salerno, 481 U.S. at 743, 751-52.
39
In short, Salerno instructs that detention is lawful, but it must be justified,
carefully limited, and fair. Today in America, however, we see pretrial
detention that is careless, largely unlimited, and unfair. It is careless because
judges often set financial conditions without even knowing whether a
defendant will be released or detained, making detention essentially random.
It is unlimited because we see detention for all classes of defendants, from
low to high risk, and from accused murderers to accused shoplifters. And it
is unfair because it is based on money.
The Purpose of Bail and No Bail
Stack and Salerno also sum up the very purposes of bail and no bail, which
follow the history of bail, discussed above. The overall purpose of bail (a
process of conditional release) can be summarized as attempting
simultaneously to: (1) maximize release – the law favors, if not demands the
release of bailable defendants and detention should be a narrow exception to
the norm; (2) maximize court appearance – the law allows jurisdictions to
limit pretrial freedom for this purpose; and (3) maximize public safety – in
virtually every state and the federal system, the law also allows jurisdictions
to limit pretrial freedom for this purpose as well.
69 In short, as noted above,
the law informs us that the purpose of bail is to provide a mechanism for
conditional release, just as the purpose of “no bail” is to provide for a
mechanism for potential detention.
These three purposes are competing purposes, and thus they form a balance
that must be weighed when considering anything related to the release or
detention of defendants pretrial. Moreover, because it is a balance, persons
should never consider one purpose in isolation (indeed, the law and the
history suggest that release is likely paramount to court appearance and
public safety). Thus, for example, if a legislature desires to pass a bail bill
focusing entirely on public safety, the American public must force the
debate to include the effects of the bill on court appearance and release. As
another example, if a jail policy results in releasing a vast number of
defendants with no consideration of the effects on public safety and court
appearance, the American public must question and analyze that policy
pursuant to the balance. And as a final example, even if someone were to

69 The overall purposes of both bail and no bail would be to (1) maximize the appropriate placement of
defendants pretrial (the law allows states and Congress to determine bail eligibility, subject to broad
fundamental legal boundaries), while simultaneously (2) maximizing court appearance and (3) public
safety.
40
produce unbiased and unflawed research showing that commercial surety
bonds have some effect on court appearance, the American public and
criminal justice leaders must assess whether to continue using surety bonds
based on how those bonds hold up to the other two purposes; if they have no
effect on public safety and they significantly hinder release, then the balance
would suggest that America should cease using them. In addition to spelling
out broad fundamental legal principles, such as due process and equal
protection, the law tells people to do this type of weighing in the bail
process. Too often, however, people do not.
Despite Stack and Salerno providing us with guidance on how to do release
and detention, the opinions in those cases unfortunately never expressly
defined “flight” or “danger to the community.” Thus, the Court left it to
American jurisdictions to glean such a definition from parsing the language
of the detention eligibility net, the various limiting processes, and, indeed,
the facts of various detention cases that formed the basis for pretrial
detention to begin with. As noted by one bail scholar, “The Court suggested
that Congress enacted the statute to reduce ‘the alarming problem of crimes
committed by persons on release,’ yet the court failed to consider whether
the statute was meant to deprive liberty to prevent any crime or only serious
crimes.”70
Nevertheless, any lack of meaningfully guiding definitions is likely due to
the fact that we simply had no real research to back any up. Only recently
have we begun to examine exactly how risky persons actually are, and, more
importantly, the likely result of that risk. Drafters of the Bail Reform Act
perhaps did the best they could by making certain assumptions – for
example, an assumption that if a person arrested for a “serious” crime
committed another crime while on release, that crime would likely be the
sort that we, as a community, would feel the need to avoid through
detention. As we will see later, pretrial research is beginning to provide the
answers needed to adequately re-draw the line between release and detention
by providing empirical evidence about risk, and, in the process, the nuances
of dangerousness and flight.
This should not detract from the fact that Stack and Salerno together still
provide valuable lessons on how to do bail and no bail – pretrial release and

70 Abhi Raghunathan, “Nothing Else But Mad”: The Hidden Costs of Preventive Detention, 100 Geo. L. J.
967, 976 (2012).
41
detention. As in other areas of the law concerning bail, however, America
has largely ignored those lessons. Even the federal government – the object
of analysis by the Court in Salerno – has allowed the federal statute to lead
to over-detention through a widening of the detention eligibility net and
rebuttable presumptions in ways the Court might not approve today.
71
Indeed, courts are now beginning to hold up various bail provisions to
Salerno and find them unconstitutional.72 Moreover, at least one high court
has issued an opinion dramatically changing the way bail is done in an entire
state.73 Finally, federal courts have begun issuing opinions in which they are
saying that America’s predominant method of detaining defendants by using
money likely violates the Equal Protection Clause of the United States
Constitution.74 These cases likely reflect the beginning of a wave of
litigation designed to bring America’s bail practices more in line with
fundamental American legal principles, which, at their core, require the
government to adequately justify its processes and to apply them fairly to all
persons.
Accordingly, when re-drawing or re-articulating the line between release and
detention, jurisdictions must remember to return to the basics underlying
these legal principles, to read and understand the lessons from the primary
bail cases, to remain mindful of the tripartite balance of lawful purposes, and
to select new demarcations only when they are adequately justified.
What Do Ratios Have to Do With This?
When thinking about the law’s guidance for re-drawing the line between
release and detention, it is natural to think of the idea of some ratio of
released to detained persons. Indeed, even before the Statute of Westminster
in England, the fact that bail and no bail comprised the entirety of the

71 Compare 18 U.S.C. § 3142 (e) (1984) with 18 U.S.C. § 3142 (e) (2016). The bigger issue in the federal
system appears not to be the law, but instead application of the law leading to unexplainable and potentially
unconstitutional detention rates. A recent article in the Detroit News reported that nationwide 57.4 percent
of federal defendants are detained pretrial, essentially making detention, rather than release, the “norm” in
the federal system. Jennifer Chambers, At Federal Court in Michigan, Most Go Free Until Trial, (Jan. 17,
2016), found at http://www.detroitnews.com/story/news/politics/2016/01/17/release-detention/78950732/.
72 See, e.g., Lopez-Valenzuela v. Arpaio, 770 F. 3d 772 (2014); Simpson v. Miller, 387 P. 3d 1270 (Ariz.
2017).
73 See State v. Brown, 338 P.3d 1276 (2014).
74 See, e.g., Pierce v. City of Velda City, No. 4:15-cv-570, 2015 WL 10013006 (E.D. Mo. June 3, 2015)
(adopting a settlement agreeing to a new bail policy and declaring that, under the Equal Protection Clause,
no defendant can be held in custody based solely on inability to post a monetary bond).
42
dichotomy meant that every release and detention system would result in a
ratio. Moreover, people often point to “Blackstone’s Ratio” – the notion,
made famous by Sir William Blackstone and as mentioned above that, “It is
better that ten guilty persons escape than that one innocent suffer”75 (as well
as the Supreme Court’s reference to that ratio in explaining the presumption
of innocence in United States v. Coffin76) – to argue that the percentage of
persons released through a tolerance of false negatives to false positives
should lead to roughly detaining 10% of the total. This argument is bolstered
by the fact that America’s singular “model” bail jurisdiction, the District of
Columbia, which uses an in-or-out release and detention system with
virtually no money bonds, and which is uniformly praised by all criminal
justice actors within the District, just happens to release defendants in the
90th percentile while maintaining high court appearance and public safety
rates. Others, who equate Blackstone’s ratio with the “beyond a reasonable
doubt” standard of proof at trial, argue that a “clear and convincing
evidence” burden for detaining someone at bail might point toward an even
larger acceptable percentage of detained defendants. And still others use the
history and the law to argue that the detention rate should be far smaller. All
these arguments seem facially reasonable.
However, anyone desiring to use any ratio as instruction for re-drawing the
line between release and detention should realize three important things.
First, Blackstone’s ratio – described as 10:1 – is merely his reformulation of
numerous prior ratios articulated by numerous authors, which range from 1:1
to 1000:1, some of which were also cited by the Supreme Court in Coffin.
77
Thus, there is historical support for both enlarging and reducing detention
based on the use of ratios.
Second, the idea of articulating a ratio concerns our tolerance with false
negatives to false positives, which are not always easy to prove. At bail,
encountering false negatives entails releasing persons predicted to succeed
(the negative being a prediction that the person will not be violent or flee)
but who fail, and encountering false positives entails detaining persons
predicted to fail but who would have succeeded (like a false alarm). Using
Blackstone’s ratio, we might say that it is better to release ten false negatives
than to detain one false positive. Unfortunately, unlike a trial where guilt and

75 William Blackstone, Commentaries on the Laws of England, Book 4, ch. 27 (Oxford 1765-
1769).
76 Coffin v. United States, 156 U.S. 432, 456 (1895).
77 See generally, Alexander Volokh, nGuilty Men, 146 U. Penn. L. Rev. 173 (1997).
43
innocence are ultimately determined, there is no way to determine that a
detained defendant would have misbehaved if he was, in fact, let out of jail.
Moreover, when we use actuarial pretrial risk assessment instruments in the
bail determination, those instruments only tell us that a particular person
“looks like” another group of similar individuals who succeeded or failed at
certain rates. They cannot predict individual risk. Finally, these same
instruments are illustrating that even the highest risk defendants still succeed
50-70 percent of the time. Lower and medium risk defendants, as a group,
often succeed at extremely high rates, often in the 80th and 90th percentiles.
Given all this, a ratio like Blackstone’s, by itself, does not always fit well
with reality. Indeed, the lack of precision in measuring risk likely means that
the number of higher risk persons we release should be vastly larger than the
number of higher risk persons we detain, simply because in America doubts
about risk at bail should be settled in favor of release, not detention.
Third, as noted by Laurence Tribe, “The very enterprise of formulating a
tolerable ratio of false convictions to false acquittals, puts an explicit price
on the innocent man’s liberty and defeats the concept of a human person as
an entity with claims that cannot be extinguished, however great the payoff
to society.”78 To Tribe, deliberately punishing a person when we have
doubts about his guilt is not only wrong, but “morally and constitutionally
reprehensible.”79 The same should be true in bail, an area of the law where
preventive detention looks substantially similar to, if not indistinguishable
from, punishment.
While some, including this author, cite to Blackstone’s Ratio to caution
jurisdictions against adopting a false belief that “one crime is one crime too
many” in bail, Blackstone’s Ratio suggests a different way to think about
crime and bail: that one person wrongly detained is one person too many.
Accordingly, we must ensure that whatever process we adopt to allow for
detention painstakingly avoids this result, and we must remember that while
some ratio might be useful as a starting point in bail reform, it is the system
that we put in place that will ultimately determine it. As noted by Tribe, “the
final balance sheet obviously matters, but the process by which it is achieved
matters more.”80 The current money bail system, it is clear, has a process
that leads to the current ratio in an unjust and arguably unconstitutional

78 Tribe, supra note 1, at 387.
79 See id. at 375-86.
80 Id. at 387.
44
fashion. To the extent that any future system uses the history, the law, the
research, and the national standards to: (1) embrace the risk of release; (2)
significantly limit detention by creating narrow detention eligibility nets as
well as other limiting processes that also include due process hearings; and
(3) move American culture toward a culture of pretrial liberty and freedom,
we should not be surprised if the actual number of released defendants
grows higher than expected.
Thus, when re-drawing the line between pretrial release and detention,
jurisdictions must remember that they should not necessarily aim toward a
particular ratio of released to detained defendants, but rather let that ratio
evolve through the creation of a rational and transparent process of narrow
detention nets and limiting processes following the law and traditional
American notions of freedom and liberty.
What Else Do the Law and History of Bail Tell Us?
The law and the history of bail are intertwined, with historical events
providing the justification for new laws, and new laws, in turn, leading to
historical events. Not surprisingly, then, the law underlying bail and certain
historical events intertwined with that law tell us other important things
necessary to consider when re-drawing the line between pretrial release and
detention. Those things include, perhaps most importantly, America’s slow
struggle with the limits of intentional and unintentional detention. The story
of that struggle begins, once again, in England in 1275.
The Big Rule
As noted previously, ever since the Norman invasion, those administering a
system of bail have been concerned with putting people in the right places.
In 1275, the Statute of Westminster helped officials do this in England by
setting out three criteria bail setters were to weigh to determine bailability:
(1) the nature of the offense; (2) the probability of conviction; and (3)
criminal history (or “ill fame” of the defendant, including whether he tried to
escape).
81 Importantly, these three elements determined bailability before the
monetary condition (the only condition in use at the time) was set. As noted
previously, once defendants were deemed bailable, they had to be released,

81 See Carbone, supra note 16, at 524-25, n. 38
45
and thus England used a system of personal sureties and unsecured financial
conditions to assure that defendants were, in fact, released.
England’s notions of both unintentional and intentional detention were thus
fairly straightforward. If bailable, few, if any, defendants were kept in jail
unintentionally. On extremely rare occasions, defendants might not be able
to find sureties, and if the official also could not persuade persons to perform
as sureties in the case, the defendant would be jailed “unintentionally” – that
is, ordered released, but detained based on the inability to meet a condition
(not so much inability to pay as inability to find anyone willing to be
responsible for the defendant). Again, however, this was exceedingly rare,
especially given the relative lack of mobility of persons, and the various
social groups that allowed English bail setters to assign sureties in any
particular case. Nevertheless, unintentional detention did happen, albeit very
infrequently.
Intentional detention of bailable defendants, on the other hand, was
forbidden. Indeed, as noted previously, various attempts by English officials
to intentionally detain bailable defendants (as opposed to unbailable ones)
led to eras of bail reform and the creation of grand jurisprudential
mechanisms – such as habeas corpus – that we take for granted today. This
notion included trying to intentionally detain defendants through the use of
unattainable financial conditions. Bail scholars have written little on the
origins of the Excessive Bail Clause in England except to note that it was
enacted as a reform due to bail setters using money to intentionally detain
bailable defendants.
82 Whether officials were simply setting the unsecured
amount so high as to dissuade all sureties from performing the surety duty,
requiring the defendant to promise an amount all knew was unattainable to
him, or, while less likely and rarely in any event, attempting to charge an
amount up front in secured form is not entirely known.83 In any event, the
general rule in England was the same: bail set to intentionally detain a
bailable defendant was unlawful, and bail leading to the unintentional
detention of bailable defendants was incredibly rare. This made it possible
for England to adhere to what this author refers to as the “Big Rule:”

82 See Carbone, supra note 16, at 528; Duker, supra note 17, at 61-66.
83 For example, Holdsworth’s account of the history states that “judges did their best to evade [habeas
corpus] by requiring prisoners, entitled to bail, to find security in such excessive sums of money that they
were unable to furnish it.” Holdsworth, supra note 17, at 118-19. Other sources are largely silent on this
point, except to say that the traditional practice was only to require promises to pay financial conditions.
Nevertheless, it appears that officials could also use their ability to “justify” a surety (or even the
defendant) for sufficiency to keep defendants in jail even while using unsecured financial conditions.
46
because bail is release and no bail is detention, bailable defendants must be
released and unbailable defendants must be detained. Indeed, this rule is so
big that anything that interferes with it causes bail reform to happen.
The Big Change (The American Overlay)
When America was formed, it embraced England’s bail rules and
administered pretrial release and detention in virtually the same way.
Unbailable defendants were detained, and bailable defendants – through the
use of personal sureties and unsecured bonds – were released. Exceptions to
the rules were rare; as in England, most defendants found sureties and it was
unusual for a person to have literally no one willing to provide the surety
service.84 Moreover, as in England, virtually all defendants were released on
recognizance, requiring only that the defendant or surety promise to pay the
financial condition only upon default.85 Nevertheless, over time differences
in beliefs about criminal justice, differences in colonial customs, and even
differences in crime rates between England and the Colonies led to more
liberal criminal penalties and, ultimately, changes in the laws surrounding
the administration of bail.86
On the other hand, while England gradually enacted a complicated set of
rules, exceptions, and grants of discretion that governed bailability, America
leaned toward more simplified and liberal application by granting a
nondiscretionary right to bail to all but those charged with the gravest

84 See Paul Lermack, supra note 32, at 505 (“Although the amount of bail required was very large in cash
terms and a default could ruin the guarantor, few defendants had trouble finding sureties.”).
85 Devine, supra note 20, at 5. See also Lermack, supra note 32, at 504 (“Provision was sometimes made
for posting bail in cash, but this was not the usual practice. More typically, a bonded person was required to
obtain sureties to guarantee payment of the bail on default.”); Charles Petersdorff, A Practical Treatise on
the Law of Bail in Civil and Criminal Proceedings, 509-519 (describing the nature of the recognizance as a
debt owed to the court, forfeitable and payable upon the defendant not returning to trial) (London, Jos.
Butterworth & Son 1824). This did not mean that abuses did not occur. Occasionally, historians would note
abuses by officials who would declare certain sureties, including the defendant himself, to be “insufficient”
in that the officials believed they would not be able to produce the unsecured amount in the event of
default. Petersdorff warns that “extreme caution should be observed, that under pretense of demanding
sufficient sureties the magistrate does not require bail to such amount as is equivalent to the absolute
refusal of bail, and in its consequences, leads to a protracted imprisonment.” Id. at 512. Nevertheless in the
famous case of John Peter Zenger for libel against the Governor of New York, it appears the court
unlawfully set bail to detain Zenger either by requiring a secured amount ten times more than Zenger’s
sworn worth, or the same amount in unsecured form knowing that Zenger did not have it and thus would
not “ask any to become [his] bail” for lack of enough counter surety. See The Tryal of John Peter Zenger,
of New York, Printer, at 4,5 (J. Wilford/London 1738), found at http://www.nycourts.gov/history/legalhistory-new-york/legal-history-eras-01/History_Tryal-John-Peter-Zenger.pdf. Zenger was detained for the
duration of this trial, but was ultimately found not guilty.
86 See Carbone, supra note 16, at 529-30.
47
offenses and by settling on bright line demarcations to effectuate release and
detention. According to Meyer, early American statutes “indicate that [the] colonies wished to limit the discretionary bailing power of their judges in
order to assure criminal defendants a right to bail in noncapital cases.”87 This
ultimately meant that persons were declared “bailable” in America prior to
assessing any “risk” beyond that solely associated with the charge.
This is a fundamental point worth explaining. In England, the Statute of
Westminster listed bailable and unbailable offenses, but bailability was to be
finally determined by officials also looking at things like the probability of
conviction and the character of the accused, which were, themselves,
carefully prescribed by the Statute. Accordingly, there was, even then,
discretion left in the “bail/no bail” determination, which was ultimately
retained throughout English history. America, on the other hand, chose
bright line demarcations between bailable and unbailable offenses, gradually
moving the consideration of things like evidence or character of the accused
to determinations concerning conditions of bail or release, presumably
assuming that those determinations would not interfere with bailability (or
release) itself.
Thus, even before some of England’s later reforms, in 1641 Massachusetts
passed its Body of Liberties, creating an unequivocal right to bail for
noncapital cases, and re-writing the list of capital cases. In 1682,
“Pennsylvania adopted an even more liberal provision in its new
constitution, providing that ‘all prisoners shall be Bailable by Sufficient
Sureties, unless for capital Offenses, where proof is evident or the
presumption great.’”88 While this language introduced consideration of the
evidence for capital cases, “[a]t the same time, Pennsylvania limited
imposition of the death penalty to ‘willful murder.’ The effect was to extend
the right to bail far beyond the provisions of the Massachusetts Body of
Liberties and far beyond English law.”89 The Pennsylvania law was quickly
copied, and as America grew “the Pennsylvania provision became the model
for almost every state constitution adopted after 1776.”90 The Continental

87 Hermine Herta Meyer, Constitutionality of Pretrial Detention, 60 Geo. L. J. 1139, 1162 (1971-72)
[hereinafter Meyer].
88 Carbone, supra note 16, at 531 (quoting 5 American Charters 3061, F. Thorpe ed. (1909)) (internal
footnotes omitted).
89 Id. at 531-32 (internal footnotes omitted).
90 Id. at 532.
48
Congress, too, apparently copied the Pennsylvania language when it adopted
the Northwest Territory Ordinance of 1787.91
This was, indeed, a big change. England determined bailability by looking at
the individualizing risk factors in addition to charge. Then, once deemed
bailable, defendants were expected to be released. America simply labeled
large classes of defendants “bailable” and then told judicial officials that the
individualizing risk factors could only be used to adjust the monetary
condition of release. And this change – what this author calls the “American
Overlay” to English bail – combined with the gradual decline of the death
penalty,
92 meant that virtually every defendant was considered to be
“bailable.” This is what America wanted – a very broad right to bail, so
broad that even capital defendants might find release if the evidence were
slight. Coupled with the “Big Rule” (discussed above), which forbade the
detention of bailable defendants, the American Overlay to English bail law
meant that virtually every defendant was meant to be released prior to trial.
As in England, there were likely rare instances of unintentional detention
when defendants were literally unknown to the communities in which they
were accused, but the system simply did not allow for the intentional
detention of bailable defendants.
Such a broad system of release works only so long as defendants return to
court in acceptable numbers, which apparently happened during the colonial
period. 93 Gradually, however, America experienced a series of remarkable
events that led to more than 150 years of struggle with both unintentional
and intentional detention.
America’s Struggle With Unintentional Detention94
The first event leading to America’s struggle with unintentional detention
(bailable defendants ordered released but unable to obtain release for
whatever reason) was the slow decline and eventual disappearance of

91 Meyer, supra note 87, at 1163-64 (citing 1 Stat. 13).
92 See Carbone, supra note 16, at 534-35.
93 Lermack reported that the forfeiture rates in Colonial Pennsylvania were “high” (from 8-12% of all cases
coming to trial), but “never so high as to cause the system to break down altogether.” Lermak, supra note
32, at 507. In addition to the rate being only subjectively “high,” it should be noted that the forfeiture rate
included civil and criminal forfeiture as well as bail for witnesses, a practice since abandoned in America.
Moreover, at that time “bail jumping” was not a crime in Pennsylvania, and the courts’ contempt power
over defendants failing to appear was rarely used. Id. at 509.
94 The states likely had their own struggles with both unintentional and intentional detention, but this paper
only explores the phenomenon in the federal system.
49
personal sureties willing to take responsibility of defendants for no money.
This, in turn, caused unacceptable friction with the “Big Rule” requiring the
actual release of bailable defendants. There are many reasons for this, but
the effect both in England and America was the same: without personal
sureties willing to take responsibility over defendants, bailable defendants
remained in jail, a condition that historically required correction. Thus, in
England, Parliament passed laws allowing judges to release defendants with
no sureties. America, on the other hand, made it legal to both profit and be
indemnified at bail, essentially allowing the commercial surety system to
operate in this country starting in about 1900. Unfortunately, and as noted
previously, this changed how judicial officers set bail, from using mostly
unsecured to mostly secured financial conditions, a change that only
exacerbated the detention problem.95
It was during the decline of personal sureties in America that judges also
began experimenting with expanding the allowances for defendants to “selfpay” the financial condition. And it was during this experimentation that
judges began quickly to realize that very few defendants could personally
afford financial conditions of bond in even modest amounts.
It is precisely at this time that the Excessive Bail Clause could have been
used to declare any unattainable financial condition to be unlawful – indeed,
such a declaration would clearly follow from a reading of the history and the
law. Instead, however, to stem the tide of constitutional claims in the
tumultuous period of declining personal sureties (and before formally
ushering in the commercial surety system), judges created a line of cases
holding, essentially, that the financial condition of a bail bond is not
necessarily excessive simply because a defendant cannot pay it.96 This line
of cases provided an expeditious solution to the immediate problem, and
proved equally effective at stemming constitutional claims when the
commercial surety system also failed to solve the issue of unintentional
detention of bailable defendants. Unfortunately, however, this meant that
unintentional detention – a condition only very rarely tolerated in England
and America until this time – would now be tolerated in much greater
numbers and, indeed, given legal justification. In short, so long as a judge

95 Indeed, by the 1920s various bail scholars began calling for its reform. See, e.g., Roscoe Pound & Felix
Frankfurter (Eds.), Criminal Justice in Cleveland (Cleveland Found. 1922); Arthur L. Beeley, The Bail
System in Chicago, at 160 (Univ. of Chicago Press, 1927).
96 For a discussion of what has been termed by this author as the “unfortunate line of cases,” see NIC
Money, supra note 30, at notes 73-82 and accompanying text.
50
did not make a record to purposefully detain, detention due to the inability to
meet a condition (so-called unintentional detention) was lawful, and
considered to be simply an unfortunate byproduct of a system of conditional
release.
The United States Supreme Court’s 1951 opinion in Stack v. Boyle did little
to help the matter. While the Court in that case did equate the right to bail
with a “right to release before trial,” and while, in his concurring opinion,
Justice Jackson expanded on this notion to say that setting bail to assure the
defendants remained in jail “is contrary to the whole policy and philosophy
of bail,”
97 the Court stopped short of saying that unattainable amounts might
violate the constitution. Instead, it disposed of the case by merely holding
that bail was not “fixed by proper methods” when the trial court set the
financial conditions primarily based on the charge and otherwise failed to
follow or allow any evidence concerning the Federal Rules’ individualizing
standards for each defendant.98 At the time Stack was decided, the only
proper purpose for limiting pretrial freedom was court appearance, and the
only condition being used to achieve court appearance was money. And
though the Court wrote that “bail set at a figure higher than an amount
reasonably calculated to fulfill this purpose [i.e., court appearance]” would
be deemed excessive, the Court did not define flight, did not say what might
be “reasonable,” and did not in any way indicate intolerance for the
historical aberration of unintentional detention. Even today, in courts across
America, judges are allowed simply to declare an amount to be reasonable,
and so long as they do not expressly say that the amount is designed to
detain an otherwise bailable defendant, the resulting “unintentional”
detention is incorrectly accepted as part of a rational justice system.
Unintentional detention of bailable defendants led to the first generation of
American bail reform in the twentieth century, and the Bail Reform Act of
196699 (and state statutes modeled after the Act) tried to reduce unintentional
detention by focusing on alternatives to the traditional money bail system.
The Act did so by encouraging release on least restrictive, nonfinancial
conditions as well as presumptions favoring release on recognizance, which
were based on information gathered concerning a defendant’s community

97 Stack v. Boyle, 342 U.S. 1, 4, 10 (1951).
98 Id. at 6,7.
99 See 1966 Act, supra note 13. Among other things, the Bail Reform Act of 1966 also began changing the
nomenclature surrounding bail by beginning to focus on the word “release” instead of “bail,” a change that
was fully realized in the Bail Reform Act of 1984, and that has endured in the federal system and many
states.
51
ties to help assure court appearance. In 1968, the American Bar Association
Standards on Pretrial Release100 made numerous recommendations designed
to reduce or eliminate the unintentional detention of bailable defendants.
Unfortunately, to date no American state has incorporated the full panoply
of laws, policies, and practices first articulated by these documents.
Today, we are more concerned with the unintentional detention of so-called
“low” and “medium risk” defendants. Unfortunately, that detention is made
worse by our clinging to a system that uses money to intentionally detain socalled “high risk” defendants. Nevertheless, it is precisely our allowance of
unintentional detention that has led to this cyclical abuse. By permitting
unintentional detention based on such a loose standard as what a particular
judge feels is “reasonable assurance,” intentional detention using the bail
process by setting an unattainable money condition – considered unlawful
for centuries – is now quite easily achieved.101 So long as the judge does not
mention his intent to detain, bond amounts in the millions of dollars can be
justified as providing reasonable assurance of court appearance and survive
appellate scrutiny. The ability to easily detain using money, in turn, obviates
any need to create a rational and fair system of moneyless preventive
detention based on risk. And as long as money remains in use for high risk
persons, it tends to bleed into cases in which defendants can be managed
safely outside of secure detention. This is a cycle that must be broken.
Accordingly, when re-drawing the line between release and detention,
criminal justice leaders must be willing to fix a system that so easily allows
the unintentional detention of bailable defendants.
America’s Struggle With Intentional Detention
As noted previously, America greatly expanded the right to bail to virtually
all defendants not facing capital offenses, and also reduced the charges for
which the death penalty might apply. It is fairly well settled among bail
scholars that “capital crimes” exceptions to release were placed in federal

100 American Bar Association Standards for Criminal Justice (3rd Ed.) Pretrial Release (2007) [hereinafter
ABA Standards].
101 The practice of using money to intentionally detain a bailable defendant by merely acting as though it is
unintentional had become so apparent and prevalent by 1986 that, in its amicus brief in United States v.
Salerno, the National Association of Criminal Defense Lawyers wrote that preventive detention was
unnecessary because excessive bail had been historically tolerated to “incapacitate the rabidly dangerous.”
See Brief for National Association of Criminal Defense Lawyers at 8, United States v. Salerno, 481 U.S.
739 (1987).
52
and state laws based on the assumption that defendants facing death were
more likely to flee than those facing less serious punishment.102 Thus,
America was used to the concept of intentional detention of capital
defendants for risk of flight, but not for anyone else or for any other reason
besides flight. America’s struggle with intentional detention started when the
country began seeing unacceptable numbers of noncapital defendants
absconding after release, and reached an apex when America began seeing
unacceptable numbers of defendants committing crimes while on bail.
Initially, the problem of flight was easily managed by judges simply setting
unattainable secured money bonds while making no record of purposeful
intent to detain; since unintentional detention was lawful, judges could
simply make a record saying that the amount seemed “reasonable,” and
appellate courts would typically uphold the decision by assuming the
detention was unintentional.
103 The problem became acute, however, when
judges saw defendants absconding despite their best efforts to keep those
defendants in jail. This is seen in cases throughout the latter half of the
twentieth century, which reveal a slow erosion of the rule against intentional
detention of otherwise “bailable” defendants – ultimately both for flight and
public safety – leading up to the Bail Reform Act of 1984.
That erosion began with cases articulating the ability of judges to detain
released defendants once a trial had begun to protect the judicial process.
104
For example, in United States v. Bentvena,105 the Second Circuit Court of
Appeals reviewed the district court’s decision to remand nine defendants
who were perceived to be disrupting an ongoing trial. In its opinion, the
Second Circuit recited a defendant’s “absolute” right to bail justified by the
presumption of innocence as well as the need for unhampered preparation of
a defense, but then stated: “Once the trial begins, the right to bail is
necessarily circumscribed by other pressing considerations,” such as
potential delay, the possibility of interfering with witnesses, and the
investment of public funds “that demand that precautions be taken to ensure
that the proceedings go forward and terminate with all possible dispatch
consistent with due process.”106

102 See, e.g., Tribe, supra note 1, at 377-79, 397, 400-02.
103 Or, in a more nuanced argument, that the decision to release was intentional, but the actual release was
neither intentional nor unintentional, as it was up to the defendant to secure the required sureties.
104 These cases are different from those in which a defendant is remanded after conviction or in which the
defendant is detained as punishment through a judge’s contempt power.
105 288 F.2d 442 (1961).
106 Id. at 444.
53
Drawing a distinction between a right to bail before trial with a right during
trial, the Second Circuit held that “the district court possessed an inherent
authority to remand the defendants into custody during trial in the exercise
of sound discretion.”107 Even though this power should be used “with
circumspection,” the court explained, here the trial court did not abuse its
discretion given frequent delays, several lost jurors, and the judge’s inability
to distinguish various individuals among a total of nineteen defendants, all
of which presented a danger that “the trial might be disrupted and never
concluded.”108
One week later, in Fernandez v. United States, United States Supreme Court
Justice Harlan reviewed the bail determinations of four of the nineteen
defendants from Bentvena, above. In upholding the district court’s denial of
bail of those four, Justice Harlan wrote: “District courts have authority, as an
incident of their inherent powers to manage the conduct of proceedings
before them, to revoke bail during the course of a criminal trial, when such
action is appropriate to the orderly progress of the trail and the fair
administration of justice.”109 Nevertheless, Justice Harlan cautioned, while
not requiring the same degree of particularization necessary for initially
admitting a defendant to bail before trial, a remand during trial must not “be
ordered on an undiscriminating wholesale basis,” and must be based on
some showing of improper defendant conduct or other circumstances
overcoming a presumptive right of release.110
Approximately one year later, Justice Douglas, sitting as Circuit Justice, was
faced with a similar intentional denial of bail. In that case, Carbo v. United
States, the district court had denied the defendant’s request for bail pending
appeal due to a “strong likelihood of flight and of further threats and even
harm to the Government’s witnesses.”111 The Ninth Circuit Court of Appeals
rejected the district court’s rationale of protecting witnesses, and, because
the trial had concluded, reasoned that denial of bail pending appeal was also
improper when based on the need to avoid disrupting a trial (as in Bentvena,
above). Nevertheless, the Ninth Circuit agreed with the district court that
bail might be denied pending appeal for purposes of flight.112

107 Id. at 445.
108 Id. at 445-46.
109 Fernandez v. United States, 81 S. Ct. 642, 644 (1961).
110 Id. at 644, 645.
111 Carbo v. United States, 82 S. Ct. 662, 664 (1962).
112 See Carbo v United States, 302 F.2d 456 (9th Cir. 1962).
54
On review, Justice Douglas upheld the denial of bail based on risk of flight,
but wrestled somewhat with the notion of intentionally denying bail for the
purpose of protecting witnesses. Nevertheless, Justice Douglas concluded:
“In my view the safety of witnesses, should a new trial be ordered, has
relevancy to the bail issue. Keeping a defendant in custody during the trial
‘to render fruitless’ any attempt to interfere with witnesses or jurors may, in
the extreme or unusual case, justify denial of bail.”113
Rounding out these opinions, the Supreme Court wrote in the 1967 per
curiam opinion in Bitter v. United States as follows:
[A] trial judge has indisputably broad powers to ensure the
orderly and expeditious progress of a trial. For this purpose, he
has the power to revoke bail and to remit the defendant to
custody. But this power must be exercised with circumspection.
It may be invoked only when and to the extent justified by
danger which the defendant’s conduct presents or by danger of
significant interference with the progress or order of the trial.114
Variations of the statements found in these cases were articulated by later
courts seeking to deny bail both during trial and after conviction. Altogether,
they formed a jurisprudential rationale for a general rule that courts have
inherent authority to remand defendants once a trial has begun to protect
witnesses or the disruption of the administration of justice, including through
flight, but only in extreme or extraordinary circumstances. As will be seen
later in this paper, this general rule would ultimately be used to help justify
pretrial detention in both the D.C. law of 1970 and the Bail Reform Act of
1984.
Passage of the 1966 Bail Reform Act likely only added further complexity to
the struggle with intentional detention. As mentioned previously, the Bail
Reform Act of 1966 attempted to reduce unintentional or “needless”
detention,115 but it did not eliminate it. As noted in a contemporaneous

113 Carbo, 82 S. Ct. at 668 (footnote and citations omitted).
114 Bitter v. United States, 389 U.S. 15, 16 (1967).
115 See 1966 Act, supra note 13 (“The purpose of this Act is to revise the practices relating to bail to assure
that all persons, regardless of their financial status, shall not be needlessly detained pending their
appearance to answer charges, to testify, or pending appeal, when detention serves neither the ends or
justice nor the public interest.”).
55
publication by two of the pioneers of the first generation of bail reform,
Patricia Wald and Daniel Freed,
In two major respects, the [1966] Act falls short of completely
revising the old bail system: it does not authorize courts to
consider danger to the community in setting conditions of
pretrial release in noncapital cases; and, while it subordinates, it
fails to eliminate money as a condition which can cause the
detention of persons unable to raise it.116
Indeed, not eliminating money as a condition of release practically
guaranteed the continuation of unintentional detention of bailable
defendants. However, the bigger issue facing courts in the next twenty years
would be the fact that the 1966 Act did not speak directly to intentional
detention. The Act itself required the release of noncapital defendants on
personal recognizance or an unsecured bond unless “such a release will not
reasonably assure the appearance of the person as required.”117 When that
determination was made, the Act then required judicial officers to impose
the “first of the following” conditions of release (expressly delineating the
legal concept of least restrictive conditions), which were then listed in order
of their perceived restrictiveness.118 Nevertheless, the Act did not specify
precisely what judicial officers should do when no condition or combination
of conditions would suffice to reasonably assure court appearance. Looking
at the provisions as a whole, Wald and Freed concluded as follows:
On balance it appears that the act neither authorizes pretrial
detention nor guarantees that it will not occur. The ambiguity
reflects recognition by many members of Congress that there
was a need for detention [for risk of flight] in certain serious
cases but no practical way yet to solve the constitutional and
drafting problems in authorizing it.119
The authors concluded that it would thus be left to appellate courts to
provide the proper boundaries.

116 Patricia M. Wald & Daniel J. Freed, The Bail Reform Act of 1966: A Practitioner’s Primer, 42 ABA J.,
940-45 (Oct. 1966) [hereinafter Wald & Freed].
117 1966 Act, supra note 13, § 3146 (a).
118 For example, the first condition was to “place the person in the custody of a designated person or
organization agreeing to supervise him,” and the fourth was to “require the execution of a bail bond with
sufficient solvent sureties, or the deposit of cash in lieu thereof.” Id. at § 3146 (A) (1), (4).
119 Wald & Freed, supra note 116, at 944.
56
Because the 1966 Act also did not address public safety at bail, intentional
detention of bailable defendants could occur after the Act through two
methods. First, believing that the defendant posed an unacceptable risk for
flight or dangerousness, a judicial official might set an unattainably high
money bond designed to detain that person. If set for reasons of public safety
– at the time an unconstitutional purpose for limiting pretrial freedom – the
judge would be forced to couch the release order only in terms of court
appearance, and to refrain from any record discussing public safety. The
notion of setting a high money bail to protect the community (called settingbail for a “sub rosa” or secret purpose) was discussed but not resolved
during the first generation of bail reform in the 1960s,120 and was ultimately
a major catalyst leading to the Bail Reform Act of 1984.
Second, a judicial official, believing that the defendant posed the same high
risk for flight or dangerousness, might simply say that there were no
conditions or combination of conditions under the Act that would manage
the risk, and thus order the defendant purposefully detained with no
conditions whatsoever. Technically, ordering intentional detention based on
dangerousness would also be sub rosa, and thus unlawful as having an
improper purpose, and so many of the court cases decided in the wake of the
1966 Act were worded only in terms of flight.
In sum, the Bail Reform Act of 1966 did not eliminate unintentional
detention and did nothing to provide boundaries for intentional detention.
More fundamentally, though, throughout history there has been both “bail,”
or release, and “no bail,” or detention, and both are intertwined. To
adequately address either one, jurisdictions likely must address both.
Accordingly, by not addressing detention, the Bail Reform Act of 1966
could never fully fix release, and thus the law left enormous gaps in
American bail practice.
These gaps are seen through various court opinions grappling with the
concept of outright detention (with no ability to gain release) versus
detention using high money bail as well as with concepts of acceptable risk.

120 The issue of “setting high bail to prevent pre-trial release” was discussed in a separate chapter of the
Interim Report of the 1964 National Conference on Bail and Criminal Justice, but the issue raised more
questions (including the question of what, exactly, preventive detention was) than answers. See National
Conference on Bail and Criminal Justice, Proceedings and Interim Report, at 149-220 (Wash. D.C., Apr.
1965).
57
For example, in United States v. Leathers, the D.C. Circuit Court of Appeals
noted the “dramatic increase” in bail appeals by persons being held on
unattainable financial conditions, and recognized “the anomaly” of trial
judges’ trying but failing to adhere to the 1966 Act.121 Defendant Leathers,
who was being held on a $1,000 bond, sought a new hearing for the trial
court to consider fashioning nonfinancial conditions as an alternative to the
unattainable money condition. In granting that hearing, the appellate court
wrote: “The authors of the [1966] Act were fully aware that the setting of
bond unreachable because of its amount would be tantamount to setting no
conditions at all. Conditions which are impossible to meet are not to be
permitted to serve as a thinly veiled cloak for preventive detention.”122
Nevertheless, later that year the same court upheld a trial court’s order to
intentionally detain – without conditions – a defendant who the court
believed was a danger to government witnesses. In United States v. Gilbert,
the D.C. Circuit Court of Appeals focused not on the 1966 Act, but instead
upon the trial court’s common law authority to intentionally detain, noting
that “[a] trial court has the inherent power to revoke a defendant’s bail
during the trial if necessary to insure orderly trial processes.”123 While the
court acknowledged the right to pretrial release under the 1966 Act, it
nonetheless wrote:
In Carbo v. United States, Circuit Justice Douglas
acknowledged that this inherent power may even extend to
custody in advance of trial when the court’s own processes are
jeopardized by threats against a government witness. He took
the view that this inherent power should be exercised, however,
only in an ‘extreme or unusual case.’
***
We are satisfied that courts have the inherent power to confine
the defendant in order to protect future witnesses at the pretrial
stage as well as during trial. Yet this power should be exercised
with great care and only after a hearing which affords the
defendant an ample opportunity to refute the charges that if
released he might threaten or cause to be threatened a potential

121 United States v. Leathers, 412 F.2d 169, 170 (1969).
122 Id. at 171.
123 United States v. Gilbert, 425 F.2d 490, 491 (D.C. Cir. 1969).
58
witness or otherwise unlawfully interfere with the criminal
process.
124
Further illustrating the struggle over intentional detention, in Gavino v.
MacMahon the Second Circuit Court of appeals cited to Carbo, but refused
to follow Justice Douglas’s suggestion that a judge’s inherent power to
detain might extend to defendants before the trial had even begun.125 The
Gavino panel wrote:
The Bail Reform Act, like its predecessor, guarantees that in a
noncapital case the defendant will have the pretrial right to
release on bail except in extreme and unusual circumstances,
e.g., where threats to a government witness would jeopardize
the court’s own processes. Although the trial judge is accorded
discretionary power during trial to revoke bail where such
drastic relief is essential to insure the orderly progress of an ongoing trial, such power must be ‘exercised with circumspection,’
and does not extend to revocation of bail before trial, which is
the situation confronted here.126
In the 1971 case of United States v. Smith, the Eighth Circuit Court of
Appeals attempted to harmonize the 1966 Act with instances in which bail
was nonetheless being denied. Citing Carbo (the case hinting at inherent
authority to detain pretrial), the Eighth Circuit focused on specific appellate
provisions tending to show that Congress did not intend release under the
Act to be absolute. The Smith panel wrote as follows:
Rule 9 [of the Federal Rules of Appellate Procedure] recognizes
that bail may be refused under appropriate circumstances by
authorizing an appeal from either a refusal of bail or from
conditions imposed that prove onerous to the defendant. The
district court in such cases must state in writing the reasons for
refusing bail or for imposing conditions of release. The right to
bail is thus not absolute but decisionally recognized and

124 Id. at 491-92 (internal citations omitted) This was a significant extension of Carbo, in which Justice
Douglas was clear that his perceived threat to witnesses was due, in large part, to the likelihood of a new
trial being ordered and thus the defendant knowing the adverse testimony likely to be given. See Carbo, 82
S. Ct. at 668.
125 Gavino v. MacMahon, 499 F.2d 1191, 1195 (1974) (internal citations omitted).
126 Id. (internal citations omitted) (quoting Bentvena, 288 F.2d at 445).
59
statutorily approved as being generally available in noncapital
cases subject to denial in exceptional cases and subject to the
imposition of reasonable conditions of release. Bail may be
denied in the exceptional case.127
Cases from the Sixth Circuit, too, illustrated that court’s internal struggle
with release and detention under the 1966 Act. In United States v. Wind, a
panel of the Sixth Circuit Court of Appeals wrote: “Since Congress did not
intend to address the problem of pretrial detention without bond in the Bail
Reform Act of 1966, the existence of extrastatutory powers to detain persons
prior to trial may be considered.”128 Citing Fernandez, Bentvena, and Carbo,
discussed above, the court noted a judge’s inherent right to revoke bail
during the course of a trial, and then, following Justice Douglas’s suggestion
in Carbo, wrote:
We are satisfied that courts have the inherent power to confine
the defendant in order to protect future witnesses at the pretrial
stage as well as during trial. Yet this power should be exercised
with great care and only after a hearing which affords the
defendant an ample opportunity to refute the charges that if
released he might threaten or cause to be threatened a potential
witness or otherwise unlawfully interfere with the criminal
prosecution.129
The Court noted discrepancies between the D.C. Circuit’s holdings in
Leathers and Gilbert, but followed Gilbert as that case dealt with denial of
bail rather than detention due to the unattainable amount.130 Just one year
later, however, another panel of the Sixth Circuit vacated a district court’s
denial of bail for a defendant charged with threatening the life of President
Ford. In that case, United States v. Bigelow, the panel cited to both Wind and
Gilbert, but declined to detain in the instant case because the defendant had
not threatened witnesses or otherwise taken steps to obstruct the trial.131
Finally, in the 1978 case of United States v. Abrahams,132 a panel of the First
Circuit Court of Appeals reviewed each of these prior cases to rule on an

127 United States v. Smith, 444 F.2d 61, 62 (8th Cir. 1971).
128 United States v. Wind, 527 F.2d 672, 674 (6th Cir. 1975).
129 Id. at 675.
130 Id.
131 United States v. Bigelow, 544 F.2d 904, 907-08 (1976).
132 United States v. Abrahams, 575 F.2d 3 (1st Cir. 1978).
60
order denying bail altogether under the 1966 Act. After discussing the
various cases, the panel found most instructive the reasoning from Smith,
above, which concluded that the 1966 Act recognized the potential for denial
of bail due to the inclusion of an appellate procedure to review those denials.
Additionally, the panel quoted approvingly the following language from a
federal district court opinion, which, the panel wrote, “here fits almost
exactly” the present case:
While the statute, § 3146 (a), does not say this in so many
words, it has been thought generally that there are cases in
which no workable set of conditions can supply the requisite
reasonable assurance of appearance for trial. To state, the
extreme case, which is not a hypothetical, the strong
presumption favoring release may disappear for a defendant
charged with a grave offense, with powerful evidence against
him, who lacks family ties or employment or resources or any
roots in the community, and is possessed of a poor record for
fidelity to court engagements. Such a defendant may have to
stay in jail pending a trial to be brought on with utmost possible
speed.133
The various courts’ ongoing struggle with release and detention pursuant to
the 1966 Act is probably best illustrated by United States v. Melville,134 the
result of which resembles the untenable “dance” around money as a
detaining mechanism that exists even today. In Melville, four defendants
were charged with conspiring to detonate a number of bombs in New York
City. A bail commissioner ordered their release on bonds with financial
conditions ranging between $100,000 and $300,000, none of which the
defendants could meet. Looking at various defendant characteristics, the
district court reviewing the Commissioner’s decision wrote: “it is
overwhelmingly likely that none of them can approach anything close to the
amount of bail prescribed for his release.”135 Moreover, the court wrote,
[I]t is apparent that in this instance, as in many others familiar
to all of us, the statement of the astronomical numbers is not
meant to be literally significant. It is a mildly cynical but
wholly undeceptive fiction, meaning to everyone ‘no bail.’

133 Id. at 8 (quoting United States v. Melville, 306 F. Supp. 124, 127 (S.D.N.Y. 1969).
134 United States v. Melville, 306 F. Supp. 124 (S.D.N.Y. 1969).
135 Id at 127-28.
61
There is, on the evidence adduced, no possibility that any of
these defendants will achieve release by posting bond in
anything like the amount which has been set.136
Despite this language, and even while taking notice of amounts that the
defendants themselves said were attainable, the district court nonetheless set
new financial conditions, considerably less than the $100,000 to $300,000
previously set, but still between two to ten times more than what each
defendant said he could meet.
Across America we still see distressing opinions such as this, with one judge
concluding one amount to be reasonable, with another judge concluding that
amount to be unreasonable and sometimes settling on a second amount, and
with both amounts being beyond what the defendant said he could pay.
Unfortunately, both unattainable amounts are equally arbitrary and yet
equally effective in detaining the defendant pretrial. The only difference is
that the second amount is given the illusion of legal legitimacy through the
gloss of an appellate opinion. The overall effect of both opinions, however,
illustrates a kind of illegitimacy that erodes our core perceptions of justice.
Lessons From the Detention Cases
The various detention cases decided both before and after the 1966 Act
(described above) illustrate three things. First, they represent a gradual
chipping away at the historical notion that intentionally denying release to
“bailable” defendants is unlawful. For centuries, attempts to detain bailable
defendants before trial on purpose have led to bail reform to cure what was
considered to be a violation of the Big Rule: bailable defendants must be
released. These cases gradually eroded that Rule, to the point where even
bailable defendants under a release-oriented statute like the Bail Reform Act
of 1966 might still be detained on purpose.
Second, the cases illustrate that the notion of detaining noncapital defendants
prior to trial for anything, let alone flight, was far from settled, with
seemingly discordant opinions even within circuits. During debates and
associated cases concerning preventive detention under both the D.C. Court
Reform Act and the Bail Reform Act of 1984, however, various supporters
of preventive detention repeatedly wrote that pretrial detention for
protecting witnesses and jurors – or even to respond to risk of flight for

136 Id. at 127.
62
noncapital defendants – was somehow already decidedly woven into the
fabric of American criminal justice.
137 The citation to Carbo in many of
those writings is especially noteworthy simply because Carbo, through an
opinion issued by a single Justice sitting as Circuit Justice, merely suggested
the propriety of applying common law detention notions to defendants
pretrial that had previously only been applicable during trial.
Indeed, when it came to intentional detention of noncapital defendants based
on flight, it appears that Congress relied on virtually no authority whatsoever
when it began codifying the practice. In the House Report accompanying the
D.C. Court Reform and Criminal Procedure Act of 1970, which was the first
federal legislative articulation of intentional detention of noncapital
defendants based on either flight or public safety, the Committee on the
District of Columbia wrote that, “Criminal defendants today may be
detained if found likely to flee regardless of the conditions of release
imposed.”138 In making this statement, however, the Committee cited no
authority whatsoever.139
Similarly, in a comprehensive and contemporaneous law review article
describing the D.C. Act, the authors stated the same conclusion – that it was
lawful to detain noncapital defendants for risk of flight – and cited to
Melville, discussed above.140 Unfortunately, and as mentioned previously,
the Melville court did not uphold purposeful or intentional detention without
conditions, but rather upheld amounts of financial conditions that merely led
to detention, a fact pattern falling more appropriately into the category of
cases dealing with unintentional detention. Indeed, while the court in
Melville wrote, in dicta, that it could foresee cases in which “no workable set
of conditions can supply the requisite reasonable assurance of appearance
for trial,” it also said that it was only assuming that a court might be able to
detain on that basis.
141 Moreover, that court found that the amounts ordered
led “practically to a denial of release conditions in a case where justification
for this extreme result is not established,” and thus the court set its own set

137 See, e.g., Blunt v. United States, 322 A.2d 579, 584 (D.C. Ct. App. 1974) (superseded by statute on other
grounds) (citing Carbo and Gilbert while reviewing the constitutionality of the D.C. pretrial detention
statute); see also United States v. Edwards, 430 A.2d 1321 (D.C. Ct. App. 1981) (citing Carbo and Blunt
while doing same).
138 H. Rep. No. 91-907, at 88, 92 (1970).
139 Citing Carbo and Gilbert, the Committee Report did note, however, that, “Defendants may be detained
prior to trial if they threaten witnesses or otherwise obstruct justice.” Id. at 92.
140 See Carl S. Rauh & Earl J. Silbert, Criminal Law and Procedure: D.C. Court Reform and Criminal
Procedure Act of 1970, 20 Am. U. L. Rev. 252, 289 (1970-71) [hereinafter Rauh & Silbert].
141 See Melville, 306 F. Supp. 124, at 127.
63
of financial conditions of release that it believed the defendants “may be
able to post.”142 Accordingly, using Melville as direct support for a
conclusion that courts may lawfully detain noncapital defendants was most
definitely misplaced.
In reality, there was no decent authority to detain noncapital defendants on
purpose based on risk of flight when the D.C. Act was enacted in 1970.
Between 1970 and 1984, the First Circuit Court of Appeals decided
Abrahams, discussed above, which relied on Melville’s dicta to become the
first federal court to publish an opinion allowing the intentional detention –
without conditions – of an otherwise bailable noncapital defendant for risk
of flight. Nevertheless, it appears that the Abrahams holding never took root
beyond the First Circuit from which it was decided. While other courts
(including a fairly long list of New York federal district courts in cases
extending beyond the Bail Reform Act of 1984) dodged the argument or
mentioned Abrahams only in passing, the First Circuit was the only Circuit
that ever cited to Stack v. Boyle143 and Abrahams as twin authority for the
proposition that bailable defendants facing noncapital charges could be
detained without bail through some extra-statutory “inherent” authority
when “no condition or combination of conditions” under the Bail Reform
Act of 1966 would suffice to provide reasonable assurance of court
appearance.
144
When enacting the Bail Reform Act of 1984, Congress nonetheless used
Abrahams as its singular precedent when it said it was “codify[ing] existing
authority to detain persons who are serious flight risks.”
145 Congress did so
despite the fact that Abrahams rested on dubious authority itself, and never

142 Id. at 129.
143 342 U.S. 1 (1951). In fact, Stack does not support detaining bailable noncapital defendants pretrial. In
addition to equating the right to bail with the “right to release before trial,” and the “right to freedom before
conviction,” that opinion also noted that setting a financial condition in order to intentionally detain a
defendant pretrial “is contrary to the whole policy and philosophy of bail.” Id. at 4, 10.
144 See, e.g., United States v. Schiavo, 587 F.2d 532 (1st Cir. 1978) (“The Standards of 18 U.S.C. § 3146,
with its ‘presumption of releasability’ apply. Only in the rarest of circumstances can bail be denied
altogether in cases governed by §3146.”) (citing Abrahams, 575 F.2d 3 (1st Cir. 1978)). The Eighth Circuit
came close to a holding like Abrahams, but that case was vacated as moot by the United States Supreme
Court. See Hunt v. Roth, 648 F.2d 1148, 1158 (8th Cir. 1981) (“We recognize that there may be instances
where no amount of bail can sufficiently protect the state’s interests. In such a case, a court may consider
the relevant factors and deny bail.”) (citing Abrahams), vacated as moot 455 U.S. 478 (1981).
145 S. Rep. No. 98-225, at 18, 1983 WL 25404, at *8. As noted previously, the panel in Abrahams reviewed
several cases for lack of clear guidance and was ultimately persuaded by three sentences from a 1969
district court case surmising, without support, “It has been thought generally that there are cases in which
no workable set of conditions can supply the requisite reasonable assurance of appearance at trial.” See
Melville, 306 F. Supp. 124, 127 (S.D.N.Y. 1969).
64
found its way beyond mostly mere mention within the First Circuit Court of
Appeals. Indeed, among the cases cited within the Abrahams opinion, those
that concerned intentional detention with no conditions were concerned
almost exclusively with a court’s authority only to purposefully detain to
protect witnesses. Purposeful detention for flight for noncapital defendants
was not only a historical aberration; it was also novel to even modern
American justice. The fundamental point is that purposeful pretrial detention
with no conditions for risk of flight by noncapital defendants was not some
deeply rooted American tradition when Congress began codifying it. The
release of all noncapital defendants was.
Third and finally, whether to protect against risk of flight or to protect
witnesses or jurors, the detention cases virtually always noted that any
exception to release should be reserved only for the “extreme and unusual
case,”146 and thus the facts of those cases instruct on what, exactly, the
courts believed to be extreme cases of risk. In Carbo, for example, the case
in which Justice Douglas surmised that safety of witnesses might justify
pretrial detention in “extreme or unusual” cases, the Justice noted that one
witness in the case at hand had received 200 threatening phone calls, had
been severely beaten, had seen an “ominous” car near his home that was
driven by associates of the defendants, all of which led the trial judge to
conclude that there was a “strong likelihood that witnesses in this case will
be further molested or threatened and perhaps even actually harmed.”147 In
Wind, the case that extended “inherent” or “extrastatutory power” to detain
without bond to the pretrial phase because such detention was not addressed
in the 1966 Act, the Sixth Circuit Court of Appeals noted that the defendant
had admitted “he would post the $1,000,000-bond and then would flee, and
that no witness would testify against him.”148 This, along with evidence that
potential witnesses refused to testify for fear of the defendant, led the Sixth
Circuit to conclude that there was substantial evidence that the defendant
“possessed dangerous propensities” toward witnesses. Even so, that court
nonetheless vacated the denial of bail because the reviewing judge
apparently also relied on additional evidence gleaned from an in camera
hearing, in which the defendant and his lawyer were excluded.149

146 See, e.g., Abrahams, 575 F.2d 3, at 8 (“This is the rare case of extreme and unusual circumstances that
justifies pretrial detention without bail.”); United States v. Schiavo, 587 F.2d 532, 533 (“Only in the rarest
of circumstances can bail be denied altogether in cases governed by § 3146.”).
147 Carbo, 82 S. Ct. 662 at 664, 668 (1962) (internal quotation omitted).
148 United States v. Wind, 527 F.2d 672 at 673 (6th Cir. 1975).
149 Id. at 674-75.
65
The facts of Abrahams, too, are enlightening. In that case, a panel of the
First Circuit noted that the defendant: (1) had three previous convictions; (2)
was an escaped prisoner from another state; (3) had given false information
at a previous bail hearing in the same case; (4) had failed to appear in the
current case; (5) had failed to appear in a case in a third state and was a
fugitive there; (6) had used several aliases; and (7) had transferred 1.5
million dollars to Bermuda within the previous two years.150 Based on these
facts, the panel concluded:
The record before us depicts a man who has lived a life of
subterfuge, deceit, and cunning. He is an escaped felon. He did
not hesitate to flee to Florida and forfeit $100,000 to avoid [a
hearing]. There is nothing in the record that suggests that bail
will result in his appearance at trial. Every indication is to the
contrary. This is the rare case of extreme and unusual
circumstances that justifies pretrial detention without bail.151
Even later, in a case decided after the Bail Reform Act of 1984, we see
mostly extreme facts justifying pretrial detention. In United States v.
Melendez-Carrion, the court upheld detention solely for risk of flight when
the defendant: (1) was found to be a member of a paramilitary, terrorist
organization or gang seeking to advance Puerto Rico independence; (2) had
knowledge of and access to various gang safe houses; (3) had assisted a
convicted felon to escape detection; (4) had recently traveled to Costa Rica
and Panama for reasons not explained to officials; and (5) had in his
possession documents reflecting various contacts in foreign countries.152
Later we will see how the United States Supreme Court absorbed risk to
jurors and witnesses into the larger notion of risk to the general public.
Nevertheless, the fundamental lesson learned from these detention cases is
that, for a number of reasons, America struggled with the boundaries of
intentional detention in ways that conflicted with the notion of bail as
release. Those reasons are numerous, and perhaps interwoven, and include a
variety of social changes seen in this country throughout the nineteenth and
twentieth centuries, including increased use of drugs and guns, more
efficient means of travel, and increased fear of crime, to name only a few.
These social changes were offset somewhat, however, by improved

150 United States v. Abrahams, 575 F.2d 3, at 4-5 (1st Cir. 1978).
151 Id. at 8.
152 United States v. Melendez-Carrion, 790 F.2d 984, at 994-95 (2d Cir. 1986).
66
policing, jails, court systems, and even laws attempting to keep up with
those changes.
Overall, from this struggle with intentional detention we see that America
had become dissatisfied with accepting widespread releases by declaring
virtually all defendants bailable and limiting the process of release only to
assuring court appearance. Due to this dissatisfaction, both unintentional
and intentional detention began to flourish, thus eroding the broad right to
bail coupled with a rule that virtually all bailable defendants should be
released. But jurisdictions should note that both kinds of detention of
bailable defendants are historical and legal aberrations. For centuries,
bailable defendants were only rarely detained unintentionally, and they
were never allowed to be detained intentionally. By the mid-twentieth
century, however, through two discreet lines of cases, America saw
countless defendants detained unintentionally, and the beginning erosion of
the time-honored rule against intentional detention of bailable defendants.
This struggle would ultimately lead to the need for some sort of fix – some
way to allow courts to answer the foundational questions of who gets
released and who gets detained based on the two subsidiary questions of (1)
“How risky is this person?” and (2) “Risky for what?”
Accordingly, when re-drawing the line between pretrial release and
detention, jurisdictions must remember that America was founded on a
broad right to release before trial, with only rare instances of unintentional
detention and virtually no instances of intentional detention. This, alone,
should lead jurisdictions to consider only the narrowest gateways toward
pretrial detention overall. Moreover, even when it was allowed, intentional
detention was based only on “rare” or “extreme” circumstances, such as in
the cases described above. In sum, when we first began intentionally
detaining people for flight and danger, flight and danger were defined to
include only the most extreme instances of facts and circumstances showing
a high likelihood of fleeing to avoid prosecution or harming identifiable
people through serious or violent crimes.
The Big Fix
It is clear that by the mid to late 1960s, America was in need of some fix.
The Bail Reform Act of 1966 was only designed to reduce the needless
detention of bailable defendants. However, as discussed above, it did not
stand in the way of either unintentional or intentional detention. Moreover,
67
it was not designed to deal with the issue of public safety. Although setting
bail for public safety – considered an unlawful purpose – was discussed at
the 1964 National Conference on Bail and Criminal Justice, the 1966 Act
only narrowly addressed public safety by allowing courts to consider
dangerousness posed by capital defendants and defendants awaiting
sentence or appeal.153
In 1970, a committee of Congress wrote that the 1966 Act went far in trying
to eliminate money as a barrier to release, but that by “totally eliminating”
dangerousness as a criterion in setting conditions of release, the Act was not
copied by the states because it “ignored the rationale behind 700 years of
legal practice.”154 In fact, the 1966 Act did no such thing because
dangerousness was never there to begin with. What America did, in fact,
was much more nuanced. It initially broadened the right to bail to virtually
all defendants. In doing so, it forced judges to consider factors one might
normally think would help with the in-or-out decision (including giving
judges some indication of dangerousness) only when deciding on the
amount of money necessary to avoid flight. All of this was designed to
follow the “Big Rule,” which said that people called “bailable” should be
released. Nevertheless, soon persons showing extremely high risk for flight
and to public safety began to interfere with our notions of both release and
detention, causing America to struggle with both unintentional and
intentional detention. Interference with release and detention causes bail
reform, and so we have endured a century of trying to provide an overall fix
designed to simply put defendants in the right places.
Nevertheless, throughout American history it was widely known that risk of
flight was the only constitutionally valid purpose for limiting pretrial
freedom, and thus basing detention on risk of future dangerousness was
simply not a lawful part of the American bail system. Indeed, when
America narrowed the eligibility for detention to mostly capital offenses, it
did so not to protect the community, but instead to protect against flight
from a defendant facing death; as noted previously, it was commonly

153 Former Section § 3148 allowed the detention of capital defendants or convicted persons awaiting
sentence or on appeal if the court believed that “no one or more conditions of release will reasonably assure
that the person will not flee or pose a danger to any other person or to the community.” 18 U.S.C. § 3148
(1966).
154 H.R. Rep. 91-907, at 840-85. Today, it is clear that many states did copy parts of the 1966 Act.
However, because the states did not eliminate secured financial conditions – in part because they were
deemed necessary to detain defendants for purposes of public safety – those states never fully fixed the
problem of unnecessary detention due to money.
68
assumed that a person facing death would flee to avoid the punishment.155
The detention cases provided some articulation of danger, but not nearly
enough to easily extend those cases to reflect danger beyond witnesses and
jurors. Moreover, the extremely limited number of flight cases also
illuminated the lack of explicit guidance from the 1966 Act as to how to
deal with extreme cases involving risk of flight.
To fix these issues, Congress passed two laws: (1) The District of Columbia
Court Reform and Criminal Procedure Act of 1970,156 and (2) The
Comprehensive Crime Control Act of 1984, 157 which contained the Bail
Reform Act of 1984. The fix involved: (a) determining up front who should
be purposefully released and detained through a detention eligibility net; (b)
making sure intentional detention was further limited through a process
capable of dealing with extreme cases of risk ultimately for both flight and
public safety; and (c) attempting to eliminate unintentional detention
altogether through significant limits on the use of money.
The D.C. Court Reform and Criminal Procedure Act of 1970
The pretrial release and detention provisions of the D.C. Court Reform and
Criminal Procedure Act of 1970 (“1970 Act”) layered provisions on top of
the Bail Reform Act of 1966 for the District of Columbia, leaving much of
the 1966 Act in place but with three important changes.
First, the 1970 Act allowed courts to consider danger to the community in
setting nonfinancial conditions of release. As mentioned previously, prior to
this Act, court appearance was the only lawful purpose for limiting pretrial
freedom in the federal system (and, at least theoretically, all American
states). Pursuant to the Act, judges were still required to release noncapital
defendants on personal recognizance or an unsecured appearance bond, but
now the Act added the following language: “unless the [judicial] officer
determines . . . that such a release will not reasonably assure the appearance
of the person as required or the safety of any other person or the
community.” 158

155 See United States v. Edwards, 430 A.2d 1324, 1326 (D.C. Ct. App. 1981) (quoting Tribe, supra note 1,
at 397, 400-02 (1970)).
156 Pub. L. No. 91-358, 84 Stat. 473 (1970) (codified at D.C. Code Ann. §§ 23-1321-1332) [hereinafter
D.C. Act].
157 Pub. L. No. 98-473, 98 Stat. 1976 (1984) [hereinafter 1984 Act].
158 D.C. Act, supra note 156, § 23-1321 (a) (emphasis added).
69
While financial conditions were retained, a new provision declared that
“[n]o financial condition may be imposed to assure the safety of any other
person or the community.”
159 This prohibition was likely added to reflect the
facts that: (1) throughout the rest of the Act judges were given broad
authority to use nonfinancial conditions and even pretrial detention to
respond to public safety;
160 and (2) then, as today, financial conditions of
release have nothing to do with public safety.161
By leaving money in the process, however, the 1970 Act did nothing new to
avoid unintentional detention. Instead, the Act relied on the existing 1966
provisions dealing with review and appeal of unattainable conditions162
while, at the same time, adding provisions designed to separate pretrial
defendants detained through unattainable release conditions from convicted
persons within secure facilities.163 Moreover, we now know that leaving
money in the process also did nothing to change intentional detention,
despite the new provisions dealing with detention for public safety purposes,
discussed below, and despite the prohibition on using money for purposes of
public safety. Due to the ease of using money to detain defendants without a
hearing, the District of Columbia largely ignored the 1970 preventive
detention provisions and used money to detain until 1992, when it added
language requiring money bonds to be attainable.164
Second, the Act provided a procedure to detain noncapital defendants for
public safety. The issues surrounding the ability to detain for dangerousness
– significant issues concerning the right to bail, due process, and the Eighth
Amendment due to its collision with American notions of liberty –
overshadowed the fact that the Act rested on a dubious premise that the
detention of bailable noncapital defendants for flight was already
permissible and widely accepted in America. Because of this premise,
however, the Act contained no provisions for explicit detention based on

159 Id. § 23-1321.
160 Rauh & Silbert, supra note 140, at 288.
161 No research has ever shown a connection between money and public safety for released defendants and
in virtually every state, the money on a bail bond cannot even be forfeited for new crimes. Money only has
a connection to public safety when judges unlawfully use the financial condition to intentionally detain an
otherwise bailable defendant, which was one of Congress’s primary reasons for enacting the 1970 Act.
162 The appeals provisions were also modified to allow appeal after the new intentional detention provisions
as well as to give the government the right to dispute and appeal certain bail decisions. See D.C. Act, supra
note 156, § 23-1324 (c), (d) (1970).
163 D.C. Act, supra note 156, § 23-1321 (h).
164 See Pretrial Justice Institute, The D.C. Pretrial Services Agency: Lessons From Five Decades of
Innovation and Growth [hereinafter D.C. Lessons], at 5, found at
https://university.pretrial.org/viewdocument/the-dc-pretrial-serv.
70
flight. In a report accompanying the Act, the Congressional Committee on
the District of Columbia only mentioned in passing that, “Criminal
defendants today may be detained if found likely to flee regardless of the
conditions of release imposed,” but cited no authority whatsoever to support
the claim.
165 Moreover, and as noted previously, the authors of a concurrent
comprehensive law review article describing the Act made the somewhat
conclusory statement that, “The Bail Reform Act of 1966 permits the
detention of . . . noncapital defendants on grounds of flight.”
166 Because
those authors could not cite to language allowing detention based on flight in
the Act, they instead cited to Melville (discussed above), the case in which a
federal district court stated in dicta that there may be extreme cases in which
noncapital defendants could be detained for risk of flight.
Nevertheless, detention to address public safety was not the only purpose of
the 1970 Act. As noted in the Committee Report accompanying the Act,
pretrial detention was designed to reduce violent crime167 as well as to:
[E]liminate from the bail system the hypocrisy of locking up
defendants, without fixed standards, through the device of
requiring a high money bond. This second objective, of
removing the practice of detaining defendants arbitrarily by
setting a bond which they can not [sic] meet, is too often
overlooked when considering this question.168
Interestingly, the detention provisions were also added to address Congress’s
seemingly urgent fear that appellate courts would soon find current bail
practices unconstitutional. In a separate report submitted for consideration of
the 1970 Act, the authors wrote:

165 H. Rep. No. 91-907, at 92.
166 Rauh & Silbert, supra note 140, at 289.
167 There appears no doubt that violent crime – including violent crime committed by those released pretrial
– was of great concern to Congress and the District of Columbia at the time this law was enacted.
Nevertheless, a reading of the entire legislative history of the 1970 Act points to language suggesting at
least some hyperbole to instill fear (“The criminal is just not punished here, and continues to roam the
streets of Washington, increasing, month by month, his murders and his rapes, his robberies and aggravated
assaults, his housebreaking, larceny and auto thefts”), H. Rep. No. 91-907, at 15, as well as racism (a set of
narrative case histories, ostensibly chosen to show “glaring deficiencies in the bail system,” nonetheless
began each case with a heading titled, “Negro, Male.”). Id. at 95-104. The bail reform provisions were, in
fact, only “one of the many facets of this bill that seeks to provide some relief to the crime problems
besetting the District of Columbia,” including providing additional funding for courts and the pretrial
services agency as well as various other provisions dealing with, for example, wiretapping, recidivist
penalties, and an overall reorganization of judicial functions. Id. at 93.
168 Id. at 82.
71
It is inconceivable that for decades de facto detention through
high money bond and absent any procedural protections could
avoid constitutional condemnation, while a measured response
to bail recidivism fully surrounded by due process protections,
the net result of which will guarantee the release of many
persons wrongfully detained, will not pass Constitutional
muster.169
Indeed, in the primary house report to the 1970 Act, the Committee on the District
of Columbia similarly wrote as follows:
Ten years from now, court decisions based on equal protection
of the law may give the indigent defendant the means to force
his release before trial if money is the barrier between jail and
freedom. Such a development could not be welcomed by a
society besieged with crime unless that society were
empowered to protect itself against the truly dangerous
defendant. In the judgment of a majority of your Committee,
the only effective means of protection is pretrial detention.170
Through hindsight, we now know that allowing secured money bonds to
exist in our pretrial release and detention systems – including seemingly
well-engineered preventive detention provisions – has continued to interfere
with the process. Also through hindsight, we now know that states have
largely ignored the above quoted warning for 35 years, just as courts have
largely ignored equal protection analysis at bail, thus resulting in the
continuation of unfair and un-transparent pretrial detention based on
wealth.171
Third, the 1970 Act added provisions designed to deal with pretrial failure,
an extremely important concept given that risk is inherent in bail, and that in
America we are expected to embrace that risk by releasing as many
defendants as possible. The Bail Reform Act of 1966 contained provisions

169 Committee Print, Statement of the Managers on the Part of the Senate Submitted Regarding the
Conference Action Upon S. 2601, The President’s Crime Legislation for the District of Columbia, at 34
(July 15, 1970).
170 H. Rep. No. 91-907, at 85.
171 Only recently have we seen cases with the potential to force the release of defendants kept in jail due to
the lack of money. See, Equal Justice Under Law, Ending the American Money Bail System, found at
http://equaljusticeunderlaw.org/wp/current-cases/ending-the-american-money-bail-system/.
72
for willful failure to appear, but not for violation of other conditions of
release, and it also had what some considered to be an inadequate contempt
provision.172 The 1970 Act added detail to the contempt section and also a
new provision dealing with violating conditions of release that allowed for
revocation and an order of detention or for prosecution for contempt.173
Essentially, a judge could order a previously released defendant back to
detention if there were clear and convincing evidence of the violation and a
finding that no condition or combination of conditions would reasonably
assure court appearance or public safety.174
A Detention Eligibility Net and Further Limiting Process
For purposes of the present discussion, it is important to consider how the
“fix” enacted through the 1970 Act can help jurisdictions to discern where to
re-draw the line between release and detention today. The legislative history
to the Act mentions the need to expand detention to “selected defendants, in
categories of offenses characterized by violence,” “the most dangerous” of
defendants who commit crimes while on bail, and “dangerous defendants in
certain limited circumstances.”
175 It attempted to accomplish this purpose by
establishing initially a detention eligibility net, and then by articulating a
further-limiting process for defendants within the net, along with procedural
due process safeguards including hearings, time limits on detention orders,
and speedy trial guarantees.
The detention eligibility net included three categories of defendants. The
first category consisted of defendants charged with “dangerous crimes,”
defined at the time to include robbery by force or threat of force, burglary or
arson of a business or sleeping premises, forcible rape or assault with intent
to commit forcible rape, and unlawful sale or distribution of certain drugs.
176
Detention was allowed for a defendant in this category, but only if the
government certified that the defendant’s “pattern of behavior consisting of
his past and present conduct” along with existing factors to determine
conditions of release meant that “no condition or combination of conditions
[will] reasonably assure the safety of the community.”177 The requirement to

172 See Rauh & Silbert, supra note 140, at 301.
173 See D.C. Act, supra note 156, §§ 1329-1330.
174 Id. § 1329.
175 H. Rep. No. 91-907, at 82, 83, 91, 181.
176 See Rauh & Silbert, supra note 140, at 290.
177 D.C. Act., supra note 156, § 23-1322
73
consider the pattern of behavior was apparently designed to avoid basing
detention on charge alone.178
The second category consisted of defendants charged with a “crime of
violence,” defined at the time to include many more crimes than “dangerous
crimes,” including second-degree murder, forcible rape, carnal knowledge of
a girl under sixteen, taking or attempting indecent liberties on a child under
sixteen, mayhem, kidnapping, robbery, burglary, voluntary manslaughter,
extortion or blackmail accompanied by threats of violence, arson, assault
with intent to commit any offense, assault with a dangerous weapon, or an
attempt or conspiracy to commit any of the above offenses. Detention was
allowed for a defendant in this category, but only if the defendant had been
convicted of a crime of violence in the past ten years, was also currently
released on bail, probation, parole, or mandatory release pending a sentence,
or was a narcotics addict.179
The Committee Report noted that detention based on dangerousness was
restricted “to those charged with serious felonies which pose risk of death or
serious bodily harm to the victim,” many of which, the Report stated, were
punishable by death in 1791.180 While likely true, this statement concerning
capital punishment ignored the fact that most states in America gradually but
purposefully enlarged the right to bail to all but capital defendants while
simultaneously reducing the number of charges eligible for the death
penalty.181
The third category consisted of defendants charged with any offense, but
only if the defendant threatened or attempted to threaten, injure, or intimidate
a witness or juror.182 The House Report to the Act cited Carbo and Gilbert
(discussed above) for this proposition, but, also as noted above, pretrial
detention (versus detention after the trial had begun) to protect witnesses and
jurors, like flight, was only barely supported in the common law.
This detention eligibility net was further narrowed by a limiting process,
which included a due process laden hearing from which a judge was
required to conclude that: (1) there was clear and convincing evidence that

178 See Rauh & Silbert, supra note 140, at 291 n. 181.
179 See id. at 292-93 (explaining D.C. Code Ann. § 23-1322 (a) (2) as well as § 23-1323, which contained
the special provisions and processes for detention of an addict); DC Act, supra note 156, §23-1323.
180 See H. Rep. No. 907, at 93.
181 See Carbone, supra note 16, at 529-535.
182 D.C. Act, supra note 156, § 23-1322 (a) (3).
74
the person was eligible for detention; (2) based on the relevant factors, there
was “no condition or combination of conditions of release which would
reasonably assure the safety of any other person or the community;” (3)
except for people believed to be obstructing justice, there was substantial
probability that the defendant committed the offense charged.183
Overall, these provisions point to a “fix” that includes a narrow detention
eligibility net combined with a further limiting detention process, but with
provisions designed to deal with the failure that is inherent in bail. While not
perfect, it provided America’s first attempt to provide for purposeful release
and detention, erring on the side of release, and with nothing hindering the
judge’s decision either way.
The Bail Reform Act of 1984
The second phase of the “big fix” came 14 years later, when Congress
passed the Comprehensive Crime Control Act, which contained the Bail
Reform Act of 1984.184 Like the 1970 D.C. Act, the 1984 Act attempted,
once and for all and for the entire federal system, to provide an in-or-out
release and detention scheme by determining up front who would be
released or detained, with limitations on intentional detention while dealing
with cases presenting extreme risk of flight or public safety. Unlike the 1970
Act, however, it was also designed to end unintentional detention through
unattainable conditions altogether.
It did all of this through two particularly significant provisions, the first of
which was a radical limitation on money bail designed to end unintentional
pretrial detention.
185 To explain this, the reader should note that the 1984 Act
provided only four alternatives to judicial officers making the release or
detention decision: (1) release the defendant on a personal recognizance or
unsecured bond; (2) release the defendant on conditions; (3) temporarily
detain a defendant for certain reasons; and (4) detain the defendant fully
prior to trial. The rest of the Act worked through each of these four
alternatives. Obviously, release on personal recognizance or an unsecured
appearance bond did not cause unintentional detention, but traditionally the

183 Id. § 23-1322 (b).
184 Pub. L. No. 98-473, 98 Stat. 1976 (1984) (codified at 18 U.S.C. §§ 3141-3150).
185 For a complete discussion of the Bail Reform Act of 1984 and a comparison of that Act with the 1966
Act, see Judge Donald P. Lay & Jill De La Hunt, The Bail Reform Act of 1984: A Discussion, 11 Wm.
Mitchell L. Rev. 929 (1985) [hereinafter Lay & De La Hunt].
75
second alternative – release on conditions – did, and so the Bail Reform Act
added perhaps its most profound provision designed to prevent unintentional
detention from occurring. While still allowing for judicial officials to use
financial conditions, the 1984 Act nonetheless stated: “The judicial officer
may not impose a financial condition that results in the pretrial detention of
the person,”
186 a provision not found in the 1970 D.C. Act. Coupled with the
other provisions, this line virtually assured that defendants would not be
detained for lack of money to pay the financial condition.187 By adding this
line, Congress intended the Bail Reform Act of 1984 to be, at its core, an
intentional in-or-out system.
Second, the 1984 Act expressly articulated that preventive detention was
allowable for both risk of flight and public safety.188 As mentioned
previously, the 1966 Act said nothing about intentional detention of
noncapital defendants for flight, and so courts struggled through their
opinions to decide whether such detention was lawful. The D.C. Act of 1970
added detention based on public safety, but left out any express authority to
detain noncapital defendants for risk of flight.
189 The 1984 Act attempted to
clear up this overall confusion by expressly listing both public safety and
court appearance as proper purposes for limiting pretrial freedom up to and
including detention.

186 18 U.S.C. § 3142 (c) (1984).
187 Despite the prohibition, courts have occasionally determined that it is not a violation of the statute if
defendants are detained when they are unable to raise the money of the financial condition. These cases are
based on faulty reasoning, however, and are undoubtedly incorrect. See, e.g., United States v. Fiddler, 419
F.3d 1026, 1028 (9th Cir. 2005) (reasoning that when a defendant is unable to meet the financial condition
but the court has determined that the amount is sufficient, it is “not because [the defendant] cannot raise the
money, but because without the money, the risk of flight [or danger to others] is too great”). The proper
interpretation of this and other cases is found in an unpublished order in United States v. Clark, No. 1:12-
CR-156, 2012 WL 5874483 (W.D. Mich. Nov. 20, 2012) (memorandum detention order), in which the
court reveals that virtually every case upholding release orders with unmet financial conditions has only
done so because the unmet condition triggered a proper detention hearing, which follows the overall intent
of the Bail Reform Act. See id. at *3 (“If Fiddler were to be read to say only that a court may circumvent
the procedural safeguards of a full detention hearing by attaching heavy financial conditions to a release
order that a defendant could not meet, using as excuse that without such financial imposition the risk of
flight would be too great, the court would clearly be defying the intent of Congress and inviting a reexamination by that body of a court’s role in setting bond. Fortunately, the reading of the statute is seldom
so circumscribed.”).
188 See 18 U.S.C. § 3142 (e) (1984).
189 As noted previously, in the legislative history Congress mentioned only in passing that it felt detention
for flight was lawful. Later documents issued by the Department of Justice indicate that DOJ believed
intentional detention for noncapital defendants based on flight was clearly prohibited by the 1966 Act. See
Special Report — Pretrial Release and Detention: The Bail Reform Act of 1984, at 2 (BJS, 1988)
(“[D]etention without bail was permitted only in cases involving capital crimes.”), found at
https://www.bjs.gov/content/pub/pdf/prd-bra84.pdf.
76
Much like the legislative history surrounding the 1970 Act, Congress
justified its authority to detain noncapital defendants pretrial in 1984 on
fairly amorphous authority. In the Senate Report accompanying the Act,
Congress wrote as follows:
The decision to provide for pretrial detention is in no way a
derogation of the importance of the defendant’s interest in
remaining at liberty prior to trial. However, not only the
interests of the defendant, but also important societal interests
are at issue in the pretrial release decision. Where there is a
strong probability that a person will commit additional crimes if
released, the need to protect the community becomes
sufficiently compelling that detention is, on balance,
appropriate. This rationale – that a defendant’s interest in
remaining free prior to conviction is, in some circumstances,
outweighed by the need to protect societal interest – has been
used to support court decisions, which, despite the absence of
any statutory provision for pretrial detention, have recognized
the implicit authority of the courts to deny release to defendants
who have threatened jurors or witnesses, or who pose
significant risks to flight. In these cases, the societal interest
implicated was the need to protect the integrity of the judicial
process. The need to protect the community from demonstrably
dangerous defendants is a similarly compelling basis for
ordering detention prior to trial.190
For the proposition that detaining defendants for threatening jurors or
witnesses provides justification for detention based on safety to the broader
public, in 1984 Congress cited to Wind and Gilbert, which, as discussed
previously, provide at least some authority for detaining defendants for that
purpose. But as to flight, Congress said it was only “codify[ng] existing
authority to detain persons who are serious flight risks”191 and cited only to
Abrahams, also discussed above. As mentioned previously, though,
Abrahams rested on dubious authority itself, and never found its way beyond
mostly mere mention within the First Circuit Court of Appeals. Thus, as
seen from the quote above, one of Congress’ primary legal justifications for
allowing the intentional detention of noncapital defendants pretrial for

190 S. Rep. 98-225, at n. 27, 28 and accompanying text (1984) (internal footnotes omitted).
191 Id. at n. 63 and accompanying text.
77
purposes of public safety – justification based on the fact that America
already allowed it for flight – appears fairly slim.
It is important to note that due to the significant debate surrounding whether
America could ever detain a person for purposes of public safety, we often
gloss over the notion that “preventive detention” as a concept involves
detaining someone preventively for either flight or public safety. To this
day, people inaccurately describe preventive detention as something only
done to address danger.
192 This is likely due to a number of factors,
including, ironically, the fact that historically when a defendant was bailable,
he or she was supposed to be actually released. Accordingly, when
jurisdictions first began discussing preventive detention in the 1960s, it was
assumed that risk of flight simply could not be used to detain persons
beyond those extremely narrow “categorical” crimes, such as capital
offenses. Thus, preventive detention began with the notion that it would be
used only for public safety. That notion, however, has gradually changed,
beginning with the detention cases, and continuing up and through the 1984
Act.
Indeed, as the 1984 Act illustrates, when courts detain today for risk of
flight, they are detaining preventively. Even in states having so-called
“broad right to bail” provisions that, for example, grant the right to release to
all but capital defendants, those states are still correctly described as having
systems of preventive detention – historically based on flight for capital
defendants. Thus, the primary novelty of the detention cases and the “big
fix,” discussed above, is not in the creation of “preventive detention.”
Rather, the novelty of the cases is in the gradual extension of preventive
detention pretrial for flight to noncapital defendants. The novelty of the “big
fix” is further extending preventive detention to defendants for purposes of
public safety and in ultimately attempting to eliminate unintentional
detention.
Nevertheless, by citing to Wind, Gilbert, and Abrahams, Congress provided
some intent concerning how limited detention should be. As noted
previously, each of those cases cautioned that a judge’s inherent authority to

192 Indeed, in Lopez-Valenzuela v. Arpaio, 770 F.3d 772, at 798-99 (9th Cir. 2014), the dissent argued that
Salerno and its discussion of due process could only provide a basis for evaluating a “no bail” provision
based on dangerousness and not flight. As Judge Fisher correctly noted in the majority opinion, however,
“the Supreme Court has never recognized – or even suggested – that distinction. Id.at 792, n.16 (citing
cases).
78
detain pretrial should be used only in “extreme and unusual cases,”
exercised with great care and only after a due process hearing.
193 Moreover,
the facts of Abrahams, also recounted above, are particularly helpful in
telling us just how extreme the risk of flight should be. Indeed, throughout
the legislative history of the 1984 Act, Congress repeatedly said it was
reserving pretrial detention for those posing “serious” risks of flight or new
criminal activity.
194 More specifically, it was reserving pretrial detention
based on public safety to a “small but identifiable group of particularly
dangerous defendants” who pose an “especially grave risk” to the
community and for whom neither conditions nor the prospect of revocation
suffice to protect the public.195
A Detention Eligibility Net and Further Limiting Process
Like the 1970 Act, Congress sought to operationalize these terms by creating
a narrow net for detention eligibility with an additional limiting process with
procedural safeguards. Unlike the 1970 Act, however, the new law
significantly broadened that net, and incorporated the use of “rebuttable
presumptions” leading toward detention in certain cases.196 As noted by one
federal appellate court, the 1984 Act made it both harder and easier to
detain:
It [made] it harder by specifying explicitly what was implicit in
prior law, namely that magistrates and judges cannot impose
any ‘financial condition’ that will result in detention. High
money bail cannot be used as a device to keep a defendant in
custody before trial. The Act [made] detention easier by
broadening the category of persons whom the officer can order
detained.197
Under the 1984 Act, defendants potentially eligible for detention fell into
six, not three categories. First, a defendant might be detained if he was
charged with a “crime of violence.”198 In 1984, a crime of violence was

193 See United States v. Gilbert, 425 F.2d 490, 491 (D.C. Cir. 1969); United States v. Wind, 527 F.2d 672,
675 (6th Cir. 1975); United States v. Abrahams, 575 F.2d 3, 8 (1st Cir. 1978).
194 S. Rep. No. 98-225, at 5, 10, 16, 21.
195 Id. at 6-7, 10, 20.
196 See 18 U.S.C. §§ 3142 (e).
197 United States v. Jessup, 757 F.2d 378, 379 (1st Cir. 1985) (internal citation omitted) (abrogated on other
grounds).
198 18 U.S.C. § 3142 (f) (1) (A) (1984).
79
defined as “an offense that has as an element of the offense the use,
attempted use, or threatened use of physical force against the person or
property of another” or “any other offense that is a felony and that, by its
nature, involves a substantial risk that physical force against the person or
property of another may be used in the course of committing the offense.”199
This definition was shaped by court opinions, but would likely have
included most of the crimes listed under the 1970 Act’s net relating to
“dangerous” and “violent” crimes.
Second, a defendant might be detained if he was charged with an offense for
which the maximum sentence was either life imprisonment or death.200
Third, a defendant might be detained if he were charged with certain serious
drug offenses with sentences of ten years or more.201 Fourth, a defendant
might be detained in any case in which he posed “a serious risk that [he
would] obstruct or attempt to obstruct justice, or threaten, injure or
intimidate, or attempt [to do the same to] a prospective witness or juror.”202
These four relatively narrow categories mirrored somewhat the 1970 Act’s
net, albeit lacking additional narrowing elements such as requiring the
government to certify certain conduct, or requiring certain preconditions,
such as the defendant currently being on pretrial release, probation, or
parole.
The next two detention eligibility categories, however, represented a
significant broadening of the net found in the 1970 Act. Category five
allowed a court to detain a defendant for any felony after conviction of two
or more crimes like those found in the first three categories.203 The sixth and
final category allowed a court to detain a defendant if he presented “a
serious risk that [he would] flee.”204 While perhaps flowing naturally from
the gradual erosion of early American law that rarely (if ever) expressly
allowed any intentional detention of noncapital defendants for risk of flight,
the 1984 Act’s allowance of pretrial detention based on a “serious” risk of
flight still represented a major shift.
Nevertheless, these wider nets were likely made necessary by Congress’s
equally important goal of eliminating unnecessary (or unintentional) pretrial

199 Id. § 3156.
200 Id. § 3142 (f) (1) (B).
201 Id. § 3142 (f) (1) (C).
202 Id. § 3142 (f) (2) (B).
203 Id. § 3142 (f) (1) (D).
204 Id. § 3142 (f) (2) (B).
80
detention – as Congress said in 1966, to remedy “the evils which are
inherent in a system predicated solely upon monetary bail”205
– and the
hypocrisy of sub rosa and standardless detention caused by money bail.
206
Indeed, it is likely that no state in the future can avoid making a similar
choice. When money is taken out of the process – i.e., when unlawful and
standardless detention is removed – it focuses one’s attention on who,
exactly, should be released and detained, and that question is initially
answered by the detention eligibility net. The less people know about risk in
general, the wider that net will likely be.
The 1984 Act also augmented these detention eligibility categories with socalled “rebuttable presumptions” toward detention based on certain
preconditions. Like the 1970 Act, the 1984 Act included a limiting process
designed to further narrow the net of detention eligible defendants, which
included a due process laden hearing from which a judge must find based on
the relevant factors that no condition or combination of conditions would
reasonably assure court appearance or public safety.207 Nevertheless, the
1984 Act also added a rebuttable presumption that no conditions or
combination of conditions would suffice to protect the public in cases in
which: (1) the defendant had been previously convicted of a crime listed in
the first three net categories (or any state or local offense equivalents); (2)
the offense for that conviction was committed while the defendant was on
pretrial release; and (3) no more than five years had elapsed since the date of
that conviction or release from imprisonment, whichever is later.208 At least
two legal scholars predicted, correctly, that the set of circumstances found in
this first rebuttable presumption was not expected to happen often.209
A second rebuttable presumption, though, pointed toward a finding of “no
conditions or combination of conditions” for both flight and public safety in
cases in which the defendant was charged with felonies punishable by ten
years or more of imprisonment covering certain serious drug cases or use of
a firearm to commit a felony.
210 According to the Senate Report, these were
considered by Congress to be “serious and dangerous offenses” committed
by defendants that pose a “significant risk” both for pretrial crime and, in the

205 H.R. Rep. 1541, 89th Congress, 2nd Sess. 8-9 (1966).
206 See H. Rep. 91-907, at 82; 98-225, at 5 (1970).
207 18 U.S.C. § 3142 (e). For a list of the elements used in this process, see the discussion on United States
v. Salerno, infra.
208 See id. at 3142 (e) (1), (2), (3) (1984); see also S. Rep. 98-225, at 19.
209 See Lay and De La Hunt, supra note 185, at 940.
210 See 18U.S.C. § 3142 (e) (1984).
81
case of drug offenders, for flight or escape to other countries.211 This
presumption was predicted to be triggered far more often.
United State v. Salerno
The “big fix” was given legal affirmation in the case of United States v.
Salerno,
212 discussed previously. In Salerno, the United States Supreme
Court concluded that the Bail Reform Act’s detention provisions did not
facially violate the Due Process Clause or Excessive Bail Clause of the
United States Constitution. In so doing, the Court made it clear that: (1)
public safety was a constitutionally valid purpose for limiting pretrial
freedom; (2) in certain circumstances pretrial detention could be used based
on predicted risk of danger; but (3) if used, pretrial detention had to be both
extremely limited and fair.213
Accordingly, the Court emphasized the importance of the various limits
concerning detention that might serve as narrowing functions.
214 As to the
detention eligibility net, the Court stated that the Act “carefully limits the
circumstances under which detention may be sought to the most serious of
crimes (detention hearings available if case involves crimes of violence,
offenses in which the sentence is life imprisonment or death, serious drug
offenses, or certain repeat offenders).”215 The Court’s use of the phrase
“most serious” accords not only with the legislative history of the Act, which
was directing detention toward a “small but identifiable” group of
defendants posing “especially grave risks,” but also with the holdings from

211 S. Rep. No. 98-225, at 19-20.
212 481 U.S. 739 (1987).
213 Id. Another interesting, but lesser known aspect of Salerno is that it essentially obviated the need to
consider safety to witnesses or jurors – sometimes tied to an articulated purpose of “ensuring the integrity
of the judicial process” – separately from any other notions of public safety. In Salerno, the principle
contention at the court of appeals level was that the Bail Reform Act of 1984 violated due process because
it permitted pretrial detention of defendants when their release would pose a danger to the community or
any person. The court of appeals expressly noted that it considered this contention to be wholly different
from what it considered to be clearly established law that pretrial detention was proper to prevent flight or
threats to persons solely within the judicial process, such as witnesses and jurors. See United States v.
Salerno, 794 F.2d 64 (2nd Cir. 1986). In the brief before the Supreme Court, the government highlighted the
anomaly presented by a scheme that would allow courts to detain persons deemed high risk to witnesses
and jurors, but not to other members of the public. See Brief for Appellee at 13, United States, Salerno, 481
U.S. 739 (1987). Had the Supreme Court not reversed the appellate court, this distinction between those
within the judicial process and those outside of it might have remained. Instead, by upholding the 1984 Act,
the Supreme Court forever expanded the notion of public safety to encompass all potential victims, whether
in or out of the judicial process.
214 The Harvard Law School Primer, supra note 3 at 27, correctly and uniquely calls these narrowing
functions “limited entry points” to any scheme of preventive detention.
215 Salerno, 481 U.S. at 747.
82
the detention cases, most of which used strong limiting adjectives
concerning risk – such as “extreme or unusual” – to keep detention within
proper boundaries. Obviously, and as discussed later, our notions of which
defendants pose which risks have been altered by current risk research.
While approving a fairly broad expansion of pretrial detention, Salerno
nonetheless provided states with the oft-quoted line that can be used to guide
American jurisdictions on the proper formulation of detention eligibility nets
and limiting processes: “In our society, liberty is the norm, and detention
prior to trial or without trial is the carefully limited exception.”216 This line is
important given the fact that American jurisdictions might differ on what
they consider to be “the most serious of crimes” or even “high risk
defendants.” Even the most crime-fearing and risk-averse jurisdictions must
come to accept that fundamental American notions of liberty require many
more persons to be released pretrial than detained. Of course, the “norm” is
not defined, which is why we must learn from history, the law intertwined
with that history, and even the facts of various detention cases to determine
the levels of risk that we must embrace.217 Jurisdictions should note an
overall theme arising from the detention cases, to the “big fix,” to Salerno,
which is that preventive pretrial detention was designed to be used only in
extremely unusual cases, on a small but identifiable group of highly risky
persons, and as a carefully limited exception to an overall presumption of
release.
Some persons argue, incorrectly, that Salerno’s statement concerning the
“norm” reflects some higher philosophical notion of American freedom. In
particular, they argue that we must look at every single defendant –
including those released on citations or summonses – before we decide
whether liberty has been preserved for most persons. But Salerno was not
decided abstractly; the Court was reviewing the detention provisions of the
Bail Reform Act of 1984, which were only triggered when persons were
arrested, charged, and brought before a judicial officer to determine release
or detention.218 Thus, when measuring release and detention under the
“norm” standard, jurisdictions should look, at least initially, at its arrested

216 Id. at 755.
217 The Court’s use of the word “norm” was likely prompted by the ACLU’s amicus brief in the case, in
which that organization argued that the government appeared to be advocating “that regulatory detention is
the norm, rather than the exception.” Amicus Curiae Brief of the ACLU, at 8, United States v. Salerno, 481
U.S. 739 (1987).
218 See 18 U.S.C. § 3141, 3142 (1984).
83
population. Importantly, however, this notion might change given certain
advances in pretrial justice.
219
The Court also emphasized the importance of the Act’s detention limiting
process and procedural safeguards, which included: (1) a “full-blown
adversary hearing,” with counsel, the ability to proffer evidence, witnesses,
and cross examination; (2) judicial guidance through standards; (3) a
requirement that the judicial official only detain after finding by clear and
convincing evidence that the defendant “presents an identified and
articulable threat” and that no condition or combination of conditions suffice
to provide reasonable assurance of public safety or court appearance; (4) a
requirement of a written findings for detention; and (5) the ability for an
immediate and expedited appeal.220
In Foucha v. Louisiana, a commitment case for a defendant found not guilty
by reason of insanity, the Supreme Court described the Bail Reform Act as a
“sharply focused scheme,” which stressed all these procedural elements,
including the limited duration of detention as well as need for the
government “to convince a neutral decision maker through clear and
convincing evidence that no conditions of release can reasonably assure the
safety of the community or any person, i.e., that the ‘arrestee presents an
identified and articulable threat to an individual or the community.’”221 In
her concurrence, Justice O’Connor noted that without concrete evidence of
dangerousness – such as a criminal conviction – courts should pay deference
to reasonable legislative judgments about dangerousness,222 but nonetheless
stressed that Salerno allowed pretrial detention only when “individuals
arrested for ‘a specific category of extremely serious offenses’ are detained
and ‘Congress specifically found that these individuals are far more likely to
be responsible for dangerous acts in the community after arrest.’”
223

219 For example, if a police agency were to use an actuarial pretrial risk assessment instrument on the street
and make the decision to arrest based on the instrument, the dynamic of the jail population might change.
This would be a rational system for determining whether to arrest someone, and yet it could potentially lead
to the arrest of only high risk persons. In such a case, a more rational system that reduces needless arrests
might be put in place but lead to decisions that – again, rationally and legally defensibly – cause liberty for
that population to be less than the “norm.” In one respect, the “norm” term represents a ratio. And thus,
once again, it is likely the rationality of the process that will ultimately determine the appropriate ratio.
220 Salerno, 481 U.S. 739, 742, 750-52 (1987).
221 Foucha, 504 U.S. 71, at 72 (1992).
222 Id.at 81.
223 Id.at 88.
84
When reviewing Salerno as a part of America’s “big fix,” jurisdictions must
also remember that the case presented a “facial” challenge to the Bail
Reform Act against Due Process and Eighth Amendment claims. A facial
challenge, as the Court noted, presents a “heavy burden,” and “is the most
difficult challenge to mount successfully, since the challenger must establish
that no set of conditions exists under which the Act would be valid.”224 Put
another way, if even one conceivable set of circumstances exists under
which the law would operate constitutionally, a facial challenge fails. Losing
a facial challenge, however, does not erode the very real possibility of
appellate courts finding individual federal cases being decided
unconstitutionally for failure to follow the various elements within the 1984
Act, and this would be true for any state that attempts to emulate the Act
through its own bail laws. For example, even if a state were to enact a
process virtually identical to 1984 scheme, the fact that a court within that
state might neglect one element – such as having defense counsel present at
the detention hearing, or, indeed, having the hearing at all – could lead to an
appellate court to declare a constitutional violation.
This is not to say that the only way to follow the Constitution is to enact a
process identical to the 1984 Act. For example, due to differences in federal
and state law, a state might provide a broader list of detention eligible
offenses than those covered under the Act.225 Moreover, it seems unlikely
that the Supreme Court would find fault in a state’s new detention eligibility
net based on empirical research demonstrating a justification for the net.
Nevertheless, Salerno highlights two fundamental problems – one current
and one future – facing America today.
Currently, the biggest problem appears not to be that jurisdictions differ over
details in their bail schemes; instead, the problem is that despite the states’
recognition of Salerno and its emphasis on limited detention, they routinely
ignore fundamental principles embodied in Salerno, which should lead –
and, indeed, have already led – to findings of constitutional violations. For
example, in Lopez-Valenzuela v. Arpaio the Ninth Circuit Court of Appeals
struck an Arizona detention provision because it was not “carefully limited,”
as mandated by Salerno.226 Similarly, the Arizona Supreme Court recently
struck a state constitutional detention provision as violating Salerno’s
requirement that provisions be “narrowly focused” on preventing the stated

224 Salerno, 481 U.S. at 745.
225 See LaFave, et al., supra note 52, § 12.3 (d), at 73, n. 96.
226 770 F.3d 772 (9th Cir. 2014)
85
harm.227 Bail scholars have predicted these and other potential constitutional
objections to various aspects of pretrial detention for some time.
228
Indeed, most state detention schemes would likely fail if merely held up to
Salerno, and the risk of constitutional violations found in individual cases
appears even greater. This is due, primarily, by the fact that the “big fix” did
not spread in any meaningful way to the states. After the D.C. Act of 1970,
states added references to public safety in bail setting fairly quickly. Even
before the Bail Reform Act was reviewed for constitutionality by the United
States Supreme Court, some states even changed their constitutional right to
bail provisions to allow for the denial of bail or release to a larger class of
defendants.229 Like the release provisions from the 1966 Act, states often
enacted the detention provisions from the 1984 Act only in part, or in some
perverted form by allowing for detention without the necessary due process
hearings. In many cases, fairly decent preventive detention provisions –
often resembling the federal law – are ignored, with judicial officials relying
instead upon the ease in which money detains. In short, many American
jurisdictions have not learned the lessons of Salerno, and instead have
apparently come to believe that the case gives states broad latitude to detain
in any way they see fit.
The big problem in the future appears to be how to take Salerno’s
fundamental principles and apply them given what we know today about
risk. The bail scheme reviewed by Salerno, and indeed, virtually every other
bail scheme in America, was based on certain assumptions, such as an
assumption that a serious charge meant the defendant posed a serious risk.
As we will see later, the risk research is causing us to re-think the
assumptions used in creating our current bail laws, and to reword them to
provide rationality today. While tempting to think that we can simply switch
from a “charge-based” detention scheme to a “risk-based” one, current
limitations in risk research mean that we must resist that temptation.
These are significant problems, but they seem less so compared to a broader
issue facing America. That issue concerns how to stop the continued growth
of pretrial detention when pretrial detention tends only to prove itself. The

227 Simpson v. Miller, 387 P. 3d 1270 (Ariz. 2017).
228 See, e.g., LaFave, et al., supra note 52, at § 12.3 (d), at 79 (“Starting with Salerno, the contention might
be that all of these state constitutional provisions violate substantive and/or procedural due process unless
the requirements emphasized in that case are engrafted onto these provisions.”).
229 See Goldkamp, supra note 3, at 15-16 (1985); Fagan & Guggenheim, supra note 3, at 415, 417-18.
86
risk of over-detaining – using money, charge, or risk – is real for either
public safety or flight, and broadening our ability to detain by any of these
measures must be curtailed by scrutinizing detention models through the law
and the research. Money, of course, over-detains in ways that are clearly
unconstitutional. But risk, too, can pose similar problems. Once we say that
every defendant is “risky” – as we do now with actuarial pretrial risk
assessment instruments – what is to keep us from gradually detaining
unconstitutional numbers of defendants, especially when detention leads to
the outcomes we seek?
Accordingly, when re-drawing the line between pretrial release and
detention, jurisdictions must remember what the law and the history of bail
tell us. Specifically, they tell us that the English system basically involved
assessing pretrial risk before declaring anyone bailable. America gradually
changed that system to one in which people were declared bailable up front,
allowing judges only to consider various risk factors in setting the amount of
the financial condition. At first, the system worked well and did not offend
the historic rule that bailable defendants must be released. Later, however,
when courts began seeing relatively higher flight and public safety risks,
those courts struggled with how to deal with both unintentional and
intentional detention. Accordingly, in the 1970s and 1980s, America needed
some fix that would provide a purposeful system of release and detention
based both on risk for flight as well as public safety. That fix needed to
provide a narrow detention eligibility net and further narrowing process to
reflect American notions of liberty, while still allowing the intentional
detention of “high” risk defendants, and while eliminating unintentional
detention by reducing the effects of money at bail.
Because that fix involved a significant overall expansion of pretrial
detention – a stark departure from earlier and much more limited release and
detention notions230
– the purposeful narrowing of detention to encompass
only extremely serious public safety and flight risks through a detention
eligibility net, a further narrowing process, and other due process safeguards
theoretically lessening the overall use of detention as a response to risk, was
a pivotal part of the solution. Unfortunately, however, the interrelated parts

230 In a comprehensive review of various modes of American preventive detention, the authors note that the
Bail Reform Act of 1984 tended not to follow the other modes of preventive detention, which greatly
narrowed what had been traditionally broad common law powers. See Adam Klein & Benjamin Wittes,
Preventive Detention in American Theory and Practice, 2 Harv. Nat’l Security J. 85 (2011) [hereinafter
Klein & Wittes].
87
to this “fix” did not successfully spread to the states, and have become
eroded to near unrecognizable levels in the federal system.
Future attempts at re-drawing the line between release and detention can be
informed by knowledge of the law and history, which both point toward
using information about risk to maintain America’s emphasis on liberty and
freedom while rationally addressing historic fears over flight and crime.
What Does the Pretrial Risk Research Tell Us About ReDrawing the Line Between Pretrial Release and Detention?
Pretrial research in all its forms (historical, legal, observational, opinion,
social science, etc.) significantly informs the field of pretrial release and
detention. By far, however, social science research – and specifically,
research concerning defendant risk – provides us with the most compelling
data for helping to re-draw the line between pretrial release and detention.
Indeed, this is one of the quintessential questions in bail and no bail today;
can risk itself serve to draw the line between release and detention, replacing
a line previously drawn (albeit somewhat ignored) by defendant charge as a
proxy for risk? In many jurisdictions, people argue the need to detain “high
risk” defendants, but this only leads to further questions. Questions such as,
what do we mean by “high” risk?; should we then release everyone who is
not “high’ risk”?; should we detain a “high” risk person even if that person
is only charged with, say, shoplifting?; concomitantly, should we release a
“low” risk person even if that person is charged with, say, murder?; and
“high” or “low” risk for what, exactly?
These questions lie at the very heart of this generation of bail reform.
Jurisdictions now know more about defendant risk than they ever knew in a
mostly charge and money-based system. Accordingly, the overarching
question becomes whether risk research, and especially the research going
into the development of actuarial pretrial risk assessment instruments,
provides us with information helpful to re-drawing the line between release
and detention. To answer that question, it is helpful to know four things
about actuarial pretrial risk assessment instruments, including: (1) what
these instruments tell us; (2) how these instruments highlight certain
significant flaws in our current system by illuminating often counterintuitive
outcomes; (3) what these instruments do not tell us; and (4) how actuarial
pretrial risk assessment and the risk research interact with the law.
88
Before we discuss these concepts, however, we must remember that “risk
assessment” is not new. We have been assessing risk at bail since at least
400 A.D., when the first group of Saxons assigned a personal surety (usually
a family member) to protect the property due as a penalty for some wrong if
the accused (or the “convicted” offender) were to flee. Since then, we have
always been concerned with risk. There is a tendency today to speak of the
advent of “risk assessment” as some new technology to solve the world’s
problems with bail, but that is incorrect. Today’s statistically-derived, multijurisdictional risk tools are simply the latest in a long line of historical ways
to assess risk.231 Indeed, every American jurisdiction today attempts to
assess risk – perhaps through intuition, or a variety of statutory factors, or
even a money bond schedule looking only at criminal charge – but it is
assessment nonetheless. Today’s statistical methods are superior to that;
indeed, they are so superior that they, alone, are likely responsible for much
of this current generation of bail reform. It is helpful to know, however, that
previous generations of risk assessment either contained or exacerbated
certain limitations to the task surrounding bail, and so we should not be
surprised that this new generation contains similar limitations. Moreover, we
should not be surprised to find that despite being exceptional at helping with
release and conditions of release, actuarial risk tools are only partially
helpful in assessing overall risk for the things we hope to address with
detention.
It helps to consider a simple thought experiment. If, in the future, America
developed an accurate way to determine, with 100% accuracy, that a certain
defendant would definitely commit an aggravated murder on a date certain,
we would likely detain that defendant and detention, in that case, might
appear infinitely reasonable. If, however, this form of risk assessment told us
that the crime he was going to commit was simple trespass, we might
reconsider detention altogether and try to fashion conditions designed to
dissuade him from ultimately making that choice. Moreover, if we knew that
a defendant was going to commit a crime on a date certain, there are likely
other things we could do outside of conditions to avoid that crime from
happening while on pretrial release, such as moving up the court date, or

231 See generally NIC Fundamentals, supra note 6. As noted in that document, a significant shift in risk
assessment occurred in 2003 with the creation, by Dr. Marie VanNostrand, of a multi-jurisdictional,
statistically-derived risk assessment tool, the Virginia Pretrial Risk Assessment Instrument. For a general
overview of risk and risk tools, see Anne Milgram, Alexander M. Holsinger, Marie VanNostrand, &
Matthew W. Alsdorf, Pretrial Risk Assessment: Improving Public Safety and Fairness in Pretrial Decision
Making, 27 Fed. Sentencing Rep., No. 4, at 216 [hereinafter Milgram, et al.].
89
simply using police surveillance as is done in virtually every other attempt to
thwart crime.
Similarly, with flight, it would depend on whether the instrument predicted
true flight to avoid prosecution versus simply forgetting a court date, and
even then the risk might disappear by simply reducing the number of court
appearances. The point is that even in a perfect system illustrating 100%
infallible risk prediction, we might still refrain from entertaining the idea of
detention because of our notions of what it means to be an American.
Current risk assessment is far from this perfect system, and so we should not
be surprised when the law – erring on the side of liberty and freedom –
trumps findings of risk based on today’s actuarial assessments.
What Do Actuarial Pretrial
Risk Assessment Instruments Tell Us?
At their core, actuarial pretrial risk assessment instruments attempt to predict
the risk of a defendant “failing” through misbehavior while on pretrial
release. In America, the two types of misbehavior the government is allowed
to address at bail are court appearance and new criminal activity, and so
these risk instruments help us to determine defendant risk for these two
outcomes as well as to guide us toward appropriate interventions designed to
reduce that risk.232 For example, if we know that a defendant is relatively
risky for failing to appear for court, we can place conditions on his or her
release designed to help provide reasonable assurance of court appearance.
Again, this generation of risk assessment using actuarial tools is better than
any other efforts of assessing risk we have done in the past, 233 and the

232 See generally Milgram, et al., supra note 231; Pretrial Justice Institute, Pretrial Risk Assessment:
Science Provides Guidance on Assessing Defendants (PJI, 2015) [hereinafter PJI Risk], found at
https://nicic.gov/library/029999.
233 For a good review of the history and then-current research of pretrial risk assessments as well as a
description of how one was constructed, see Christopher T. Lowenkamp, Richard Lemke, & Edward
Latessa, The Development and Validation of a Pretrial Screening Tool, 72 Fed. Prob. 2 (2008); For a
general description of risk assessment generations, see Sarah L. Desmarais & Jay P. Singh Risk Assessment
Instruments Validated and Implemented in Correctional Settings in the United States: An Empirical Guide
(CSG, 2013) [hereinafter Desmarais & Singh]; the American Bar Association Standards on Pretrial Release
trace the development of empirical risk since the 1920s, ending with VanNostrand’s Virginia instrument,
see ABA Standards, supra note 100, Std. 10-1.0 (b) (i) (commentary), at 57, n. 22; see also Council of
State Governments, Risk Assessment: What You Need to Know (2015) (“Risk assessments are absolutely,
statistically better at determining risk than the old ways of doing things.”), found at
https://csgjusticecenter.org/reentry/posts/risk-assessment-what-you-need-to-know/; Charles Summers &
Tim Willis, Pretrial Risk Assessment: Research Study (BJA, 2010) [hereinafter Summers & Willis], found
at https://www.bja.gov/Publications/PretrialRiskAssessmentResearchSummary.pdf.
90
literature suggests that they are significantly better than clinical (i.e., largely
subjective) prediction. As noted by researchers Sarah Desmarais and Jay
Singh, “There is overwhelming evidence to suggest that assessments of risk
completed using structured approaches produce estimates that are both more
accurate and more consistent across assessors compared to subjective or
unstructured approaches.”
234
This is true in the pretrial setting, and using these tools – empirically-based
actuarial instruments that classify defendants by differing levels of pretrial
risk through weighted factors – is considered to be an evidence-based
practice, and is often a critical prerequisite to adopting other best practices in
the pretrial field.235 These assessment instruments provide standardization
and transparency, help avoid arbitrary decision making, and help to
maximize our pretrial goals. Moreover, by telling us pretrial risk in a more
accurate way, these tools also help us to follow the so-called “risk
principle,” which instructs jurisdictions to expend less or no resources on
lower risk persons and more resources on higher risk persons, and which
thus includes a requisite finding of risk to allocate resources properly. The
risk principle is well known in the post-conviction field, and has equal
relevance to pretrial release and detention decisions.
236
Overall, actuarial pretrial risk assessment instruments are invaluable to the
process, and, by themselves, an exceptional justification for eliminating
money at bail. They can help courts and justice systems with virtually all
issues concerning release (including structuring and evaluating supervision
strategies, crafting responses to violations, assessing the efficacy of bond
“types,” evaluating jail populations, helping to encourage more summonses
and citations and even providing some rationale to emergency releases,
when necessary), and they can assist with detention. Using them can even
lead to more confidence in data processes and systems policies. Moreover,
they are always improving; as noted previously, the Arnold Foundation’s
pretrial risk assessment tool, developed in 2013, has various attributes
tending to ameliorate many of the concerns from previous generations.

234 Desmarais & Singh, supra note 233, at 1 (citing Stefania Aegisdottir, et al., The Meta-Analysis of
Clinical Judgment Project: Fifty-Six Years of Accumulated Research on Clinical Versus Statistical
Prediction, 34 The Counseling Psychologist, 341 (2006)).
235 See generally PJI Risk, supra note 232.
236 Milgram, et al., supra note 231, at 216-17; Marie VanNostrand, & Gena Keebler, Pretrial Risk
Assessment in the Federal Court [hereinafter VanNostrand & Keebler] (Washington, D.C.: Office of
Federal Detention Trustee, 2009), found at
https://www.ncjrs.gov/App/publications/abstract.aspx?ID=250813.
91
Using these tools to better follow the law, by helping courts to determine
reasonable assurance of court appearance and public safety, by making sure
that pretrial liberty on least restrictive and otherwise lawful conditions is the
“norm” (with no intentional or unintentional pretrial detention outside of any
particular lawful process for doing so), by assuring that pretrial detention is
done in a “carefully limited” way,237 and by helping courts to follow other
fundamental legal principles, is a legal and evidence-based practice, the very
thing that American jurisdictions are attempting to achieve in the pretrial
field. Pretrial risk assessment tools are not designed to replace professional
discretion, but rather to enhance pretrial decision making by coupling
instinct or experience and objective assessment, which research has shown
produces the best results.238
How Actuarial Pretrial Risk Assessment Instruments Are Created
To understand whether we can use these actuarial tools to re-draw the line
between release and detention, it is important to know how they are created,
and we will do so using the Colorado Pretrial Assessment Tool (CPAT) as a
primary example, as virtually all pretrial risk assessment instruments in use
today are similar to the CPAT in their construction. Overall, an actuarial
pretrial risk assessment instrument uses scientific methods and data
collection to test variables for their predictability of certain relevant
outcomes, which, in the pretrial field, are failure to appear for court and new
criminal activity (and, occasionally, failure to abide by other conditions).
While a group of researchers might test hundreds of potential variables (such
as previous FTAs, level of charge, etc.) the end result is a set of variables
(such as 8 in Virginia, or roughly 70 in Washington, D.C.) that, when used
together and in the manner in which they are weighted, are best at predicting
the two outcomes that the law allows us to consider pretrial. For example, in
creating the CPAT, the researchers tested over 150 variables resulting in a
tool having 12 factors, which, when weighted and considered together,
provide the best set of factors to predict risk.239

237 United States v. Salerno, 481 U.S. 739, 750 (1987).
238 See Milgram, et al., supra note 231, at 219-20.
239 See Pretrial Justice Institute/JFA Institute, The Colorado Pretrial Assessment Tool, at 13 (PJI/JFA,
2012) [hereinafter PJI/JFA].
92
In some ways, the set is better than using Colorado’s statutory factors – none
of which are weighted and many of which did not end up on the tool at all.240
For example, “age at first arrest” is a statistically-derived risk predictor on
the CPAT, but it is not listed among the statutory factors judges are
encouraged (and were once mandated) to consider when setting bail.241
Similarly, judges in Colorado are statutorily encouraged (and were once
mandated) to consider prior failures to appear (which also appears on many
risk tools around America), but due to data limitations, that factor does not
appear on the CPAT.
242 In other ways, however, the set is inferior to certain
statutory variables, which allow judges to inquire into facts and
circumstances that may provide nuance to severity or type of risk.
Nevertheless, an important point to remember is that based on variations in
population, data collection methods, adequate data sources, and other
variables, actuarial pretrial risk assessment instruments differ among the
jurisdictions that use them. In 2009, VanNostrand and Keebler identified
nine statistically significant and policy relevant predictors of pretrial
outcomes in the federal system to guide decision makers in the release and
detention process in the federal courts.243 Two years later, Mamalian
examined studies undertaken in the previous decade and summarized the six
most common pretrial risk factors identified by those studies.244 Despite
their commonalities, however, she advised caution in extrapolating those
factors due to their nuanced differences.
That same year, Bechtel, Lowenkamp, and Holsinger completed a
comprehensive meta-analysis to examine numerous risk factors and to
identify which factors were statistically associated with pretrial failure.245
Most recently, the Pretrial Justice Institute listed 17 risk factors variously
linked to six widely-used assessment tools; interestingly, no single factor

240 See Colo. Rev. Stat. § 16-4-103 (2016).
241 Like most states, there is a catch-all in the Colorado statutes that allows judges to consider “any other
facts,” but before the CPAT was created, it is highly unlikely that judges ever considered age at first arrest.
See id., § 16-4-103 (h), (i), (j).
242 This is somewhat counterintuitive, and seems to be an aberration from many other tools, which do
include prior failures to appear as a risk factor. It is widely believed that the variable’s inability to predict is
due to Colorado’s data collection surrounding missed court dates.
243 VanNostrand & Keebler, supra note 236.
244 Cynthia A. Mamalian, State of the Science of Pretrial Risk Assessment (PJI/BJA, 2011), found at
https://university.pretrial.org/viewdocument/state-of-the-science.
245 Kristin Bechtel, Christopher Lowenkamp, & Alex Holsinger, Identifying the Predictors of Pretrial
Failure: A Meta-Analysis, 75 Fed. Prob. 78 (2011) [hereinafter Bechtel, et al.], found at
https://www.ncjrs.gov/App/Publications/abstract.aspx?ID=258843.
93
was included in all six tools.246 In sum, the differences among various
assessment tools (including the weighting of the predictive factors
themselves), nuances in meaning behind predictive factors, and even
operational or legal and cultural differences mean that certain elements that
are considered predictive in one jurisdiction may not be considered
predictive in another. Accordingly, no single list is produced as definitive,
and jurisdictions must recognize the necessity of continually validating any
proposed set of predictive factors to their local populations.
Nevertheless, many of the factors found in the various tools are similar, and
tend to fall into two groups: (1) static (unchanging factors typically
pertaining to criminal justice history or involvement, such as history of
FTAs or prior convictions); and (2) dynamic (changing factors typically
pertaining to community stability, such as employment or residence).
247 Of
these two groups, static factors are emerging as the strongest predictors of
pretrial misconduct,248 although some researchers have argued that dynamic
factors and the defendant interviews often needed to ascertain them likely
have independent value. Nevertheless, the combination of the acute need for
research-based pretrial assessment in America and budgetary considerations
means that an assessment instrument using only static factors that does not
take a defendant interview to complete is likely to become even more
popular in the future.
In 2013, researchers funded by the Laura and John Arnold Foundation
created such a tool, named the Public Safety Assessment (PSA). It uses nine
static factors (which are weighted and tested to be race-and-gender-neutral)
to accurately predict the risk that a defendant will commit any new crime,
commit a violent crime, or fail to appear for court.
249 The PSA was created
using an extremely large defendant population, making the tool initially
generalizable to all states. It is currently being tested in multiple American
jurisdictions, and a recent study in Kentucky reported that after six months

246 PJI Risk, supra note 232, at 3.
247 See Marie VanNostrand & Christopher T. Lowenkamp, Assessing Pretrial Risk Without a Defendant
Interview (LJAF, 2013) [hereinafter VanNostrand & Lowenkamp], found at
http://www.arnoldfoundation.org/wp-content/uploads/2014/02/LJAF_Report_no-interview_FNL.pdf.
248 See Bechtel, et al., supra note 245; VanNostrand & Lowenkamp, supra note 247, at 5.
249 See Laura and John Arnold Foundation, Developing a National Model for Pretrial Risk Assessment
(LJAF, 2013), found at http://www.arnoldfoundation.org/wp-content/uploads/2014/02/LJAF-researchsummary_PSA-Court_4_1.pdf.; VanNostrand & Lowenkamp, supra note 247, at 15-18; Laura and John
Arnold Foundation, Public Safety Assessment: Risk Factors and Formula (LJAF, 2016) [hereinafter LJAF],
found at http://www.arnoldfoundation.org/wp-content/uploads/PSA-Risk-Factors-and-Formula.pdf.
94
of using the tool, that state was able to release more defendants pretrial and
yet reduce pretrial crime by nearly 15%.
250
After testing the various theoretical predictors of risk, the researchers
helping a jurisdiction develop a risk assessment instrument typically create a
graph showing the varying levels of misconduct associated with rising
scores. Colorado’s misconduct graph looked like this:
251
Diagnostic Plots of Misconduct Rate by Points on the Risk Assessment Scale with Cumulative
Sample Proportions.
Note. FTA/Filing refers to either a FTA or a new filing.
Based on this data pattern, researchers in Colorado then decided where to
divide the data into groups to represent categories of risk. These categories
can be created in different ways, and Colorado ultimately used a so-called
“natural breaks” method, which examined the data for places along the
graph where data naturally cluster together or break apart. No matter which

250 Laura and John Arnold Foundation, Results From the First Six Months of the Public Safety Assessment –
Court in Kentucky (LJAF, 2014), found at http://www.arnoldfoundation.org/wpcontent/uploads/2014/02/PSA-Court-Kentucky-6-Month-Report.pdf.
251 PJI/JFA, supra note 239, at 14.
95
risk instrument one uses, the user will quickly notice that the data have been
reduced into certain categories, often corresponding to scoring on the tool
and telling persons the predicted success rates for the various categories. In
Colorado, the natural breaks led to the creation of four categories, which
were refined and are now represented in the graph below:
252
Revised Risk
Category Risk Score Public Safety
Rate
Court Appearance
Rate
Overall Combined
Success Rate
1 0 to 17 91% 95% 87%
2 18 – 37 80% 85% 71%
3 38 – 50 69% 77% 58%
4 51 – 82 58% 51% 33%
(Average) 30 78% 82% 68%
From this chart, one can see that, for example, a defendant scoring from 0-
17 places him or her in Category One, which represents the lowest risk or
the best chances for success with a predicted public safety rate of 91% and a
predicted court appearance rate of 95% (another graph, not included here,
illustrates that in Colorado, about 20% of defendants will fall into this
category). Likewise, Category Four defendants – who represent
approximately 8% of all defendants arrested and brought to jail – are
predicted to succeed at 51% and 58% levels for public safety and court
appearance, respectively. One complicating factor with the Colorado tool
(and similar tools) is that the “overall success rate” – that is, how many
defendants remain completely arrest free and return for all court hearings –
is lower than the categories separately. This rate, comprised of defendants
who succeed at both outcomes simultaneously, is somewhat smaller merely
because it is rarer for defendants to remain both crime and FTA free than to
remain only crime or only FTA free.
Virtually all risk instruments operate this way, with risk scores transferring
to categories based on cutoffs that are largely determined by researchers or
the jurisdictions using the instrument. Thus, in Colorado, when a defendant
scores as a category one, the summary document tells the court that this
particular defendant looks like other, similar defendants, who, when
released, have performed at these levels. We do not know whether this
particular defendant will perform the same and, in fact, we will never know
whether this particular defendant will perform the same until he or she is
released. The difficulty for any judicial official setting bail is to try to
determine if this defendant is like the vast majority who will succeed, or if

252 Id. at 18.
96
he or she will be among the nine or five percent of CPAT Category One
defendants who fail.
These instruments have significant utility in determining what conditions to
use to manage pretrial risk for released defendants. In short, knowing a
defendant’s risk category, along with other information gleaned from a
pretrial services interview combined with knowledge of the risk literature,
can allow a pretrial services agency officer to recommend some set of
research-based techniques designed to manage risk of released defendants.
For example, in Denver, Colorado, the pretrial services agency has learned
what has also been shown in national research: through supervision, a
jurisdiction can significantly improve overall success rates of even the
highest risk defendants when they are, in fact, released.
253
How Do the Risk Research and Actuarial Pretrial Risk
Assessment Instruments Illuminate Flaws in the Current
System?
Much of the knowledge we have gained from the research used to create
actuarial pretrial risk assessment instruments illuminates dramatic flaws in
our current system of pretrial release and detention. It does this primarily by
showing that criminal charge, while in many cases some part of defendant
risk, is only a small part of defendant risk. For example, in the revised
validated Virginia tool of 2009, “primary charge type” (i.e., whether the
charge is a felony or misdemeanor) was only one of eight factors necessary
to predict risk; a 2016 modification from primary charge type to “charge is
felony drug, theft, or fraud” indicates a more nuanced and superior measure,
but still remains only one of several factors.254 In the Florida tool, “current
most serious charge” is one risk factor of eleven, and, in fact, that tool
weighs a “current property charge” at four times a “violent charge.”
255 In
Colorado, the study – admittedly counterintuitively – found that the statistics

253 Denver Dept. of Pub. Safety, Denver Pretrial Servs. Prog. CY15 Ann, Rep. at 7 [hereinafter Denver
Annual Report], available from the author or the Denver agency. This is consistent with other, national
research showing that general supervision can increase court appearance and public safety rates for released
defendants showing moderate and high risk in significant numbers compared to defendants without such
supervision. See Harvard Law School Primer, supra note 3, at 16-17.
254 See Marie VanNostand & Kenneth Rose, Pretrial Risk Assessment in Virginia, at 13 (CDCJ/VCJA,
2009); Mona J.E. Danner, Marie VanNostrand, & Lisa Spruance, Race and Gender Neutral Pretrial Risk
Assessment, Release Conditions, and Supervision (Luminosity, 2016).
255 James Austin, Avi Bhati, Michael Jones, & Roger Ocker, Florida Risk Assessment Instrument, at 13
(JFA Inst., 2012) [hereinafter Austin, et al.,], found at https://university.pretrial.org/viewdocument/floridapretrial-ris.
97
“failed to show that the nature (e.g., person or property crime) or severity
(felony, misdemeanor) of the defendant’s current charge was statistically
significantly related to pretrial misconduct.”256
Most recently (and importantly, due to the large defendant population used
to test the tool), the Arnold Foundation’s PSA tool showed charge type and
severity were not predictive for failure to appear or new criminal activity,
but “current violent offense” was one of five factors used to create its socalled “violence flag.”257 The way that current charge type and severity is
used in these instruments lies in stark contrast to the way jurisdictions have
used them previously in pretrial release and detention. Previously,
jurisdictions looked almost exclusively at charge, assumed risk based on
charge to set some arbitrary financial condition – with amounts rising as
charges appear more and more serious – and then waited to see what
happened.
Indeed, these more nuanced examinations of defendant risk have turned
much of what we have believed about “risk based on charge” on its head. As
long as America has been a country, we have operated on a somewhat
intuitive assumption that “the higher the charge, the higher the risk.” This
notion is grafted onto our constitutions and statutes and is a primary part of
our current release and detention policies and practices such as through the
use of monetary bail bond schedules, which assign rising dollar amounts to
increasingly serious crimes. Salerno was decided using certain charge-based
assumptions, and in its opinion, the Supreme Court specifically noted that
the Bail Reform Act “operates only on individuals who have been arrested
for a specific category of extremely serious offenses. Congress specifically
found that these individuals are far more likely to be responsible for
dangerous acts in the community after arrest.”258 Compared to the research
available today, these findings are likely now somewhat simplistic.259
Instead, risk assessment research tells us what logic should suffice: we see

256 PJI/JFA, supra note 239, at 20.
257 LJAF, supra note 249, at 3.
258 United States v. Salerno, 481 U.S. 739, 750 (1987).
259 Congress’s findings on this matter were derived primarily from two studies. The first study apparently
only showed that a certain percentage of defendants committed crimes while on release, a finding necessary
to conclude that public safety should be a consideration at bail. In addition, Congress used a second study
of felony defendants in the District of Columbia and made certain assumptions about risk based on whether
any particular defendant was (higher risk) or was not (lower risk) on a surety bond. See S. Rep. No. 98-225,
at 6-7.
98
low and high risk defendants facing charges of all types260 and, indeed, some
of our riskiest individuals released pretrial (as measured by a risk tool) are
facing the least serious charges such as non-violent misdemeanors or
property offenses.
261
This point is crucially important to understand when it comes to re-drawing
the line between release and detention. As we will see in detail later, creating
a detention eligibility net by using an actuarial pretrial risk assessment
instrument from the current generation of instruments (or creating an
unlimited charge-based net while using the tool to sort defendants later) is
significantly flawed, which points to jurisdictions continuing to use criminal
charge to initially delineate whom to release and detain. But unless those
jurisdictions are able to somehow also justify that charge-based
determination – that is, unless they can show some research that a defendant
facing a particular charge (or one of a particular group of charges) is
somehow at an elevated risk to do the thing that society hopes to avoid –
then we likely have no justification to initially detain anyone pretrial. The
risk tools consistently tell us that, when it plays any part at all, current
charge is only one small part of defendant risk, and that we see persons
showing all levels of risk for all charges. Knowing this, jurisdictions must
tread lightly when crafting a detention eligibility net based on criminal
charge.

260 For example, in New York, the New York City Criminal Justice Agency has tracked local pretrial rearrest data for many years. In 2005, that agency reported that the pretrial arrest rate for a 2001 sample of
defendants was 17%, but that those re-arrests were split nearly evenly among persons charged with violent
felonies, non-violent felonies, and misdemeanors. Moreover, defendants were re-arrested for all types of
crimes (with certain nuanced variations), and for all different charge types and severity levels as compared
to the initial charge. Qudsia Siddiqi, Research Brief No. 8: Pretrial Re-Arrest Among New York City
Defendants, at 3-4 (NYCCJA, 2005).
261 Support for this notion is not always easy to find in the published material. Nevertheless, support can be
found by looking at how risk instruments occasionally weigh levels of crime when levels of crime are
actually deemed a predictor of risk; for example, and as noted previously, the Florida risk tool weighs
property crimes higher than both drug and violent crimes for risk. See Austin et al., supra note 255, at 13.
Support is sometimes found when jurisdictions keep comprehensive data about their pretrial program. For
example, in Mesa County, Colorado, local jurisdiction data collection revealed varying risk scores for all
separated charge categories, including “high risk, low charge” and “low risk, high charge” groupings. The
American Bar Association Standards on Pretrial Release note “some evidence that the risk of nonappearance or criminal behavior may actually be greater for persons charged with relatively minor nonviolent offenses (e.g., prostitution, retail theft, numbers-running, small-scale drug possession) than for
some persons charged with more serious crimes.” ABA Standards, supra note 100, Std. 10-5.1(b)
(commentary), at 104 (citing John S. Goldkamp, Two Classes of Accused: A Study of Bail and Detention in
American Justice (Cambridge, MA: Ballinger, 1979); John S. Goldkamp, Michael R. Gottfredson, Peter R.
Jones, & Doris Weiland, Personal Liberty and Community Safety (New York: Plenum Press, 1995)).
99
As we learn more about who, exactly, is “risky” pretrial, we are faced with
certain dilemmas surrounding our charge-based assumptions. For example,
jurisdictions are often comfortable with separating out sex offenders for
special punishments, and, indeed, before they are convicted, jurisdictions are
equally comfortable with giving persons charged with sex offenses higher
money bail conditions due to the serious nature of the charge.262 In any given
case, however, risk assessment can illustrate that a person accused of a sex
offense poses very little risk whatsoever. If so, what should we do about
persons charged with sex offenses? We are used to demanding high bond
amounts, an act that simultaneously assumes high risk and signals our
beliefs about the seriousness of the alleged crime. But if particular charges
only play one small part of defendant risk, and if an accused sex offender
poses little risk, are we willing to let that defendant out under minimal
supervision, as the risk research would suggest? The same issue is raised in
the case of a “high risk” person charged with a “lower level” crime. Do we
release the low risk defendant accused of murder but detain the high risk
homeless defendant accused of trespassing, and which of these situations are
perhaps more appropriately addressed outside of bail?
Risk assessment research also illuminates flaws in the traditional charge and
money-based system by showing that defendants are simply not all that risky
(relatively speaking) to begin with, and that failure is much less likely than
we probably assumed when we had no empirical data to back it up. Indeed,
if one looks at the research behind any particular pretrial risk assessment
instrument, he or she will see that lower risk defendants are incredibly
successful, operating in predicted risk categories with success rates in the
90th percentiles.263 Moreover, “high risk” defendants in most instruments are
often predicted to succeed more than half of the time,264 and can actually
succeed at higher rates than predicted when released with conditions

262 Indeed, Arizona added certain sexual offenses in its “categorical” no bail constitutional provision. See
Simpson v. Miller, 377 P. 3d 1003 (Ariz. Ct. App. 2016).
263 See, e.g., PJI/JFA, supra note 239, at 18.
264 See PJI Risk, supra note 232, at 4 (showing even Kentucky’s overall success rate for level 5 (higher risk)
defendants is 64%). A risk instrument could be created to include a “high risk” category in which
defendants failed at, say, 80% levels, but the number of defendants covered by that category would be
significantly lowered.
100
designed to manage risk.265 In short, during that small window we call the
pretrial phase of a criminal case, defendants are not as risky as we think.266
Additionally, when defendants fail, the failures are simply not as bad as the
failures we have historically articulated that we wish to avoid. For example,
when it comes to public safety, America has historically articulated that it
wants to avoid extremely dangerous persons committing serious or violent
crimes while on bail. Under the traditional charge and money method, our
assumptions regarding risk meant that risk and resulting failure were tied to
the charge; accordingly, for example, if we arrested a bank robber, we might
assume that he or she was risky to commit another bank robbery (or
something equally serious). This assumption thus justified the notion of
setting higher bond amounts for more serious crimes. Research surrounding
pretrial risk assessment, however, tells us that when people fail by
committing new crimes, they are not typically the kinds of crimes we fear.
For example, in Washington D.C., while 91% of released defendants remain
arrest free, 98% remain arrest free for a crime of violence while on pretrial
release.267 This gets at a more nuanced discussion concerning the question of
“risk of what,” which is discussed under the section of this paper titled,
“What Do Actuarial Pretrial Risk Assessment Instruments Not Tell Us?”
This is not to say that certain defendants – especially defendants charged
with serious or violent offenses – do not commit crime while on bail. Indeed,
as will be shown later, the research on violent crimes provides some
empirical justification for a charge-based detention eligibility net covering
violent offenses that simply does not exist for other categories. Overall,
however, defendants are not as risky as we think, and the ones who are
extremely risky are often hard to spot due to the rarity of the event.
The notion that the research tells us that defendants are less risky than we
think is clouded by the fact that “risk” is largely determined subjectively by
the researchers creating the instruments and the jurisdictions adopting them.
For example, and as noted above, when the CPAT was created in Colorado,

265 In 2013, Denver showed an overall success rate of 58% for its highest risk defendants, much higher than
the CPAT’s predicted overall success rate of 33%. See Denver Annual Report, supra note 253. Readers are
reminded that the “overall” success rate is typically lower than either of the individual success rates for
court appearance and public safety, which are 51% and 58%, respectively.
266 This can be due to many reasons, including, logically, the shorter periods of time defendants are
watched. The need only to assess risk during the pretrial period is one reason why jurisdictions should use
caution when looking at research that evaluates outcomes beyond the pretrial window.
267 See D.C. Pretrial Performance Measures, found at https://www.psa.gov/?q=data/performance_measures.
101
the researchers plotted a line indicating failures based on risk assessment
scores. That graph was then used to create cutoffs, initially by the
researchers simply dividing the data into quarters. Later, local researchers
looked for the “natural breaks” to create different cutoffs, which were
molded, as well, by local criminal justice leader input.
Together, the researchers and Colorado officials decided who belonged in a
category and what to call it. In Colorado they used numbers, labeling the
lower risk defendants as “in Category One” and higher risk defendants as “in
Category Four.” With equal confidence and propriety, however, Colorado
could have used only two categories, or six, or could have made it so a
Category Four included only 2% of all defendants, or could have named the
categories, “extremely low,” “low,” “medium low,” and “slightly above
low.” For these and other reasons, the categories and cutoffs differ across the
country, and represent fairly subjective notions concerning varying
jurisdictional tolerance (or likely intolerance) for risk. Most relevant to this
paper, the subjective aspects surrounding these instruments, by themselves,
makes them potentially inadequate for deciding whom to release and detain
pretrial in the first instance based solely on prediction.
Together, these two notions – the notion that defendants are simply not as
risky as we think (especially for the things we fear) coupled with the notion
that we define risk somewhat subjectively – become crucially important
when we consider perhaps the most deceptively dangerous thing that many
actuarial pretrial risk assessment instruments do: they subtly tell us that all
defendants are risky simply because they are all labeled as “risky.”
Historically, being ignorant of actual defendant risk allowed us to avoid
dealing with risk altogether, even though it is the primary consideration at
bail. Courts could make certain assumptions about the charge, assign an
amount of money reflecting either those assumptions or their sense of
seriousness surrounding the charge, and yet be largely unaware of detailed
information that might cause them to re-think release or detention in any
meaningful way.
Although this problem has existed to some degree before, in this generation
of bail reform courts are increasingly handed information in the form of
scientific, statistically-based risk instruments labeling every defendant as
“risky” and containing detailed information showing that even so-called
“low risk” defendants fail. Judges are then told that although the defendant
standing before them is likely to succeed, his or her individual risk cannot be
102
predicted. Given this information, it seems natural to assume that those
courts will likely err on the side of over-conditioning versus underconditioning, on detention versus release. In other words, having now been
given statistical data showing with mathematical precision that some
defendants in every risk category will undoubtedly fail, can we hope courts
will still follow American law by embracing the risk associated with release?
Without some significant increase in bail education, asking courts to release
more defendants pretrial (a goal in American bail since the country was
founded) while simultaneously showing them statistical evidence of failure
seems destined to lead only to the opposite outcome: more detention.
Risk assessment also illuminates flaws in the current system by allowing us
to see exactly who is in jail based on risk category, and for the most part this
helps to generally confirm our surmise that many of the wrong people are in
and out of jail. Whenever a person conducts a study either directly or
indirectly examining jail population by risk, we see that there are significant
numbers of low and medium risk defendants in jail, and that there are
occasionally higher risk defendants out of jail.268 This, of course, is a
monumental finding of the sort that has led to bail reform throughout the
history of England and America. Throughout history, whenever the wrong
people are in jail pretrial, bail reform occurs as a natural correction.269
Obviously, from any incarcerated defendant population, one will see
instances of a “low risk” person who, in fact, presents a higher risk of the
sort not necessarily measured by current statistical instruments. 270
Nevertheless, thinking about these things in the aggregate, we must view the
issue of who should be in and out of jail in the context of the issues
discussed above. Knowing that we define risk and create the cutoffs
subjectively, acknowledging that almost all defendants are risky but are
actually far less risky than we have assumed, and knowing that they are not
so risky for the things we actually fear or seek to address through detention,
should lead us to conclude that there are, in fact, far more persons in jail

268 See, e.g., Michael R. Jones, Unsecured Bonds: The As Effective and Most Efficient Pretrial Release
Option, at 12 (PJI, 2013) (showing both “low” risk persons held in jail and “high” risk persons released
from jail).
269 See generally NIC Fundamentals, supra note 6.
270 Indeed, in any particular jail population, there will undoubtedly be instances where defendants assessed
“low” risk by a risk assessment tool will nonetheless pose an identifiable and unmanageable risk to do
something bad if released, just as there will be instances where defendants assessed as “high” risk by a risk
assessment tool will be manageable if released. Assessing risk for the jail population nonetheless gives
jurisdictions at least a broad idea about who is in their jails.
103
pretrial who likely should be released, and far fewer persons out of jail who
likely should be detained.
What Do Actuarial Pretrial Risk
Assessment Instruments Not Tell Us?
The answer to this question is perhaps the most significant answer when
deciding how to incorporate empirical defendant risk into re-drawing the
line between release and detention. As noted previously, many jurisdictions
have begun making wholesale changes to their release and detention
practices by replacing their mostly charge-based system with a mostly riskbased one. To do this, they are using actuarial pretrial risk assessment
instruments to guide them so as to – putting it somewhat simplistically –
detain higher risk defendants and release lower risk ones. This articulation of
purpose is attractive, but it sets up a system that is in need of further analysis
and, ultimately, rational justification.
Preliminarily, some of what actuarial pretrial risk assessment instruments do
not tell us is tied to what they do tell us. For example, because risk
assessment instruments tell us primarily who is likely to succeed only in a
particular jurisdiction, they do not necessarily tell us who is likely to succeed
in all jurisdictions.271 Likewise, because properly created risk assessment
instruments tell us a prediction of a narrow band of misconduct for only the
pretrial period, they do not tell us risk in the long term, and thus they should
not be used, for example, for program placement, pleas, or to otherwise aid
in the sentencing decision. Moreover, in many cases risk instruments do not
necessarily tell us how defendants will become more or less risky with
conditions or some other treatment; for that, we must rely on other research
or experience.
272 For purposes of this paper, however, actuarial pretrial risk
assessment instruments do not tell us three important things: (1) individual
risk; (2) detail concerning “risk of what;” and (3) protective factors that
offset risk and what to do with assessed risk.

271 The Arnold Foundation’s Public Safety Assessment is an attempt to create a single risk instrument
capable of being used across all American jurisdictions. It has the potential to do this through its creation
from an extremely large data set. Nevertheless, the tool is still being tested and validated to specific
jurisdictions’ populations.
272 General risk research tends to illustrate the “risk principle,” which suggests using more interventions or
supervision for higher risk persons and less interventions or supervision for lower risk persons. See
Milgram, et al., supra note 231, at 216. As noted previously, in some American jurisdictions, experience
has led people to believe that pretrial supervision (versus detention) for “high” risk persons can result in
better outcomes than predicted by any particular assessment tool.
104
Individual Risk
First, actuarial pretrial risk assessment instruments do not tell us individual
risk. Instead, they predict individual risk based on how a group of similar
defendants performed under like circumstances. The general inability to
assess individual risk has been described by LaFave, et al., as presenting, at
least arguably, a “fundamental constitutional defect” under any legal theory
because to reliably detain any individual who would, in fact, miss court or
commit a new crime while on pretrial release, a judge would have to also
detain those who ultimately would not fail.273 As noted previously, however,
risk assessment at bail has been done since at least 400 A.D., so risk
prediction of this sort is not the sort of government action that would
necessarily shock the conscience and thus lead automatically to a finding of
unconstitutionality. Moreover, and as also noted by LaFave, it is a defect
that will likely remain tolerated to some degree as the Supreme Court itself
has written that “there is nothing inherently unattainable about a prediction
of future criminal conduct.”274 Of course, the Court wrote this sentence
before we ever had empirical evidence showing just how many incorrect
predictions we might actually have. Accordingly, jurisdictions tempted to
move toward incorporating laws or policies designed to detain all “high risk”
defendants should do so with caution simply because most actuarial pretrial
risk assessment instruments tell us that, more often than not, a “high risk”
person will typically succeed if released while additional risk research has
shown that these persons will succeed at even higher rates with certain
interventions such as pretrial supervision. True individual risk (of the sort
we desire to know prior to using pretrial detention) is thus something that
must be ascertained from something beyond current actuarial tools operating
with existing cutoffs.
LaFave’s concern once again raises the issue of false positives at bail – an
incorrect prediction that someone is either dangerous or a flight risk – when
the decision to detain such persons is unfalsifiable.
275 If an actuarial pretrial

273 LaFave et al., supra note 52, §12.3(f), at 81-82.
274 United States v. Salerno, 481 U.S. 739, 751 (1987) (quoting Schall v. Martin, 467 U.S. 253, 278
(1984)).
275 Generally speaking, if predicting risk of violence, for example, a true positive would be a person
predicted as violent who subsequently commits a violent offense. A true negative would be a person
predicted to be nonviolent who does not subsequently commit a violent offense. A false positive would be a
person predicted as violent who proves to be nonviolent, and a false negative would be a person predicted
to be nonviolent who subsequently commits a violent offense. While false negatives are also extremely
105
risk assessment instrument tells us that a defendant looks like a group of
similar defendants labeled as “high risk” for public safety, but the same tool
also tells us that “high risk” defendants who are released will succeed more
often than they fail, detaining all “high risk” persons just to make sure we
capture all crimes for this group of persons will inevitably lead to
significantly high numbers of false positives. This problem is exacerbated by
the fact that risk research tells us that only an extremely small number of
“high risk” defendants commit serious or violent crimes when released. If a
jurisdiction detains 100 defendants just to make sure it reaches the one
defendant who will commit a violent crime, then despite what the Supreme
Court says about prediction, that jurisdiction will undoubtedly have false
positives in unconstitutionally high numbers.
Caleb Foote called the people we allow to be in the category of false
positives, “a dehumanized second-class category of persons” who are, in
fact, “expendable.”276Authors Jeffrey Fagan and Martin Guggenheim
similarly write that it is helpful to view false positives as “individuals
deprived of their liberty for utilitarian purposes” – that is, persons “jailed not
to stop them from any wrongdoing but in order to throw a wide enough net
to cover others, who, if not stopped, would endanger society.”277
Nevertheless, and as those authors also suggest, while decisions diminishing
the rights of convicted persons for the collective good might in some
instances be acceptable, at bail they are decidedly less so. Accordingly,
when re-drawing the line between release and detention, jurisdictions should
be informed of our inability to predict individual risk as well as our ability to
recognize that individual rights likely outweigh claims of utility.
Moreover, if we are truly concerned about not detaining persons who would,
in fact, succeed if we released them, then we must also deal with base rate
problems. A base rate is simply the rate at which a thing that we are trying to
predict happens naturally in the population of interest.278 As Stephen
Gottfredson explains, the difficulty of predicting becomes more problematic

important in bail, they are not unfalsifiable because the defendants are released. See Dean J. Champion,
Measuring Offender Risk: A Criminal Justice Sourcebook, at 73 (Greenwood Press, 1994).
276 Caleb Foote, Comments on Preventive Detention, 23 J. of Legal Ed. 48, 52 (1970).
277 Fagan & Guggenheim, supra note 3, at 428. In this document, Fagan & Guggenheim describe the results
of a “natural” study caused when the courts required judges to release juveniles the judges had already
determined to be dangerous, thus making the decision falsifiable and thus subject to analysis.
278 See Stephen D. Gottfredson, Prediction: An Overview of Selected Methodological Issues, at 25, in 9
Prediction and Classification: Crim. Just. Decision Making (U. of Chicago Press, 1987) [hereinafter
Gottfredson].
106
whenever the base rate either increases or decreases from 50%.279 When
base rates are high, “the difficulty involves developing bases to make
predictions that improve on randomness.”280 But when base rates are low,
prediction is only good if it improves upon the base rate.
In bail, for example, we are trying to predict flight and violent or serious
crime during pretrial release. Unfortunately, however, these things are
actually very rare and so the base rates are extremely low. Noted legal
philosopher Andrew von Hirsch explains what makes criminal conduct
generally resistant to prediction:
(1) It is comparatively rare. The more dangerous the conduct is,
the rarer it is. Violent crime – perhaps the most dangerous of all
– is the rarest of all. (2) It has no known, clearly identifiable
symptoms. Prediction therefore becomes a matter of developing
statistical correlations between observed characteristics of
offenders and criminal conduct.281
And when it comes to statistical correlations, unless we can predict a
relatively rare event the same or better than its actual rate, we will have
problems with false positives. For example, if we are concerned with
reducing violent pretrial crime, but only 1% of defendants are known to
commit violent pretrial crime, then our prediction method leading to
detention must do better than simply letting all defendants out of jail, for
letting all defendants out of jail will yield results that are right 99% of the
time.
There are fundamental issues with how America is beginning to almost
reflexively adopt actuarial pretrial risk assessment instruments as a panacea
to bail problems. Perhaps the most important issue is that by adopting these
instruments, we have adopted their definitions, and thus we call all
defendants “risky.” And yet, when defining that risk, we have moved away
from worrying about flight to worrying about all FTAs, and from worrying
about serious or violent crime while on release to worrying about any and all

279 Id.
280 Fagan & Guggenheim, supra note 3, at 426.
281 Andrew von Hirsch, Prediction of Criminal Conduct and Preventive Confinement of Convicted Persons,
21 Buff. L. Rev. 717, 733 (1971-71) [hereinafter von Hirsch]. Von Hirsch goes on to explain that what
makes violence so particularly difficult to predict is both rarity and situational quality; that is, violence does
not apparently adhere to certain individuals, but, instead, can happen to any person based on a number of
variables beyond the characteristics of the defendant.
107
crime. This, in turn, has altered our base rates. This is discussed in greater
detail later in this paper, but jurisdictions should be mindful of how this
issue is manifesting in this generation of bail reform by reading through the
following brief explanation.
As noted previously, ever since America began intentionally detaining
noncapital defendants in the 1960s, we articulated a common desire only to
detain defendants who presented an unmanageable risk of willful flight to
avoid prosecution or a risk of serious or violent criminal activity while on
release. Those two things were incredibly rare, however, and so it was
difficult for any generation of statistical risk assessment to predict them.
And because they were hard to predict, any method for dealing with them
was likely to lead to a staggering number of false positives. For these and
other, mostly political reasons, we thus began (perhaps unwittingly) to
change the definitions of the things we wanted to do at bail; instead of flight,
we began articulating a desire to avoid all failures to appear, no matter how
benign, and instead of serious and violent crimes, we began articulating a
desire to avoid all criminal activity, no matter how minor. Doing so actually
allowed us to reduce false positives because it is easier to predict things that
happen far more frequently. While avoiding all crimes and failures to appear
is an appropriate goal of pretrial release, doing so becomes problematic
when we allow the risk of those things to lead to pretrial detention.
For example, and as noted previously, if we say we care about a defendant
committing a violent crime while on bail, but if only one of 100 defendants
will commit a violent crime while on bail, releasing all defendants will be
99% correct, and any prediction method will have to be at least that correct
(or better) to eliminate false positives. Increasing the base rate, however, can
help with prediction. Accordingly, if we develop an assessment that can
show a particular group is, say, 50% likely versus 1% likely to commit a
violent crime, we have increased the base rate, making prediction somewhat
less error prone. We would still have false positives, but not nearly so many
as if the base rate remained so low.
Unfortunately, in America we have created assessments with base rates
hovering around 50% for subgroups not necessarily by better predicting the
violent crimes, but instead by including more and more minor crimes to our
measurement of public safety. By adding those minor crimes, “it becomes
increasingly difficult to demonstrate a need for societal protection of the
degree of urgency that could conceivably warrant the kind of pretrial
108
deprivation of liberty [we would see].”282 Adding more and more crimes to
our definitions of public safety may have been unavoidable, because “it is
only when we allow a wide, standardless definition of pretrial danger that
the efficacy of the predictions even makes sense.”283 But do we really want
to be a country that uses secured detention to respond to a risk of committing
minor crimes, such as drug use or low level property offenses, simply to get
at the one or two persons who are extremely high risk to commit a serious or
violent crime? As noted by Fagan & Guggenheim, when we add petty and
minor offenses into our decision standard for dangerousness, we “run[] the
risk of predicting everything and nothing at the same time.”284
In sum, “data based on infrequently occurring behavior has low predictive
utility.”285 Three ways to deal with this problem include: (1) continually
narrowing the focus of the risk instruments (or the concept of risk generally)
to screen out higher and higher numbers of false positives by better
predicting the low number of true positives for the thing we seek to avoid;
(2) using cutoffs to identify a subgroup that has a much higher incidence of
the thing we wish to avoid and try to predict from that group; or (3) using
cutoffs for subgroups but also re-defining the thing we are attempting to
predict more broadly so that it includes defendants who have higher base
rates of pretrial misconduct of around 50%.With the research in America, as
noted above, we have tended to do option number three. Most risk
instruments today include subgroups of “risky” defendants, often with a
cutoff for “high risk” defendants with base rates for pretrial misconduct
hovering around 50 to 60%, which seems rational and which research
suggests is an appropriate rate to avoid false positives.
286 But in our ongoing
attempt to predict something that is hard to predict, we have (again, likely
unwittingly) re-defined the risk that we seek to avoid quite broadly to
include all failures, and simultaneously moved consideration of that
definition to the detention decision. American notions of freedom and
liberty, however, would suggest that we instead define those things quite
narrowly. Accordingly, until the science concerning risk begins to do
options number one and two, above, we should view what we have done
through option number three with extreme caution and be ready to override

282 Id. at 738.
283 Fagan & Guggenheim, supra note 3, at 447.
284 Id. at 448.
285 Hazel Kemshall, Understanding Risk in Criminal Justice, at 66 (Open University Press, 2003).
286 See von Hirsch, supra note 281, at 737; Fagan & Guggenheim, supra note 3, at 426 (and sources cited
therein).
109
our decision making frameworks based on actuarial pretrial risk assessment
instruments to effectuate higher rates of release.
Consider the adoption of the CPAT as a more specific example. In Colorado,
policy makers and researchers created cutoffs on that tool from one through
four, with “Category Four” representing defendants most likely to fail
relative to other defendants.287 Approximately eight percent of all defendants
assessed were predicted to end up in Category Four, and research showed
that defendants in that category tended to fail for new criminal activity 42%
of the time (succeed 58% of the time), and fail to appear for court 49% of
the time (succeed 51% of the time). Because these base rates are near 50%,
predictions of this subgroup are more likely to be free of false positives than
groups with lower base rates.
288 But these base rates for the subgroup are
only high because Colorado defined “public safety” as “a filing for any new
felony, misdemeanor, traffic, municipal, and petty offense, and was not
limited to a more narrowly defined set of crimes that involve a form of
physical or emotional harm to one or more victims.”
289 Colorado thus likely
improved upon the base rate for predicting crime while on bail, but it has
done so only by re-defining public safety to include far more minor crimes
versus only serious or violent crime while on bail.
Similarly, in Colorado, risk of failure to appear was defined broadly on the
CPAT, which labels a defendant as a failure for missing a single court date
out of possibly 10 or more court dates. The base rate for a defendant
willfully failing to appear for court to avoid prosecution, however, is
undoubtedly quite low. In creating the CPAT, Colorado has thus likely
improved upon the base rate for “flight,” but it has only done so by
attempting to redefine flight to mean “failure to appear.”
It should be noted that in these examples, we have been looking only at
subgroups for “high risk” defendants with higher base rates for new criminal
activity that approximate 50%. Despite jurisdictions broadly re-defining
public safety and flight to improve upon the rates, other subcategories of
defendants still have much lower base rates even for those broadly-defined
categories of failure (e.g., defendants in a “low risk” category might only fail

287 See generally PJI/JFA, supra note 239.
288 See von Hirsch, supra note 281, at 737. As Gottfredson explains, the difficulty of predicting becomes
more problematic whenever the base rate either increases or decreases from 50%. See Gottfredson, supra
note 278, at 25. When base rates are high, “the difficulty involves developing bases to make predictions
that improve on randomness.”
289 PJI/JFA, supra note 239, at 18, n. 23 (emphasis in original).
110
5% of the time), meaning that detaining defendants found within these
categories will likely lead to a high and possibly unacceptable number of
false positives.290 All of these things suggest that if we do not use caution,
jurisdictions will likely over-detain “high risk” defendants due to our
extremely broad definitions of public safety and flight, and over-detain
everyone else due to the definitions as well as extremely high base rates and
potential false positives.
Detail Concerning “Risk of What?”
Second, as mentioned above and intertwined in any discussion of base rates
and false positives, actuarial pretrial risk assessment instruments do not tell
us the important question concerning the nature and severity of the risk. In
short, they do not adequately answer the fundamental question of, “risk of
what?”291 For example, when we are told that a defendant is a Category Four
(“higher risk”) on the CPAT, that designation not only does not tell us
whether that particular defendant will succeed or fail (indeed, the “high risk”
group itself succeeds over 50% of the time), it also does not tell us what that
defendant who fails is likely to do to cause that failure. In Colorado, it could
mean risk of a new filing for anything from a petty or traffic offense all the
way to homicide. For flight, it could mean missing a bus on the way to court
all the way to moving to Venezuela. Making matters more complex, this
same question would likely be answered differently in other jurisdictions
using other tools.
The issue is not unknown to the pretrial field. In 2007, Dr. Marie
VanNostrand noted that, “Although pretrial risk assessment instruments in
most instances do well in predicting the likelihood of danger to the
community (often measured by new arrest pending trial) there is no known
research that explores the nature and severity of the new arrest.”292 Although

290 See PJI/JFA, supra note 239, at 18. For example, a Category One defendant on the CPAT has predicted
base failure rate of 9% for new criminal activity and 5% for failure to appear for court (with base rates
based on the much broader definitions of the two outcomes). For a judicial in-or-out decision based on a
predictive instrument to provide value, that decision would have to do better than the base rate for pretrial
crime and FTA because releasing all defendants in that category would likely yield 91% and 95% success
rates. To do better than the base rate and to avoid false positives (if, indeed, it could to any acceptable
numbers), though, an instrument would have to settle by missing a number of true positives. See von
Hirsch, supra note 281, at 733-35.
291 See Harvard Law School Primer, supra note 3, at 22 (stating that “[a] simple designation of ‘high risk’
may not tell a decision-maker whether that reflects risk of arrest for a serious violent crime”); Summers &
Willis, supra note 233, at 4 (“research is needed on the severity or type of risk identified by PRAIs”).
292 Marie VanNostrand, Legal and Evidence-Based Practices: Application of Legal Principles, Laws, and
Research to the Field of Pretrial Services, at 17-18 (NIC/CJI, 2007).
111
we are likely getting closer to exploring that nature and severity – the Arnold
Foundation’s PSA Tool includes a so-called violence flag, which “flags
defendants presenting an elevated risk of committing a violent crime”293
–
we are still far from the kind of research that would settle nagging doubts
about using risk assessment for certain functions, like using them as the sole
basis to detain.
This overall concept is so important that it requires further and separate
emphasis from previous discussions surrounding the topic. As noted earlier
in this paper, our previous bail schemes often operated on an assumption that
a person arrested for a particular crime was either unmanageably risky for
flight or to commit the same or similar crime if released, an assumption that
led mostly to detention eligibility nets based on certain serious criminal
offenses. But the risk of an armed robber committing another armed robbery
is far different from the risk of that robber trespassing. It would help to know
the distinction. And yet, actuarial pretrial risk assessment instruments are
created, and success or failure is ultimately measured, by defining the “risk
of what” differently than the risk America historically has sought to address.
For example, throughout the history of America we have been concerned
with flight – the kind of willful flight to avoid prosecution that would hinder
our ability to bring a defendant to justice in a legal process that relies on the
moral deterrence of written laws and requires freedom before conviction.
When America gradually began to allow intentional detention of noncapital
defendants based on flight, it was only allowed in “the rare case of extreme
and unusual circumstances,”294
a concept that followed into the Bail Reform
Act of 1984. In most jurisdictions, however, risk and failure are measured by
a defendant missing any single court date out of any possible number of
court dates. Indeed, this proxy measure is the national standard for
measuring this particular outcome.295 Essentially, we are assessing and
measuring risk of merely failing to appear for court, which, in virtually all
cases, is quite far from flight. While judges certainly have legitimate
concerns over making sure that defendants also do not forget their court

293 Public Safety Assessment: Risk Factors and Formula, at 1 (Arnold Found. 2016), found at
http://www.arnoldfoundation.org/wp-content/uploads/PSA-Risk-Factors-and-Formula.pdf.
294 United States v. Abrahams, 575 F.2d 3, at 8 (1st Cir. 1978).
295 Measuring What Matters: Outcome and Performance Measures for the Pretrial Services Field, at v, 3
(NIC 2011) [hereinafter NIC Measuring].
112
dates, or miss their buses to get to the court,296 these concerns likely do not
rise to a level justifying pretrial detention.
Likewise, in the latter half of the twentieth century, America became
concerned with defendants committing new crimes while on bail, and the
examples and statistics used to justify massive changes to our detention
schemes – starting with the 1970 D.C. Act and continuing through state
constitutional changes and Salerno – reflected our desire to enact measures
to address extreme risk of defendants committing serious and violent crimes
during the pretrial phase of the criminal case. Indeed, when America began
to allow intentional detention of capital and noncapital defendants based on
risk to public safety, it articulated a desire to “reduce violent crime” during
the pretrial period,297 committed by “the most dangerous of …
defendants.”
298 While the Bail Reform Act of 1984 broadened the purpose of
detention to include reducing nonphysical harms in addition to physical
violence,299 it was nonetheless still directed toward a “small but identifiable
group of particularly dangerous defendants”300 who pose “an especially
grave risk to the safety of the community.”
301 In most jurisdictions, however,
risk and failure concerning public safety are indicated by a defendant being
charged with any criminal offense, a definition of public safety vastly
broader than American history suggests. The national standard for
measuring this outcome urges jurisdictions to count a defendant as having
failed if he or she is charged with any offense that “includes a prosecutorial
decision to charge” and “carries the possibility of incarceration or
community supervision upon conviction.”302
Essentially, we are often assessing and measuring risk of the potential for
committing nearly all criminal offenses, which, in virtually all cases, is quite
far from the sort of public safety risk we have historically sought to address.
In 1970, Congress supplied ten examples to justify detention based on
danger, and nine of the ten involved persons charged with violent felonies,

296 There is now a great deal of literature showing that court date reminder programs dramatically increase
court appearance rates among defendants. This same literature also suggests that a significant portion of
failures to appear are due not to willful flight, but to other factors, such as forgetfulness.
297 H. Rep. 91-907, at 82 (1970).
298 Id. at 83.
299 See S. Rep. 98-225, at 12-13 (examples included corrupting a union or the risk that a defendant would
engage in drug trafficking).
300 Id. at 6.
301 Id. at 5. The Court in Salerno also mentioned the necessity of factors designed to gauge the “nature and
seriousness of the danger posed by the suspect’s release.” 481 U.S. 739, at 743.
302 NIC Measuring, supra note 295, at 3.
113
and all ten involved persons who subsequently committed violent felonies
while on pretrial release. Today in Colorado, the CPAT considers a
defendant to have “failed” if he or she violates a traffic offense and misses a
single court date for any reason, and does not distinguish between risk of
failure to appear and risk to public safety (something the newer instruments
are doing).303 While not routinely found in the published literature, email
correspondence with developers and users of other tools shows similar
issues. For example, the Florida risk assessment tool considers a defendant
to have “failed” if he or she violates a municipal ordinance leading to a
summons or citation to appear.304 Moreover, in an email to the author of this
paper, an official in one state pretrial services department said, “We count
everything. Arrests and citations from speeding to capital murder.”305
Occasionally, and on their own, jurisdictions will make individual
determinations that arrests for certain crimes should not be included as
failure despite its definition within the tool, but this only adds to the
somewhat random nature of risk instrument use between locales. Again,
while jurisdictions may have legitimate concerns over making sure that
people do not miss a single court date for any reason, and that they refrain
from all criminal activity while on pretrial release, these concerns, too, likely
does not rise to a level justifying pretrial detention.
This primary defect in using actuarial pretrial risk assessment as the sole
basis for detention – the fact that we are measuring something different from
the threat we seek to address – is likely more fundamental, constitutionally
speaking, than the defect of allowing risk prediction generally because it can
lead to over-detention based on circumstances (failure through a nonviolent
infraction, for example) that do not necessarily constitute a legitimate state
interest for detention to begin with. In short, while actuarial pretrial risk
assessment instruments may be the best current method of reliably assessing

303 See PJI/JFA, supra note 239, at 19. Likewise, while the Virginia Pretrial Risk Assessment Instrument
measures an FTA as a failure only if it results in a capias (coming slightly closer to a finding of willfulness
for court appearance even if warrants are overused), it considers an arrest for any new jailable crime to be a
failure. Because of the limitation to jailable offenses, it includes felonies and Class 1 and 2 misdemeanors,
but not traffic, local ordinance violations, or class 3 or 4 misdemeanors, which are “fine only” offenses. See
Marie VanNostand & Kenneth Rose, Pretrial Risk Assessment in Virginia, at13 (CDCJ/VCJA, 2009);
Mona J.E. Danner, Marie VanNostrand, & Lisa Spruance, Race and Gender Neutral Pretrial Risk
Assessment, Release Conditions, and Supervision (Luminosity, 2016); Va. Code Ann. §§ 18.2-10, 18.2-11.
Examples of Class 3 and 4 misdemeanors include destruction of property, public intoxication, regulatory
and license enforcement issues.
304 Email correspondence from bail researcher to Timothy R. Schnacke, Sept. 1, 2016 (name withheld for
privacy).
305 Email correspondence from bail practitioner to Timothy R. Schnacke, Sept. 1, 2016 (name withheld for
privacy).
114
probabilities of individual risk, at their core they are likely only measuring a
portion of the risk necessary to trigger pretrial detention.
Thus, it is helpful to think of pretrial risk leading to detention as having two
components: (1) the risk of what we fear or seek to address, which includes
the “extreme or unusual” risk of flight and serious or violent crime while on
pretrial release; and (2) the risk that virtually all actuarial pretrial risk
assessment instruments measure, which is typically the risk of FTA and
public safety as measured by any new crime while on pretrial release. The
risk measured by the assessment instruments can be used for 99% of
everything at bail, including some small part of detention, but it cannot be
used solely to detain. As articulated in the Harvard Law School Primer on
Bail Reform, risk assessment as measured by a tool is perhaps a necessary
but not sufficient basis to trigger a hearing on detention, but only if those
tools are “geared specifically to the risk of re-arrest for violent or serious
crime, as opposed to instruments that lump together re-arrest for serious and
non-serious crime or do not distinguish between re-arrest and nonappearance.”306 In fact, because of the many things that risk tools do not tell
us, it is likely only appropriate to use them as one factor in the detention
decision after some other triggering event.
Protective Factors That Offset Risk and What to Do With Risk
Actuarial pretrial risk assessment instruments also do not tell us various
protective factors that offset assessed risk and what to do with assessed risk
once jurisdictions have measured it. As mentioned previously, simply
labeling every defendant as “risky” might subtly lead jurisdictions toward
over-detention and over-supervision, but this is likely still preferable to a
system of release and detention based on money. Nevertheless, the issue of
overestimating risk is likely due, at least in part, to the fact that pretrial risk
assessment instruments focus on risk factors rather than so-called “protective
factors,” which are “variables that can be shown to decrease the likelihood
of failure,” and which can help to better determine individual versus
aggregate risk.307 Additionally, once risk is measured, the instruments do not

306 Harvard Primer, supra note 3, at 27.
307 Summers & Willis, supra note 233, at 4-5; see also John Jay College Prisoner Re-Entry Institute,
Pretrial Practice: Building a National Research Agenda for the Front End of the Criminal Justice System
(Oct. 26-27, 2015) [hereinafter John Jay], at 29 (statement of the Vera Institute of Justice describing the
need for some assessment of strengths instead of just risks). This document provides an invaluable
overview of pretrial research, including what is currently available and what is still needed as of the date of
publication. In Maine, researchers created “one of the very few” pretrial assessments to include protective
115
tell jurisdictions what to do next. Thus, it is helpful for jurisdictions to
remember that knowing defendant risk is simply not enough; other social
science research must be used to tell us “what works” to achieve our lawful
pretrial goals, and the law must provide our overall boundaries for using risk
assessment, including whether it should ever be used to draw the line
between release and detention.308
Accordingly, when re-drawing the line between pretrial release and
detention, jurisdictions must remember that actuarial pretrial risk assessment
instruments do not tell us individual risk, adequate detail concerning “risk of
what,” and things that offset risk along with what to do with assessed risk.
Understanding this can be a crucial part of justifying either a detention
eligibility net or a further limiting process as well as in crafting rules or laws
designed to effectuate the in-or-out decision. These limitations are a
hindrance only – and I must emphasize only – when people wish to use
actuarial pretrial risk assessment instruments as the sole basis for pretrial
detention.
How Does Risk Research Interact With the Law When It
Comes to Re-Drawing the Line Between Pretrial
Release and Detention?
Actuarial pretrial risk assessment instruments, and specifically what they do
and do not tell us, illuminate important new interactions with fundamental
legal theories. Each new interaction would likely fill its own volume, but we
will briefly consider a few here, including how risk assessment interacts
with due process, excessive bail, and equal protection.
Initial Balancing Issues
Initially, each of these legal theories requires some sort of balancing test,
which, in turn, requires courts to assess the government’s means it has
employed to meet a lawful government objective. Assuming that a
defendant’s liberty interest is fundamental, requiring strict or at least some

as well as risk factors. See Two Rivers Reg. Jail/USM Muskie School of Pub. Serv./Vol. of America, M
Risk: Pre-trial Risk Assessment, Maine Demonstration Project, 2 (BJA, 2011).
308 The research and the law are intertwined in this aspect. If the research shows that a particular condition
of release does not work to achieve our lawful goals, it would be irrational or unreasonable to set it, and
thus courts would likely conclude its imposition itself to be unlawful.
116
“heightened” scrutiny,309 and assuming that crime control and court
appearance are compelling government interests, a jurisdiction would have
to show that the way it administers pretrial detention for some class of
defendants is necessary to protect its compelling interest of reducing
defendant crime or flight while on pretrial release. The argument that
actuarial pretrial risk assessment cannot meet this test because the tools
cannot predict individual risk largely has failed (as noted previously, the
Supreme Court has there is said “there is nothing inherently unattainable
about a prediction of future criminal conduct.”).310 Nevertheless, today’s risk
research would likely require the government to provide more detail than
historically provided in its articulation of a compelling interest sufficient to
trigger potential pretrial detention.
Specifically, the government would likely need to provide some research
leading to findings that a certain type of indicator or combination of
indicators (such as charge) is likely to lead to higher risk for pretrial failure –
a finding made difficult by the research itself. Moreover, if the government
wished to use actuarial pretrial risk assessment to determine detention
eligibility, it would also likely have to articulate a compelling interest that
not only overrides the risk of over-detaining, but also an interest in
protecting all of society from pretrial crime that includes things like traffic
offenses, for that is included in what those tools measure. It would require
the government to articulate the need to protect the administration of justice
not only from willful flight to avoid prosecution, but also from a single FTA
for a court hearing that may or may not even be necessary.
The notion that the government needs to take greater care in articulating its
compelling interest is beginning to show up in court opinions. In LopezValenzuela v. Arpaio, the Ninth Circuit Court of Appeals struck an Arizona
detention provision due, in part, to the government’s inability to articulate “a
particularly acute problem,” quoting one of the elements mentioned by the
United States Supreme Court in Salerno.311 The Ninth Circuit noted:
The record in Salerno contained empirical evidence
establishing that the legislation addressed ‘a pressing societal
problem,’ and the law operated only on individuals ‘Congress

309 See United States v. Salerno, 481 U.S. 739 (1987); Lopez-Valenzuela v. Arpaio, 770 F.3d 772 (2014).
310 United States v. Salerno, 481 U.S. 739, 751 (1987) (quoting Schall v. Martin, 467 U.S. 253, 278
(1984)).
311 770 F.3d 772, at 782-84.
117
specifically found . . . are far more likely to be responsible for
dangerous acts in the community after arrest.’ This evidence
figured prominently in the Court’s decision to uphold the Bail
Reform Act.312
While there may still be persons “far more likely to be responsible for
dangerous acts in the community,” today’s actuarial pretrial risk assessment
instruments: (1) do not tell us precisely who they are; (2) illustrate that there
are fewer of them than we ever believed; and (3) in fact, show that even the
“highest risk” defendants – a label jurisdictions have largely made up – often
succeed more than fail, and can include persons posing only a high risk to
commit some infraction while on release. The same is true for flight. In
short, the empirical evidence points to less of a pretrial crime problem than
we likely ever thought existed before. This, in turn, makes it more difficult
for the government to justify detention.
Excessive Bail Generally
Beyond the above balancing issues faced in any of the three legal challenges,
an excessive bail challenge in most states requires the court to determine
whether a condition is reasonable – excessive bail is often defined as
“unreasonable” bail and non-excessive bail is defined as “reasonable” bail313
– and the general test is whether a court needs a particular condition (or
detention) to provide “reasonable assurance” of public safety or court
appearance.
314 And thus the same risk research and attributes of actuarial
risk assessment that make it more difficult to justify a particular balance –
what it tells us, what it does not tell us, and the fact that the risk instruments
measure something different than what we seek to address through detention
– means that a court would likely be on solid footing under the Excessive
Bail Clause by releasing all defendants pretrial. For example, if a court knew
that it could not predict individual risk, knew that most of even the highest
risk defendants would succeed pretrial, and knew that what made defendants
“high risk” to begin with was a subjective determination (through definitions
and cutoffs) that those defendants might commit virtually any crime while

312 Id. at 783.
313 See, e.g., In re Losasso, 24 P. 1080, 1082 (Colo. 1890) (“bail must be reasonably sufficient to secure the
prisoner’s presence at the trial”); People v. Lanzieri, 25 P.3d 1170, 1175 (Colo. 2001) (“The right to
reasonable bail . . . following arrest lessen[s] the impact of an unlawful arrest.”); Ex parte Ryan 44 C. 555,
558 (Cal. 1872) (Bail is excessive when it is “unreasonably great, and clearly disproportionate to the
offense involved, or the peculiar circumstances appearing must show it to be so in the particular case.”).
314 See Stack v. Boyle, 342 U.S. 1, 10 (1951).
118
on release, it would likely be deemed reasonable for that court to release all
defendants pretrial. This presents a monumental shift in thinking from
traditional or historical excessive bail analysis, which, through a discussion
of “reasonableness,” allowed courts to simply compare equally arbitrary
numbers associated with different cases to come to a result.
Due Process and Equal Protection Generally
Looking broadly at due process and equal protection, understanding risk
research highlights issues of fairness. For due process, is it fair to consider
detention based solely on actuarial pretrial risk assessment instruments when
those tools have subjective or political elements? Is it not arbitrary or
unreasonable to detain all “high risk” persons when we only seek to keep
one or two from committing a violent or serious crime? When attempting to
treat similar persons similarly pursuant to equal protection analysis, can we
justify detaining both the “high risk” defendant who might commit an
extremely violent crime and the “high risk” defendant who might only
commit a traffic offense? Is an arrest enough to trigger detention when many
non-arrested persons pose even higher risks?
More particularly, Salerno (as well as case law leading to it) specifically
informs that to survive due process scrutiny a proper detention provision
requires both a net and a further limiting process to make sure that detention
is the “carefully limited exception” to release.315 Specifically, Salerno
approved the Bail Reform Act’s limitation of detention to “a specific
category of extremely serious offenses,”
316 a net created by Congress based
on certain assumptions associating higher risk to those charges. Moreover,
any limiting process developed in the wake of Salerno would likely require
an adversary hearing to at least determine by clear and convincing evidence
that no condition or combination of conditions could reasonably assure
public safety or court appearance, and would need to craft the overall
assessment to focus on “the nature and seriousness of the suspect’s
release.”
317
Based on this, important questions loom for jurisdictions seeking to change
their release/detain dichotomies. Can actuarial risk assessment, which in
most cases labels persons as “high risk” based on their likelihood of

315 See Salerno, 481 U.S. 739 at 755 (1987).
316 Id. at 750.
317 Id. at 743, 750.
119
committing virtually any crime on release, ever be considered an adequate
basis for creating a detention eligibility net? Likewise, can our further
limiting processes use an actuarial tool when the tools themselves appear to
broaden certain nets? The PSA Court’s violence flag is a step in the right
direction, and that tool, like others, can be an invaluable tool for all aspects
of release, but many of the fundamental shortcomings of prediction still exist
when considering the somewhat drastic remedy of pretrial detention.
Fair Notice
One of those shortcomings deals with the fact that preventive detention
based solely on risk gets dangerously close to violating due process based on
the premise that in America, “we insist upon limiting the criminal law to
enforceable rules about the specific conduct in which men may or may not
engage rather than confining all persons with criminal propensities before
their deeds are done.”
318 Put another way, Herbert Packer wrote, “[i]t is
important, especially in a society that likes to describe itself as ‘free’ and
‘open,’ that a government should be empowered to coerce people for what
they do and not for what they are.”319 Accordingly, “‘the criminal law ought
to be presented to the citizen in such a form that he can mold his conduct by
it, that he can, in short, obey it.’ Due process forbids punishment that one
has no assured way to avoid.”320 In sum, people should be able to order their
lives to be able to stay out of trouble, and the law should be written in clear
ways to discourage discriminatory enforcement. Author Christopher
Slobogin writes as follows:
The constitutional version of this principle is vagueness
doctrine, which as a matter of due process requires invalidation
of statutes that do not sufficiently define the offending conduct.
The purposes of vagueness doctrine are to ensure citizens have
notice of the government’s power to deprive them of liberty and
concomitantly to protect against the official abuses and the
chilling of innocent behavior that can occur if government
power is not clearly demarcated.321

318 Tribe, supra note 1, at 394-95.
319 Christopher Slobogin, Defending Preventive Detention, at 70 (Oxford Press 2009 ) (Eds. Paul H.
Robinson, Stephen P. Garvey, Kimberly Kessler Ferzan) [hereinafter Slobogin] (quoting Herbert Packer,
The Limits of the Criminal Sanction, 74 (1968)).
320 Tribe, supra note 1, at 395 (quoting L. Fuller, The Morality of Law, at 105 (1964)).
321 Christopher Slobogin, A Jurisprudence of Dangerousness, 98 N.W. U. L. Rev. 2, 18 [hereinafter
Slobogin Dangerousness] (2003-04) (internal footnotes omitted).
120
This concern should be foremost in jurisdictions’ minds even though the
Supreme Court has labeled preventive detention “regulatory restraint” and
not “punishment” in the traditional sense.322 In short, “Vagueness doctrine
should govern the scope of preventive detention laws even if it is assumed . .
. that such laws are not ‘criminal’ in nature.”323 This accords with analyses
by other legal scholars, who have commented on the Court’s application of
“fair notice” outside of the criminal law.324 Indeed, Eugene Volokh writes
that at least one recent Supreme Court opinion suggests that “fair notice”
might apply “whenever there’s any legal effect, even a modest one that falls
far short of criminal punishment.”325
Vagueness has been largely ignored in the past when bail schemes were
designed to detain persons based only on terms such as “dangerousness” and
“community safety,” but it is highly relevant today as jurisdictions try to
make sense of the risk research and how that research applies to making an
initial determination about release and detention. In sum, the notion of
adequately describing triggering conduct is crucial to the criminal law
generally and equally so when discussing pretrial detention. Indeed, the fact
that we have laws on the books describing failure to appear for court or
committing new crimes while on release is a way of giving advance notice to
persons that those things will bring some governmental response during the
bail process. Under a theoretically pure charge-based detention eligibility
net, a person may reasonably believe that he or she will not be detained
pretrial unless he or she is charged with committing a crime within the net.

322 Slobogin writes that despite consideration of a logical syllogism that preventive detention is not
punishment (i.e., punishment occurs after conviction; with preventive detention there is no conviction;
accordingly, there is no punishment), “[I]f a liberty deprivation pursuant to a prediction fails to adhere to
the logic of preventive detention . . . then it can become punishment” when held up to the general due
process requirement that “‘the nature and duration of commitment bear some reasonable relation to the
purpose for which the individual is committed.’” Id. at 13 (quoting Jackson v. Indiana, 406 U.S. 715
(1972)).
323 Id. at 18.
324 See, e.g., Theodore J. Boutrous, Jr. & Blaine H. Evanson, The Enduring and Universal Principle of
“Fair Notice,” 86 So. Cal. L. Rev. 193 (2013).
325 Eugene Volokh, The Void-for-Vagueness/Fair Notice Doctrine and Civil Cases (June 21, 2012), found
at http://volokh.com/2012/06/21/the-void-for-vagueness-fair-notice-doctrine-and-civil-cases/. That opinion,
from FCC v. Fox Television Stations, 132 S. Ct. 2307 (2012), applied the fair notice doctrine to a regulated
entity, and even mentioned “reputational injury” beyond even regulatory “punishment” as a basis for relief.
Id. at 2318-19. Vagueness applies both to ensure that affected persons know what is required of them so
they may act accordingly as well as to ensure that “those enforcing the law do not act in an arbitrary or
discriminatory way.” Id. at 2309. A “risk-based” detention eligibility net implicates both concerns: persons
will not be able to glean how to keep from being “risky,” and the somewhat arbitrary nature of the risk
tools themselves (along with the ability for overrides) can easily lead to arbitrary enforcement.
121
That reasonableness evaporates when that net is described only in terms of
risk, using actuarial pretrial risk assessment instruments based on
subjectively broad definitions and labels of “risk,” public safety, and flight,
and on aggregate determinations of risk, which reflects the conduct of others
that cannot be controlled by any particular individual.
326
Take, for example, the new constitutional right to bail provision enacted in
New Jersey, a state that desired to move from a “charge and money-based”
release and detention system to one based more on empirical risk. The
previous constitutional language articulated a right to bail for all defendants,
“except for capital defendants when the proof is evident or the presumption
great,” a broad right to bail provision modeled after the Pennsylvania law of
1682.327 Theoretically, under this prior language, persons would know that
unless they committed a capital crime, they would have a right to bail. As
mentioned before, that right – historically meant to be a right to release – has
been eroded over time and practically eviscerated through the use of money.
Nevertheless, the right was there for all who did not commit capital crimes.
The new bail language, however, states than any person can be denied
pretrial release “if the court finds that no amount of monetary bail, nonmonetary conditions of pretrial release, or combination of monetary bail and
non-monetary conditions would reasonably assure the person’s appearance
in court when required, or protect the safety of any other person or the
community, or prevent the person from obstructing or attempting to obstruct
the criminal process.”328 By itself, this language would be wholly incapable
of warning individuals of what conduct might lead to pretrial detention. The
lack of adequate conditions might be determined subjectively, or even based
on adequate government resources. The new provision is presumably based
on notions that certain defendants are dangerous and flight risks, but even if
the constitution expressly said so, persons would have a difficult time
ordering their lives to somehow remain un-dangerous or un-risky for flight,
however those things might be defined.
The New Jersey statute limits the constitutional detention language to
“eligible defendants,” who are persons charged with indictable crimes,
which are equivalent to felonies elsewhere, and “disorderly persons

326 As noted in the Harvard Law School Primer, supra note 3, at 22-23, and n. 195, “While an individual’s
conduct is within his control, that individual cannot control the aggregate conduct of others who share some
characteristic deemed relevant for the risk assessment instrument.”
327 N.J. Const. art. I, § 11 (2013).
328 Id. (2017).
122
offenses,” which are equivalent to a broad range of misdemeanors including
possession of marijuana under 50 grams, simple assault, shoplifting of less
than $200 worth of merchandise, resisting arrest, underage possession of
alcohol, bad checks, and possession of a fake ID. While not posing the acute
subjectivity problems of a risk-based net, this is an extremely broad chargebased detention eligibility net – much broader than the net reviewed by the
U.S. Supreme Court in United States v. Salerno.
329 Within that net, a
provision in the New Jersey law allows a prosecutor to move to detain
eligible defendants for a certain array of clearly defined crimes (such as a
crime with punishment of life in prison), but also for “any other crime for
which the prosecutor believes there is a serious risk that: (a) the eligible
defendant will not appear in court as required; (b) the eligible defendant will
pose a danger to any other person or the community.”330 This nearly
limitless net, coupled with a statutory mandate to use statistically-derived
risk assessment to determine release and detention,331 makes it virtually
impossible for anyone to conduct themselves in ways that would clearly
avoid pretrial detention.
In New Jersey, as in the rest of America, no one should fear that pretrial
detention – sometimes lasting weeks or months – will be possible for the
vast majority of crimes, just as no one should fear the death penalty as a
possible punishment for all crimes. And whatever actuarial pretrial risk
assessment instrument is ultimately used in that state, it might have the same
limitations discussed above – for example, it might label risk levels
somewhat subjectively; it might determine its cutoffs subjectively and
possibly even for political purposes;
332 and it might define a risk to public
safety and flight in such broad terms as to make virtually all defendant

329 Even defenders of preventive detention have written that, “If the state’s preventive detention power is
not limited by the requirement that it prove some affirmative act that is predictive of a legislatively defined
danger, then the government, not the individual, controls if and when the government intervenes.
Conditions, dispositions, and thoughts, even if highly predictive of danger and identified as such, cannot be
the ‘point of no return’ described by Packer because there is no identifiable ‘point’ at which they can be
avoided. Slobogin, supra note 319, at 3-4 (quoting Herbert Packer, The Limits of the Criminal Sanction, at
74 (1968)). While this affirmative act need not always be articulated as a crime, it is, by far, the most
rational way to do so within the criminal law. Moreover, that conduct, when articulated in terms of a crime,
must be narrow. When arguing before the Salerno Court in support of the Bail Reform Act of 1984, the
government itself sought to assure the Court that by using a limited charge-based net, the Act would not
“grant federal courts a roving commission to ferret out dangerous individuals wherever they may be
found.” See Brief of United States of America, United States v. Salerno, 1986 WL 727530, at 12 (1986).
330 N.J. Stat. Ann. § 2A:162-19 (2017).
331 See Id. § 2A:162-17 (3) (a), (b).
332 See Harvard Law School Primer, supra note 3, at 21 (writing that a given characterization or definition
of a risk level “is a policy judgment, not a statistical one”).
123
conduct potentially detainable.333 As explained by von Hirsch, such a system
would eviscerate any safeguard based on giving persons the ability to avoid
the coercive effects of the law and to determine their own fates:
An individual would have little choice as to whether he is
confined or remains at large. His liberty would depend not upon
his voluntary acts, but upon his propensities for future conduct
as they are seen by the state. Far from being able ‘to identify in
advance the space which would be left free to him from the
law’s interference,’ his liberty would depend upon predictive
determinations which he would have little ability to foretell, let
alone alter by his own choices.334
As noted previously, these issues should not be ignored based on the notion
that preventive confinement is not technically deemed “punishment.” Even
if regarded as simply precautionary (or “regulatory,” as explained by the
Supreme Court in Salerno), preventive detention provisions based solely on
risk can still provide little guidance to persons hoping to avoid incarceration.
Moreover, whether punishment or not, the detention of persons who are not
actually dangerous – the so-called false positives – is nonetheless unjust:
The force of this argument – that preventive confinement of the
false positives is essentially unjust – does not, in fact, depend
upon whether such confinement is classified as punishment.
Even if it is regarded as a precautionary, rather than a punitive
measure, the justification of preventively confining an
individual would depend upon his actually being dangerous.
The individual is being deprived of his liberty because, if he
were to remain at large, he would interfere with the liberty of
others by committing crimes. If he is not in fact dangerous, this
justification simply collapses; and what we have left is
gratuitous suffering imposed upon a harmless individual.335
Accordingly, when re-drawing the line between pretrial release and
detention, jurisdictions must remember how risk research and actuarial

333 This is not much different from other charge-based state bail schemes today, which often include broad
detention eligibility nets and that do not adequately define terms such as “danger” and “community safety.”
334 von Hirsch, supra note 281, at 746 (quoting H.L.A. Hart, Punishment and Responsibility 181-82
(1968)).
335 Id. at 743, n. 74.
124
pretrial risk assessment instruments interact with the law, especially the law
surrounding excessive bail, due process, and equal protection. Each of these
fundamental legal principles suggest, again, that we should err on the side of
release and on constantly narrowing any system of detention that can be
justified through the research, the history, or the law. Jurisdictions must
remember to fully define flight and danger – the “risk of what” – when
articulating the test for detention, and to carefully avoid vagueness when
crafting laws that impact human liberty.
Can We Use An Actuarial Pretrial Risk Assessment
Instrument Solely As Our Eligibility Net When We Re-Draw
the Line Between Pretrial Release and Detention?
Using one of today’s actuarial pretrial risk assessment instruments solely to
draw a line between release and detention – for example, by saying that a
particular state will detain only “high risk” individuals as measured by a risk
tool or by creating an unlimited charge-based net to be sorted out later by
risk tool, while tempting, would be wrong. It is tempting because the idea
contains superficial logic, and it gets us back to the historical ease of
assessing risk prior to labeling a defendant either “bailable” or “unbailable.”
Unfortunately, however, the various notions discussed above – that while the
risk research used to create them is fairly unassailable, the structure and
application of current risk tools through labels and cutoffs is somewhat
subjective336 and political; that they lead to overestimates of risk; that they
do not eliminate the problems with base rates and false positives; that they
do not necessarily even measure the type of risk to which we are trying to
respond; that they are vague when used as a standard; and that all of these
things implicate and potentially offend fundamental legal notions underlying
bail – mean that we: (1) must never use them solely to determine release or
detention in the first instance based on risk; (2) must never use them in
creating our detention eligibility nets; and (3) must never use them to
automatically determine defendant detention within a wide charge-based net.
Jurisdictions using so-called “bail guidelines,” “matrices,” or other such
documents that guide courts toward detention in certain cases, must also

336 Allowing detention based on a finding that “no condition or combination of conditions” suffice to
provide adequate assurance of public safety or court appearance ( a standard often used today) is equally
subjective, and has additional problems associated with resources (jurisdictions with fewer resources are
likely to detain more defendants based on this standard). Detaining based on the actuarial risk tool requires
special caution because its subjective nature and lack of adequate definitions to determine severity of risk
are somewhat masked by the outward appearance of objective science.
125
understand the limitations in using actuarial pretrial risk assessments
instruments that have an overinflated role in the detention decision.
While the Supreme Court has previously rejected arguments against using
prediction in the release or detention process, those arguments likely only
failed because there existed a backstop – in the form of a charge-based
detention eligibility net – to restrain detention to constitutionally acceptable
levels. The Court was faced with using prediction only among a small set of
defendants facing extremely serious crimes. It was not faced with a
prediction method that could potentially lead to detention of countless “high
risk” individuals on relatively minor charges. Everyone is risky, and some
persons simply walking down the street today would be deemed “high risk”
if they were merely stopped and measured with an actuarial tool.
Accordingly, jurisdictions must determine, in advance, when it is proper to
assess this pre-existing risk. It may not be proper after stopping someone for
a traffic violation, but it might be proper after arresting a person on a violent
felony. But in both cases, and wherever that line is ultimately drawn, the risk
used to detain someone pretrial should be the kind of risk to which we seek
to respond with detention.
Thus, jurisdictions must constantly remind themselves that everything we
have learned from the history of bail, the law surrounding release and
detention, and the pretrial research points to discerning a different kind of
risk to detain than that currently provided by actuarial pretrial risk
assessment instruments today. Put another way, actuarial risk assessment
tools provide the best way to measure the kind of risk that those tools
measure. But until they adequately answer whether a defendant poses a
substantial and unmanageable risk of willful flight versus simply failure to
appear for court, and risk of committing a serious or violent crime against
knowable persons versus the risk of committing any crime against
potentially all persons, they should never be used solely to determine
detention eligibility in the first instance (i.e., based on prediction alone).
Accordingly, and most importantly, when re-drawing the line between
pretrial release and detention, jurisdictions must remember not to use results
from the current generation of risk assessment instruments to create their
detention eligibility nets, which should more appropriately be based on
justifiable and limited categories of criminal charge. In sum, there are two
types of risk today. There is the risk as measured by the risk tool, and there
is the risk that we may use to detain. While risk assessment instruments can
126
be helpful tools, jurisdictions must “look under the hood” of these
instruments to determine exactly what they show, and be prepared to use
“risk as measured by the tool” perhaps primarily for determining conditions
of release for defendants outside of the eligibility net as well as defendants
within the net who are nonetheless released into the community.
Will Future Actuarial Pretrial Risk Assessment Instruments
Theoretically Be Sufficient to Function as a Detention
Eligibility Net?
As noted previously, a perfect pretrial risk assessment instrument would give
jurisdictions a 100% probability that a particular person would do the
particular thing we fear during pretrial release. Today we are far from that
perfect tool, but the research continues to improve. Indeed, today we are
now better able with some assessment instruments to predict the risk of a
defendant committing a violent crime while on release. This ability, while
groundbreaking, is still likely not enough to overcome the legal and policy
problems associated with relying on aggregate risk to determine detention.
For example, an assessment tool might tell us that a person looks like other
people who are risky for violent behavior, but it still will not tell us that this
defendant is so risky. Moreover, while the tool may indicate “high risk,” it
will not tell us why “high” was determined to be at that particular cutoff, and
it will still likely overestimate risk and lead to an unacceptable number of
false positives. For a while, at least, it may still differ jurisdiction to
jurisdiction. Importantly, it will not provide any basis for persons to guide
their behavior to avoid being labeled risky and detained. And finally, due to
all of these issues, without boundaries it will likely lead to assessing and
potentially detaining defendants charged with any criminal offense, possibly
violating the Due Process, Equal Protection, and Excessive Bail Clauses as
well as American norms that include more risk tolerance for minor crimes.
While nearly unimaginable to think it could happen, in the worst case it
could nonetheless lead to a government rounding up all “risky” or
“dangerous” individuals on any charge, knowing that being so labeled will
lead to detention.
This author has heard the argument that, in some distant future, police
officers will not arrest as many persons (possibly using a risk tool for that
decision), and then a near-perfect risk tool – meaning it performs the same
or better than base rates, eliminates nearly all false positives by somehow
127
better identifying individual risk, is the same in every jurisdiction, is
somehow created in a way that reduces or eliminates subjectivity and
politics, and that adequately assesses risk for flight (versus FTA) and serious
and violent criminal activity (versus all criminal activity) – will be used to
sort defendants into a net. Even then, and despite other remaining issues
addressed throughout this paper, jurisdictions will still need to further sort
defendants based on charge or risk having those schemes declared unlawful
based on multiple legal theories. The need for a charge-based net exists no
matter how or when we use even near-perfect actuarial risk tools in the
process.
Nonetheless, jurisdictions must be constantly reminded that using actuarial
pretrial risk assessment instruments to do anything is nonetheless far
superior to using money, as we have in America since the 1800s. Money has
no empirical justification and offends legal principles far more readily and
completely. Moreover, actuarial pretrial risk assessment is better than using
charge-based schemes that have no basis or justification underlying them.
Thus, if a jurisdiction simply said that it intended to change its pure chargeand-money-based system with one based on results from a risk tool, it would
likely be preferable to the previous system and might even be deemed
rational and lawful by an appellate court. This paper, however, looks to
create an “ideal” process in an era when money will not be available to
detain; an era when jurisdictions must articulate, up front, who, if anyone,
they may purposefully initially detain pretrial. While someday the pretrial
research may reach a point at which it will overcome the various hurdles
associated with solely using actuarial risk tools to reach that ideal, in this
author’s opinion, today is not that day.
What Do the National Standards Tell Us About Re-Drawing
the Line Between Pretrial Release and Detention?
The national standards concerning pretrial release and detention provide
concrete recommendations based on the law and the research. Both the
American Bar Association (ABA)337 and the National Association of Pretrial
Services Agencies (NAPSA)338 have standards, but because the current
NAPSA Standards are virtually identical to the ABA Standards for the

337 ABA Standards, supra note 100.
338 Standards on Pretrial Release (3rd ed.), Nat’l. Assoc. of Pretrial Servs. Agencies (Oct. 2004).
128
relevant provisions (moreover, at the time of this writing, they were being
updated) this paper will only briefly describe the ABA Standards.
For the most part, the ABA Standards reflect notions underlying the “big
fix” as found in the 1970 D.C. Court Reform Act and the Bail Reform Act of
1984. Thus, they recommend an in-or-out decision making process that is
fair and transparent and that has nothing, like money, impeding the decision
to release or detain. Likewise, following the opinion in United States v.
Salerno, the Standards attempt to create primarily a charge-based detention
eligibility net along with a detention hearing procedure both to further limit
detention and to provide the appropriate due process protections necessary to
deprive one of his or her liberty. Broadly, the Standards provide justification
for a narrow detention eligibility net (and thus a broad presumption of
release) by stating that the law favors release pending trial, which “is
consistent with Supreme Court opinions emphasizing the limited permissible
scope of detention.”339
The current detention provisions in the ABA Standards are found in
Standards 10-5.6 through 10-5.10. These provisions are part of an overall
scheme allowing for three separate triggers leading to pretrial detention. The
first trigger occurs if a defendant violates a condition of release, including a
new crime or willful failure to appear for court, and the court considers
revocation of release followed by a detention hearing.340 If a judicial officer
finds probable cause for a new crime while on release or clear and
convincing evidence of a violation of other conditions, that officer may
follow Standard 10-5.8 to initiate a detention hearing.
The second trigger occurs whenever persons are charged with a crime and
are: (1) already on release pending trial on another charge; or (2) on release
pending sentencing or appeal; or (3) on probation or parole for any offense,
and “may flee or pose a danger to the community of to any person.”341 When
this occurs, the Standards recommend temporary detention for a
recommended three days, “to allow time for the jurisdiction or court that
released the defendant in the original case to decide whether to modify
release conditions, initiate a revocation hearing, or lodge a detainer before

339 ABA Standards, supra note 100, Std. 10-1.1 (commentary), at 38 (citing Salerno and Stack v. Boyle, 342
U.S. 1, 4-5 (1951)).
340 See Std. 10-5.6, at 116-17.
341 Id., Std. 10-5.7, at 120.
129
the arresting jurisdiction takes action on the new charges.”342 At the end of
the period of temporary detention, the court must initiate a full detention
hearing or release the defendant on conditions.
The third trigger involves making a determination in the first instance based
on prediction that a defendant presents an unmanageable risk of either flight
or public safety warranting secure confinement.343 This initial detention is
the primary subject of discussion within this paper; nevertheless, when
considering where to re-draw the line between release and detention,
jurisdictions should question whether they have the ability to temporarily
detain defendants in certain circumstances, and, perhaps more importantly,
whether they have the ability to revoke a bond and order detention if a
defendant willfully violates fundamental conditions of release.344 These are
elements of a proposed model process, which is revealed later in this paper.
The Detention Eligibility Net
As mentioned previously, the Standards create a detention eligibility net
(limited by charge except for risk to witnesses and jurors) for pretrial
detention in the first instance (based solely on prediction), which is
articulated in Standard 10-5.9, and which is wider than the net articulated in
the previous editions of the Standards. A brief history of the evolution of the
current net is helpful to the instant discussion.
The ABA Standards Relating to Pretrial Release were created in 1968, and
while those Standards contained provisions for revocation of release, the
first time that the Standards articulated recommendations for a general
procedure for pretrial detention was in the Second Edition, published in
1979.345 Those recommendations were an admitted attempt merely to
“alleviate” some of the continuing problems associated with the bail system
that were not fixed in the first generation of bail reform (and the first edition
of the Standards), including defendants being held due to lack of money.346
The 1979 Standards attempted to do this primarily by establishing “[a] fair

342 Id. (commentary) at 123.
343 See id., Stds. 10-5.8 through 10-5.10.
344 States are advised to check their case law, as occasionally courts interpret constitutional bail provisions
to either allow or deny the ability to detain a “bailable” defendant, even when that defendant has committed
a new crime or willfully failed to appear for court.
345 See American Bar Association Standards, Pretrial Release (approved Feb. 12, 1979) [hereinafter 1979
ABA Standards].
346 Id. (introduction), at pp. 10.5-10.6.
130
system for detaining individuals who have engaged in specific pretrial
conduct demonstrating dangerousness, or who cannot meet monetary
conditions necessary to deter flight.”347
This edition of the Standards was clear in expressing that it was not
recommending pretrial detention premised on a general prediction of
dangerousness – something it still saw as “constitutionally dubious,” and
which had been seen as having largely failed through disuse in the District of
Columbia.348 Nevertheless, these 1979 Standards did attempt to provide a
fair and constitutionally acceptable way to detain defendants without using
money “when there is no way to assure their reappearance or because they
have demonstrated that they constitute an unacceptable risk to the
community.”349
As noted above, they did this primarily by articulating a detention eligibility
net triggered either by specific defendant conduct while on pretrial release,
or by instances when a defendant could not meet the monetary condition.
Specifically, commentary to 1979 Standard 10-5.9 read as follows:
There are four ways in which the procedures in this standard
can be triggered: (1) by a judicial determination . . . that
monetary conditions are necessary to assure reappearance and
the defendant’s failure to satisfy those conditions; (2) by a
judicial determination . . . that there is probable cause to believe
that a defendant has willfully violated a condition of release; (3)
by a judicial determination . . . that there is probable cause to
believe that the defendant has committed a new crime while on
pretrial release; or (4) upon a formal complaint executed by the
prosecutor, a law enforcement officer, or a representative of the
pretrial release agency alleging that the defendant is likely to
flee, threaten or intimidate witnesses, or constitute a danger to
the community.350
While the fourth category appears quite broad, the Standards further
narrowed it by requiring judicial findings based on specific defendant

347 Id. at pp. 10.98.
348 Id. Std. 10-5.9, at p. 10-98. It is widely known that the District of Columbia preventive detention
provisions were not used until money was removed as potential means to detain. See D.C. Lessons, supra
note 164, at 5.
349 Id. Std. 10-5.9, at p. 10.99.
350 Id. at p. 10.100.
131
conduct – such as new criminal activity or breach of a release condition –
that “demonstrate[es] in a concrete way that he or she poses an unacceptable
risk to the community and ought to be detained.”351 Likewise, for flight,
using the fourth category had to be accompanied by a showing either that the
defendant could not make his or her monetary condition, or that he or she
had violated some other condition designed to provide reasonable assurance
of court appearance. The Standards then provided recommendations for a
“procedurally fair and rigorous” due process detention hearing – the kind of
hearing that had been included in the 1970 D.C. Act and was later part of the
Bail Reform Act of 1984, which was ultimately reviewed by the Supreme
Court in United States v. Salerno.352
In sum, the 1979 Standards proclaimed as follows:
Pretrial detention, under the circumstances and with the
protections provided for in this standard, is clearly
constitutional. Standard 10-1.2 requires that the release of every
defendant be conditioned on the defendant’s refraining from
criminal activity and interfering with witnesses, and standard
10-5.2 empowers a judicial officer to impose additional
nonmonetary conditions of release to ensure the defendant’s
appearance in court, protect the safety of the community, and
prevent intimidation of witnesses. With one exception, every
category of defendants detained pursuant to standard 10-5.9
would have violated one of these conditions. . . . All this
standard does is to provide a procedurally fair mechanism for
determining when such a violation has occurred.
The only circumstance in which detention is not premised on a
violation is when the defendant is unable to meet monetary
conditions necessary to ensure reappearance. In these
circumstances, this standard merely provides an added layer of
procedural protection for a defendant who would, in any event,
be detained under the traditional bail systems.353

351 Id. at p. 10.101.
352 See id. at pp. 10.96-10.97. There are some key differences, however. For example, the 1979 Standard
requires courts to use normal criminal trial evidentiary rules when premising detention upon a new criminal
offense.
353 Id., at p. 10.102.
132
It was an admittedly middle-ground solution, designed to “soften, if not
eliminate” the conflicts posed by using money.354 But the 1979 Standards
did provide a strong statement against what has become commonplace
today: pretending that a defendant has not been detained simply because he
was ordered released on unattainable conditions: “[These Standards] end the
hypocrisy of pretending that defendants too poor to post bail have been
‘released’ on monetary conditions. Such defendants obviously have not been
released; they have been detained, and it follows that they should be
afforded precisely the procedural protections granted to other detained
defendants.”355
In 1986, the ABA released supplements to the 1979 Standards, in which that
organization ultimately recommended procedures for denying initial release
for certain defendants based on concepts of pure preventive detention.
Specifically, the ABA noted as follows:
While the 1979 standards (standard 10-5.9) recognized
‘dangerousness’ and while they took into consideration issues
regarding the safety of the community, those factors did not
enter into play until a defendant violated a condition of release,
committed a new offense, or otherwise demonstrated by acts or
omissions that continued release would be inimical to
community safety and the orderly administration of criminal
justice.356
Accordingly, the 1986 supplements revised the 1979 Second Edition
Standards to “recognize[] the legitimacy of initial preventive detention for a
certain limited class of defendants when their dangerousness has been
proved under specific criteria and with appropriate procedural
safeguards.”357 Specifically, those supplements retained the 1979 provisions
that allowed detention based on violating conditions, committing new crimes
while on release, and the inability to meet monetary conditions.
Nevertheless, the Standards added provisions for initial preventive detention

354 Id. at p. 10.103.
355 Id. at p. 10.103. The need for this hypocrisy-ending stance is more necessary than ever today. Ever since
the Excessive Bail Clause was interpreted to mean that defendants do not have a right to “bail” they can
“meet,” courts have used monetary conditions to detain, typically with no due process hearing whatsoever.
It is perhaps the greatest failure in American bail law today.
356 American Bar Association Standards, Pretrial Release (Explanatory Note), at 7S (Supp. 1986)
[hereinafter 1986 Supplements].
357 Id. at 8S (internal citations omitted).
133
with a net consisting of the following categories of detention eligible
defendants: (1) defendants charged with a violent felony allegedly
committed while on pretrial release, probation, or parole in connection with
another violent felony; and (2) defendants charged with a violent felony who
had been convicted of another violent felony within the past ten years.358
Interestingly, the Standards considered and rejected a proposal to include a
net of “any crime of violence” (settling instead only on felonies), and
whether directed at persons or property (settling instead only on persons).
359
Technically speaking, a person arrested for a violent felony who then is
arrested for another violent felony while on release could be detained for
violating a condition of release for his first alleged offense; the new standard
would make detention appropriate for the second alleged offense and
included release on probation and parole as qualifying preconditions. The
bigger difference in overall detention policy, though, was in allowing for
detention for defendants charged with violent felonies whose,
pattern of behavior, consisting of past and present conduct, and
specifically including a conviction for at least one felony
involving violence within the preceding [ten] years, supports a
judicial finding that no condition or combination of conditions
will reasonably assure the safety of any person and the
community, or reasonably prevent intimidation of a witness and
interference with the orderly administration of criminal
justice.360
This Standard took care, however, to caution jurisdictions that while a prior
violent felony conviction was a predicate to the detention determination,
those jurisdictions should not use a prior violent felony conviction as the
sole basis for detention. As noted in the analysis, “a prior conviction,
standing alone, cannot legitimate a preventive detention order; conversely, a
person without a prior felony record cannot be detained preventively, even
though there is a fear that persons may be harmed or the effective
administration of criminal justice imperiled.”361

358 Id. Std. 10-5.4, at 10.28S-32S.
359 See id. (analysis), at 31S.
360 Id. Std. 10-5.4 (a) (i) (B), at 28S.
361 Id. (analysis), at 32S.
134
Nevertheless, by recognizing the legitimacy of preventive detention based
on prediction of danger beyond what had previously been considered as
some “inherent” judicial ability to detain in extremely rare cases, this
recommendation took a significant step toward initial purposeful detention
of noncapital defendants in America. Compared to today’s Standards (and
laws based on those standards), the 1986 net appears fairly narrow by
continuing to base detention on individual defendant conduct. That net
would be considerably broadened, however, in the next (and current) edition
of the ABA Standards, which was approved in 2002 and published with
commentary in 2007.
The current edition of the Standards reserves detention for four categories of
defendants, which are “intended to encompass those defendants most likely
to present a danger or fail to appear.”362 The categories are as follows: (1)
defendants charged with a crime of violence or a dangerous crime; (2)
defendants charged with a “serious” offense who are already on release on a
different case that is also a serious offense unless the defendant was on
release pending sentencing or on appeal (if the defendant was on probation
or parole, the underlying conviction must be for a serious and violent or
dangerous offense); (3) defendants charged with serious offenses who pose
“a substantial risk . . . [to] fail to appear for court or flee the jurisdiction;”
363
or (4) defendants charged in any case “who pose a substantial risk of
obstructing justice or threatening, injuring, or intimidating prospective
witnesses or jurors.”364
Commentary to this Standard provides that the “substantial risk” component
to this fourth category of detention eligibility “requires that there be a
showing of facts pointing to unacceptable behavior by the defendant (such as
intimidating witnesses) if released. The facts could be found in the risk
assessment prepared by the pretrial services agency and/or in evidence
provided by the prosecution.”365 However, based on the earlier discussion
within this paper concerning a risk assessment instrument’s tendency to
measure aggregate risk as well as the risk of something somewhat different
from the sort of flight or danger of historical concern in America, it is likely
that the better evidence will often be found outside of the risk tool. This
notion is reinforced in Standard 10-5.8, which includes commentary

362 ABA Standards, supra note 100, Std. 10-5.9 (commentary) at 130.
363 Id. Std. 10-5.9, at 129
364 Id. (commentary), at 132.
365 Id.
135
suggesting that certain evidence of relevant riskiness will not necessarily be
found in the risk instrument, but rather in things such as the weight of the
evidence against the defendant or the arguments of counsel.366 Overall, this
fourth category tends to follow the history of intentional detention in
America, which first found justification for detaining noncapital defendants
when facts and circumstances tended to show the potential for specific bad
behavior to a discreet group of persons (specifically, witnesses and jurors) if
released. As noted previously, detaining such defendants was believed to be
within a court’s “inherent” power to conduct trials.367
The Standards leave it up to individual jurisdictions to define “crime of
violence” and “serious” offenses, but do note that serious crimes would
“clearly encompass some offenses that are not violent or physically
dangerous.”368 As noted previously, the 1970 D.C. law defined “dangerous”
crimes to be narrower than “violent” crimes, and the subjective nature of
these terms should be considered when attempting to re-draw the line
between release and detention.
There are pros and cons associated with the previous and the current
detention eligibility nets from the Standards. The most obvious change over
time is a slow progression toward more opportunities for purposeful
detention, including a clear widening of the detention eligibility net from the
previous versions to the current Standards. As noted previously, this may
have been prompted – as with the Bail Reform Act of 1984 – by the
Standards’ incorporation of language recommending judicial officials not to
impose financial conditions that result in pretrial detention due to inability to
pay, thus requiring some honest method for detaining risky defendants.
369
Nevertheless, it appears that the details concerning this overall broadening of
detention eligibility was based more on prevailing assumptions of defendant
risk rather than on actual research. Indeed, because of troubling questions
over various aspects of the 1970 D.C. detention provisions, the 1979 edition

366 Id. Std. 10-5.8 (b) (commentary), at 127.
367 Theoretically, at least, there are some factual scenarios here (such as a defendant personally obstructing
justice by bribing jurors) in which defendants would not necessarily be detainable for flight or general
public safety purposes. Following Salerno, most jurisdictions have dropped the overall historical distinction
between witnesses and jurors versus the general public by defining public safety to include any law
violation. Bribery, in this sense, would be considered a public safety “failure.”
368 Std. 10-5.9 (a) (commentary), at 130.
369 See Std. 10-5.3 (a), at 110. Though the Standard says that judicial officers “should not impose” financial
conditions leading to detention, commentary to that Standard states that the provision “prohibits” judicial
officers from doing so. Id. (commentary), at 112.
136
of the Standards were clear in requiring defendant conduct rather than a
“generalized prediction of dangerousness” to trigger possible detention.370
Moreover, when the 1986 Supplements were published adopting initial
preventive detention for a certain small class of dangerous defendants, they
cited no research suggesting that defendants were higher risk when facing
violent felonies (instead, they also required another form of defendant
conduct, a prior violent felony, to trigger detention eligibility) under an
apparent assumption that those defendants should be considered higher
risk.371
Likewise, under the current set of Standards, there is no research cited for
why the detention eligible categories in that set, as opposed to any earlier
set, are thought “to encompass those defendants most likely to present a
danger or fail to appear.”372 The current Standards, for example, make an
assumption that certain defendants facing “serious” charges are at a higher
risk to flee, but they base that assumption not on research but the idea that
some defendants in that category would likely have access to large amounts
of money showing motivation to abscond.373 The Standards should not be
faulted for making these assumptions. Rather, the Standards merely reflect
the way America was thinking prior to any research contradicting those
assumptions, which was the same thinking that created the 1970 D.C. Act,
the Bail Reform Act of 1984, and, indeed, the opinion in Salerno. Overall,
the current Standards appear to have done the best job possible given that
there was very little so-called risk research, such as the kind that might
indicate which, if any, crimes may or may not be associated with higher risk
to fail.
Support for this supposition is suggested through discussion earlier in the
Standards concerning a recommendation for release on recognizance. In that
particular Standard, the ABA articulates its recognition that the risk research
might, in fact, point to counterintuitive conclusions. Nevertheless, it makes
the case for why “risk” surrounding a “more serious” crime is qualitatively
different than risk for a “less serious” crime, even when the risk might be

370 1979 ABA Standards, supra note 345, Std. 10-5.9 (history), at p. 10.98.
371 See 1986 Supplements, supra note 356, Std. 10-5.4, at 10.28S.
372 ABA Standards, supra note 100, Std. 10-5.9 (commentary), at 130.
373 Id. at 132.
137
quantitatively higher for the lower offense:
Empirically, there is some evidence that the risk of nonappearance or criminal behavior may actually be greater for
persons charged with relatively minor non-violent offenses
(e.g., prostitution, retail theft, numbers-running, small-scale
drug possession) than for some persons charged with more
serious crimes. However, if a person charged with a serious
offense does in fact commit a similar offense while on release,
the costs to society of the subsequent offense are much greater
than if a defendant charged with a minor offense commits
another minor offense.374
Once one moves from specific instances of defendant conduct to empirical
estimates of individual defendant risk to detain based on aggregate data, one
must find justification for why a particular group of defendants may be
treated differently than others. The above quote thus suggests an attempt to
find a rationale for making different decisions based on charge given that the
risk research (perhaps counterintuitively) often illustrates that some persons
charged with serious crimes are not empirically risky, and some people
charged with less serious crimes are empirically risky. Jurisdictions thinking
of moving toward a more risk-based release and detention system and away
from a primarily charge-based system are directly confronted with this
research. The Standards, therefore, supply a rationale for drawing a line
between release and detention that might withstand scrutiny from the courts:
All things being equal, it is likely necessary to treat certain serious or violent
crimes differently at bail based simply on shared concerns about risk
tolerance.
The Further Limiting Process
Standard 10-5.8 in the current edition of the Standards provides the main
section for detention, and allows pretrial detention after a due process
hearing in which “the government proves by clear and convincing evidence
that no condition or combination of conditions will reasonable ensure the
defendant’s appearance in court or protect the safety of the community or
any other person.”375 The requirement of clear and convincing evidence

374 Id. Std. 10-5.1 (commentary), at 104 (internal footnote omitted). This Standard deals with release and
setting conditions of release, but the rationale is relevant to determining the detention eligibility net.
375 Id. Std. 10-5.7, at 124.
138
reflects the Standard’s intention “to emphasize the deliberately limited scope
for using secure detention. It places a significant burden on the prosecution
to present facts demonstrating why such detention is essential and why the
risks of flight or dangerousness cannot be met through some type of
conditional release.”376 The rest of that Standard includes factors to be used
in deciding whether no conditions will suffice, and includes only one
rebuttable presumption toward detention for persons charged with a capital
offense or an offense punishable by life without parole.377
Finally, Standard 10-5.10 provides recommendations for the requisite due
process hearing necessary for pretrial detention. In the main, it mirrors
provisions found in the current D.C. statute as well as the federal statute,
which was reviewed by the United States Supreme Court in Salerno.378
Overall, this process – arguably including the additional provisions requiring
status reports, immediate appeals, and accelerated trials for detained
defendants – serves as one that further limits detention even within the
eligibility net. To assure that detention is the “carefully limited exception”379
to release, the Standards thus recommend both a net and a further limiting
process designed to withstand legal scrutiny for their rationality and
justification.
Accordingly, when re-drawing the line between pretrial release and
detention, jurisdictions must remember that the ABA Standards, too,
recommend a purposeful in-or-out system, with nothing – like money –
hindering the release or detention decision. Likewise, they must remember
that the Standards reflect the law, the history, and the research at bail to
provide for recommendations that constantly urge jurisdictions to err on the
side of release, to create rational, fair, and transparent but extremely limited
preventive detention schemes, and to provide ample justification for
whatever process is ultimately approved. Nevertheless, jurisdictions must
also hold these aspirational recommendations up to what we know today
about risk, and realize that both the detention eligibility net and further

376 Id. Std. 10-5.8 (a) (commentary), at 127.
377 Id. Std. 10-5.8 (c). This is different from the Federal statute, which has rebuttable presumptions covering
a great many more defendants. The Standards provide no rationale for why they limit the use of rebuttable
presumptions, but it is likely tied to a broader philosophical stance concerning whether defendants should
be forced to shoulder any burden in a criminal case. For a number of reasons, the model crafted within this
paper includes no rebuttable presumptions toward detention.
378 See United States v. Salerno, 481 U.S. 739, at 751-52 (1987).
379 Id. at 755.
139
limiting process articulated by the Standards may need some alteration to
provide adequate legal justification today.
What Have The States Done Up Until Now to Re-Draw the
Line Between Release and Detention?
Every state has already drawn a theoretical line between pretrial release and
detention. Typically, that line is drawn in a state’s constitution – 41 states
have constitutional right to bail provisions – and when these provisions were
enacted, they represented each state’s articulation of who should be given a
right to bail, or release, and who could potentially be denied that right by
being eligible for “no bail, or detention. Most of those early provisions
granted the right to release to everyone except capital defendants, and only
later added additional charges to the “no bail” side. Historically, anyone
deemed bailable was to be released, and so when states articulated a right to
bail for all except capital defendants, for example, those states were at least
theoretically saying that their detention eligibility net consisted of persons
charged with capital offenses, and that everyone else was intended to be
released. States that later changed their right to bail provisions to, for
example, add defendants facing violent felonies to the persons potentially
ineligible for bail were saying that the detention eligibility net consisted of
persons charged with capital offenses and violent felonies and that everyone
else was intended to be released. Once again, these nets were often only
based in theory, as practical application using money has eroded the nature
of these distinctions.
These early detention eligibility nets often contained the requirement of a
finding of “proof evident or presumption great” as to the charge, which
added an evidentiary component to when a person could be detained and to
make sure there was a way out of the net when the evidence was weak.
380
This point is important to reinforce: although persons might fall into a
detention eligibility net, they could still be released, meaning that there was
no automatic detention. Overall, these early models still presented the two
fundamental components of any detention provision today: (1) a detention
eligibility net, and (2) some further limiting process.
As noted previously, these theoretically pure models of release/detain
dichotomies have been complicated by American practice, which gradually

380 See Carbone, supra note 16, at 531-32.
140
began to allow the unintentional detention of bailable defendants through the
use of money. Throughout the history of bail in England and America, the
idea that a bailable defendant – a defendant who, today, would not be within
the detention eligibility net – might be detained was deemed to be so
backward and wrong that it typically led to bail reform. Today, our
understanding of a clear in-or-out system, articulated as bail (release) and no
bail (detention), is clouded by the fact that our practical administration of
bail is completely aberrant to historical notions. Today, we say that a
defendant is bailable and yet detain him. We order a defendant to be
released, and yet allow a condition of that release to keep him in jail.
Nevertheless, taking a step back, one sees that every state has already
articulated where it intends the line to be drawn between release and
detention. It just so happens that the states have drawn that line in
dramatically different variations, and that bail practice and the use of money,
in any event, have confused our understanding of the dichotomies. The
fundamental point is that states have already drawn theoretical lines between
release and detention, and so any changes to those dichotomies today means
that states are merely re-drawing those lines.
Wayne R, LaFave’s treatise on criminal procedure still provides the best
breakdown of the various state dichotomies, as represented in their right to
bail provisions (even though it is slightly out of date due only to very recent
activity in this area).381 Based on LaFave’s correct analysis of the issue,
states may be placed in one of the three following groups:
(1) states having no right to bail in their constitutions (nine states, akin to
the federal system operating under the United States Constitution);
(2) states having “broad” or “traditional” right to bail provisions (now
likely 19 states, modeled after the Virginia law of 1682);
(3) states with constitutional “right to bail” provisions that have been
amended since the 1980s to provide for additional preventive
detention that is typically (but not always) charge-based and often
premised on public safety (now likely 22 states).
States within the first group still typically have release/detain dichotomies in
their statutes; indeed, the lack of a constitutional right to bail provision
allows relative ease in creating vigorous preventive detention provisions in

381 See LaFave et al., supra note 52, § 12.3 (b), at 55.
141
the statutes or court rules. These statutory dichotomies, like those in other
states’ constitutions, can be explicit – for example, West Virginia states
expressly that, “A person arrested for an offense not punishable by life
imprisonment shall be admitted to bail by the court or magistrate. A person
arrested for an offense punishable by life imprisonment may, in the
discretion of the court that will have jurisdiction to try the offense, be
admitted to bail.”382 They can also be implicit – for example, North Carolina
articulates the right for virtually all noncapital defendants to have
“conditions of release determined.”
383 Still others require a much closer
examination, and can lead to an actual line that is quite different from any
theoretical line between release and detention.
States within the second group provide the most straightforward articulation
of a theoretical bail/no bail or release/detain dichotomy. For example,
Alabama’s Constitution provides that “all persons shall, before conviction,
be bailable by sufficient sureties, except for capital offenses, when the proof
is evident or the presumption great.”384 Other states have added to the net of
“capital offenses” certain categories of crimes (often called “categorical
exceptions” to the right to bail), such as crimes carrying a penalty of life
imprisonment,385 or individual crimes, such as treason.386
States within the third group are often called “preventive detention” states,
although any state potentially denying bail for either risk of flight or public
safety (including so-called broad right to bail states excepting only capital
defendants from the right to bail) can be said to have preventive detention.
As already noted, when America began discussing preventive detention, the
discussion surrounded danger only because it was commonly believed that
intentional detention of noncapital defendants due to risk of flight was first
prohibited, and then later gradually allowed through the courts’ inherent
power. Gradually, however, as the courts (and later the federal statutes)
began slowly to allow such detention, the distinction between flight and
danger has blurred. Today, the concept of preventive detention should not be
limited only to notions of detention for dangerousness, as both flight and
dangerousness are constitutionally valid purposes for limiting pretrial
release, up to and including detention.

382 W. Va. Code, § 62-1C-1 (2015).
383 See N.C. Gen. Stat. § 15A-533 (b) (2015).
384 Ala. Const. § 16.
385 See, e.g., Nev. Const. art. I § 7.
386 See, e.g., Or. Const. art. I, § 14.
142
As LaFave correctly notes, the states in this last group are likely best further
categorized in three ways: (1) states authorizing preventive detention for
certain charges, combined with the requirement of a finding of danger to the
community; (2) states authorizing preventive detention for certain charges,
combined with some condition precedent, such as the defendant also being
on probation or parole; and (3) states combining elements of the first two
categories.387
Most recently, New Jersey changed its constitutional provision from “broad
right to bail” language (all persons bailable except capital defendants, proof
evident presumption great) to preventive detention language allowing the
denial of pretrial release whenever the court determines that “no amount of
monetary bail, non-monetary conditions of pretrial release, or combination
of monetary bail and non-monetary conditions would reasonably assure the
person’s appearance in court when required, or protect the safety of any
other person or the community, or prevent the person from obstructing or
attempting to obstruct the criminal process.”388 This represents a massive
enlargement of the theoretical detention eligibility net in an attempt to redraw the line between release and detention based on “risk” and not charge.
As noted previously, the new statute in that state limits detention to “eligible
defendants,” but that definition includes defendants charged with any
indictable offense (akin to any felony in other states) or charged with any
disorderly persons offense (akin to most misdemeanors in other states.)389
Thus, the detention eligibility net is extremely broad.
New Mexico, too, recently passed a change to its constitutional right to bail
provision. New Mexico’s previous constitutional provision was in the form
of LaFave’s preventive detention state subgroup number two, which allowed
detention for persons charged with capital offenses or felonies with certain
preconditions, such as felonies after the conviction of two previous felonies.
The new language now includes “risk-based” language by allowing the
denial of bail for defendants charged with any felony if the prosecutor
“proves by clear and convincing evidence that no release conditions will
reasonably protect the safety of any other person or the community.”390
While not necessarily providing express authority to detain based on risk of

387 See LaFave, et al., supra note 52, at § 12.3(b), 56-59.
388 N.J. Const. Art I, par. 11.
389 See N.J. Stat. Ann. § C.2A:162-15 (2017).
390 Senate Joint Resolution 1, found at http://www.sos.state.nm.us/uploads/files/CA1-SJM1-2016.pdf.
143
flight,391 the new language does, like New Jersey, greatly enlarge the
detention eligibility net from only certain felonies with preconditions to all
felonies. The court rules, which might further limit this net, have not yet
been crafted at the time of the writing of this paper.
It is important to note that virtually every state constitutional provision
found in groups two and three, above, are likely vulnerable to constitutional
attack on various grounds, but mostly on grounds derived from the opinion
in United States v. Salerno.
392 LaFave, et al., point out the vulnerabilities
from lack of procedural safeguards,393 but the provisions are equally
vulnerable due to the apparent lack of justification for dramatically enlarging
the detention eligibility nets and the lack of decent limiting processes.
Indeed, at least two recent court cases have begun what will likely be a long
jurisprudential march toward determining the limits of preventive detention
in the states. The first, Lopez-Valenzuela v. Arpaio, which has been cited for
various points previously discussed in this paper, ruled that an Arizona
detention provision was not “carefully limited” as required by the Supreme
Court from a reading of Salerno.
394 The second, an Arizona Supreme Court
case, is also significant because the court looked at a “no bail” provision for
certain sex offenses that was added to that state’s list of so-called categorical
offenses – like capital offenses – that are potentially detainable if the court
finds that the proof is evident or the presumption great” as to the
commission of crime.395 Holding that provision up to Salerno, the Arizona
Supreme Court ruled the provision to be unconstitutional on its face because
it was not narrowly focused on accomplishing the government’s stated
objective.396 Theoretically, the Arizona court’s analysis in that case would
make vulnerable any charge-based detention provision that relies only upon
a finding of “proof evident, presumption great,” of which there are many

391 The constitutional language itself only allows detention based on danger. Although the provision later
says, “A person who is not detainable on grounds of dangerousness nor a flight risk in the absence of
bond,” this language does not provide express authority to detain for risk of flight. An introductory
paragraph to the ballot language includes a statement that indicates the state perhaps intended to also allow
detention based on risk of flight: “Proposing an amendment to Article 2, Section 13 of the Constitution of
New Mexico to protect public safety by granting courts new authority to deny release on bail pending trial
for dangerous defendants in felony cases while retaining the right to pretrial release for non-dangerous
defendants who do not pose a flight risk.” Id.
392 See Salerno, 481 U.S. 739 (1987).
393 See LaFave, et al., supra note 52, § 12.3 (b), at 61, 84.
394 See supra notes 226, 311, and accompanying text.
395 Simpson v Miller, 387 P. 3d 1270 (Ariz. 2017).
396 See id.
144
across America.397 At the very least, the opinion signals what will likely be a
new wave of court cases examining the substance and justification for how
states have currently drawn the line between pretrial release and detention.
Accordingly, when re-drawing the line between pretrial release and
detention, jurisdictions must remember how states have drawn those lines in
the past, but recognize that, in many cases, those lines might not stand up to
scrutiny under existing law. Jurisdictions are cautioned not to look to other
state laws as examples or models unless those examples adequately follow
the basic fundamental principles outlined in this paper; indeed, most
detention eligibility nets in current state laws have been gradually widened
based on false assumptions, fear of crime, and the fact that detention proves
its own worth. As of the date of this writing, a truly exceptional “model” bail
provision dealing with the line between release and detention has not been
enacted. Moreover, because the federal system has greatly expanded its own
detention eligibility net, has misused various rebuttable presumptions
leading toward detention, and has otherwise adopted practices leading to
over-detention, that law should only be used as a model in the sense that it
broadly requires a deliberate in-or-out process with minimal use of money.
Do We Have to Eliminate Money at Bail Before We Re-Draw
Our Line Between Pretrial Release and Detention?
There are many powerful arguments for eliminating money at bail, including
that money bail is ineffective and unfair. Nevertheless, jurisdictions do not
have to rid themselves of money bail in order to create a rational line
between release and detention; however, they must rid themselves of
money’s ability to detain. Historically, money’s ability to detain in the form
of secured financial conditions has interfered with every state’s initial
attempt to draw a meaningful and purposeful line between release and
detention.398 And despite some attempts to dissuade the use of certain

397 See id. The opinion’s analysis is somewhat strained compared to that of the state court of appeals;
indeed, the Arizona Supreme Court disagreed with both the state court of appeals as well as the Ninth
Circuit’s federal analysis of the federal constitutional claim, and implied that certain categorical no bail
provisions with much less procedural due process might withstand scrutiny if those provisions include
charges that pose “inherent risks” that can justify the denial of bail.
398 As noted previously, detaining someone using money on purpose is unlawful. Unintentional detention,
while allowable in America under some unfortunate 8th Amendment analysis, is currently under attack on
other grounds, including that it violates the Equal Protection Clause. This and other papers have also
articulated numerous other reasons for why using money might be irrational and unfair, and thus violate the
law. Nevertheless, even if the law does not eliminate money’s ability to detain, it does not remove the
negative consequences most likely to affect the overall goal of this paper, which is to determine which
145
blatantly unlawful practices, we have been unable to keep money from
causing systemic problems, including massive interference with who we feel
should be released and detained pretrial. Jurisdictions can choose to leave
money in the system – indeed, both the federal and D.C. pretrial systems
retain money while eliminating money’s ability to detain – but jurisdictions
should also realize that money might be taken from them. In this generation
of bail reform, many national organizations are crafting litigation strategies
designed to rid the country of money bail. Elimination of money bail will, in
turn, force all jurisdictions to make sure their lines between release and
detention are drawn correctly, and to change them if they are not. The
fundamental point is that money need not be eliminated from the pretrial
system, but money the way we have used it for over 100 years must be – and
likely will be – eliminated in order to create a transparent and workable
demarcation between those we seek to release and those we seek to detain
prior to trial.
Accordingly, when re-drawing the line between pretrial release and
detention, jurisdictions must remember that leaving secured money bonds in
the process will likely thwart any efforts to set up a fair and effective release
and detention system. Making sure that money does not detain is thus a
crucial prerequisite to creating that system.
Will We Need to Make Sure We Have Some Resources – Like
Pretrial Services Functions – To Make Everything Work?
In 1970, Congress created authority for the pretrial services agency in the
District of Columbia to supervise defendants in the community (in addition
to creating bail reports), which was seen as a necessary component of the
purposeful in-or-out process being enacted. Likewise, in 1984 Congress
considered pretrial services functions to be a critical component of the
overall change from a traditional money-based system to the federal in-orout system using virtually no money whatsoever. This recognition of the
need for at least some minimal supervisory resources is akin to the history of
probation in America, which once used financial conditions of release, but
which eventually replaced money with some community supervision as a
more effective and fair way to achieve the goals of probation. While it is

defendants to release and which to detain pretrial. In short, leaving money in the system will significantly
diminish, if not completely preclude, a jurisdiction’s attempts to decide whom to release and detain, and to
see those decisions purposefully effectuated.
146
occasionally argued that many jurisdictions currently over-supervise
defendants – indeed, the research would support using far less supervision
on low and medium risk defendants to achieve maximum outcomes – some
resources are likely necessary for states to move to a purposeful bail scheme
that desires to lawfully release and detain the appropriate people. Those
resources should be in the form of traditional pretrial services functions.
Will We Have To Change Our Constitutions?
States that have right to bail provisions in their constitutions do not
necessarily have to change them, but they might want to do so nonetheless.
If, in fact, America eliminates money bail (or even money’s ability to detain
at bail), or if a state either voluntarily changes or is forced to change certain
bail practices leading to more purposeful pretrial release and detention,
states that are happy with their current release/detain dichotomies can simply
leave their constitutions alone. For example, if a state currently has a broad
right to bail (reserving potential detention only for defendants charged with
“capital offenses, where the proof is evident or the presumption is great”),
and the state removes money’s ability to detain, the state merely has to ask
whether it is acceptable to release everyone except defendants facing capital
charges. If the state feels that a wider detention eligibility net is necessary –
and if that state can, in fact, justify a larger net – it might want to change its
constitution.
Justifying whatever new net a state hopes to create will be a crucial part of
this question. Justification is discussed at length below, but for now
jurisdictions should realize that the law requires proper justification for
detention provisions, and recent court cases are forcing states to examine
whether they have supplied sufficient justification. If, for example, a state
has a relatively narrow, charge-based detention eligibility net, and after
reading this paper, that state realizes that it simply cannot justify a wider net,
it will not have to change its constitution so long as the first net does have
some valid justification. If the first net also does not have adequate
justification, the state can leave the provision alone, but it risks having that
language struck later on constitutional grounds. The myriad variables
associated with this decision make the undertaking somewhat complex.
Moreover, even if money (or its ability to detain) is not eliminated, a state
may still want to create a system that dramatically reduces the use of money,
which will, in turn, similarly require the state to articulate whom it intends to
147
release and detain with more precision. For example, even if money’s ability
to detain is not eliminated, states may want to create a more rational process
for detention that does not rely upon unattainably high monetary conditions
of bond to detain certain unmanageable defendants. This, too, may lead to a
desire to change any particular constitutional provision. Nevertheless, it
bears repeating that even states that have attempted to create more rational
in-or-out processes leading to purposeful detention based on risk have seen
those processes ignored when money is left in the system. The fundamental
point is that states do not have to change their constitutional right to bail
provisions, but it may be sound practice to do so based on their desire to
properly articulate who is eligible for detention and how to effectuate that
decision.
The exact wording of any proposed change to a state constitution will likely
depend upon philosophical considerations. For example, if a state has a
broad right to bail provision and wishes to release virtually everyone
pretrial, no change to the constitution may be necessary. If, however, the
state wishes to detain certain high risk defendants, it may need to create that
authority within its constitution. Philosophically speaking, if that state
desires to create the authority to detain so-called “high risk” defendants
within a charge-based eligibility net and to release virtually everyone else,
but also wants to limit future legislative determinations that would gradually
erode the presumptive right to release, the state would need to add detail to
its constitution to forestall those determinations. If not so concerned, the
state can use broad language granting legislative authority to prescribe the
detention process, and it will be the two things together – constitutional bail
provision and implementing legislation – that will be reviewed for overall
legality. The same concept governs current constitutional provisions that
reserve detention for “violent” or “serious” crimes. States leery of legislative
erosion of rights have actually defined such phrases in the constitution
itself.
399
The risk research, too, will likely impact a state’s decision on particular
language. Actuarial pretrial risk assessment instruments do not necessarily
currently predict the type of risk we hope to address through detention, but
they might in the future. Moreover, someday risk research might definitively
point to a certain group of defendants whose conduct makes them extremely

399 See, e.g., Tex. Const. art. I, § 11a.
148
high risk for flight or new serious or violent crime. These details must be
considered when a state desires to change a document such as a constitution.
Accordingly, when re-drawing the line between pretrial release and
detention, jurisdictions must remember that they may wish to change their
constitutional bail provisions, but that they do not have to. The decision to
change will be based on individual state notions of liberty and freedom, but
should be made with the knowledge that bail reform appears to be forcing
jurisdictions to legally justify their constitutional provisions and may
ultimately remove money (or at least the ability of secured financial
conditions to detain) from the existing system.
Will We Have to Change Our Statutes/Court Rules?
Most states will find that they will have to make significant changes to their
statutes and court rules in this generation of bail reform. For example, the
reduction or elimination of money’s ability to detain will necessarily lead to
substantial changes in most state laws, which are often primarily designed
around a money-based system. As another example, the courts may begin
ruling that various state-articulated detention nets are unlawful under United
States v. Salerno,400
as was done in the Ninth Circuit case of LopezValenzuela v. Arpaio, in which the court ruled that the relevant detention
provision and lack of due process protections violated the federal
constitution.401 Rulings like this will undoubtedly also force states to
“engraft such protections into the applicable provisions in the state
constitutions, statutes and court rules to forestall” invalidation of preventive
detention schemes on federal constitutional grounds.402
Thus, when re-drawing the line between pretrial release and detention,
jurisdictions must remember that it will likely be necessary to change their
statutes and court rules to respond to the various elements underlying the
third generation of bail reform. Jurisdictions should also remember that it
will be the totality of their detention process – their constitutional bail
provisions (if they have one) along with their processes as articulated in their
statutes, rules, or even court opinions – that will be analyzed for
justification, rationality, and fairness.

400 See United States v. Salerno, 481 U.S. 739 (1987).
401 Lopez-Valenzuela v. Arpaio , 770 F. 3d 772, at 791-92 (2014).
402 LaFave et al., supra note 52, §12.3 (b), at 61.
149
Will We Have to Change Our New Bail
Guidelines/Praxes/Matrices?
In this generation of bail reform, various jurisdictions have begun creating
bail guidelines, praxes, or matrices to reflect new notions dealing with
defendant risk. Originally, these matrices were designed to replace
traditional money bail schedules, which are documents that assign money
amounts to various charges. They have since been seen as a valuable way to
engage various criminal justice stakeholders in discussions about risk
tolerance, release and detention philosophy, defendant supervision and
responses to pretrial violations. Rather than to assign money amounts to
charges, many of these new matrices often place risk assessment scores
along the vertical axis (left side) of a grid, and then various charges or
charge categories along the horizontal axis (top) of a grid. The boxes where
these two things intersect represent decisions about presumptive conditions,
supervision strategies, and risk tolerance. For example, the intersection
between “low” risk and a nonviolent misdemeanor charge on a particular
matrix might lead to a box with a presumptive release, presumptive
conditions, and even presumptive supervision strategies. More recent
matrices are using risk of failure to appear along one axis and risk of new
criminal activity along the other, along with criminal charge and the PSA
violence flag as other considerations to suggest various pretrial options.
150
The following is a risk/charge matrix, made up, but typical of several seen
across America:
Most Serious Charge
Pretrial
Risk
Category
Less Serious
Misdemeanor
More Serious
Misdemeanor
Non-Violent
Felony
Driving
Under the
Influence
Domestic
Violence
Violent
Felony
Lower
Recognizance
Release with
Court
Reminder
Recognizance
Release with
Court
Reminder
Recognizance
Release with
Court
Reminder
Recognizance
Release with
Basic
Supervision
Recognizance
Release with
Basic
Supervision
Recognizance
Release with
Enhanced
Supervision if
Released; or
Detained
Medium
Recognizance
Release with
Basic
Supervision
Recognizance
Release with
Basic
Supervision
Recognizance
Release with
Basic
Supervision
Recognizance
Release with
Enhanced
Supervision
Recognizance
Release with
Enhanced
Supervision
Recognizance
Release with
Enhanced
Supervision if
Released; or
Detained
Higher
Recognizance
Release with
Basic
Supervision
Recognizance
Release with
Enhanced
Supervision
Recognizance
Release with
Enhanced
Supervision
Recognizance
Release with
Enhanced
Supervision
Recognizance
Release with
Enhanced
Supervision
Recognizance
Release with
Enhanced
Supervision if
Released; or
Detained
151
The following is a risk/risk matrix recently created and used in one
American county:
Despite their value, the matrices being used today can be misleading to
jurisdictions. For example, sometimes a matrix will label a box
“presumptive detain,” even though it would be unlawful to detain under that
state’s current constitutional release/detain dichotomy. In other cases, the
matrices are no better than traditional bail schedules, as they include money
amounts in the boxes and are simply using risk versus charge to administer a
wealth-based bail system.
Accordingly, some of these matrices should already be changed to reflect the
actual law in their states, and still others should be changed to rely more on
evidence-based research. In any event, as jurisdictions begin to dig deeper
into their own laws concerning release and detention, and especially as those
states begin studying the pretrial research, they will undoubtedly find that
152
various aspects of these matrices must be changed. On the other hand, once
appropriate changes are made to a state’s legal structure, the creation and
operation of guidelines, matrices, and praxes can operate neatly within that
structure.
153
Part II – If We Change, To What Do We Change?
This generation of bail reform appears to be leading states to change from
the traditional charge-and-money-based system to something new. So far,
that new thing has been labeled a “risk-based” or “risk-informed” way of
doing bail, and involves assessing all defendants for their risk using actuarial
pretrial risk assessment instruments, trying to detain only so-called “high
risk” defendants, and using the law and the research to release everyone else
on varying levels of supervision. While superficially simple, this paper
illustrates just how complex such an undertaking can be.
In fact, it is the risk research itself that triggers our need to slow down and
systematically justify everything we intend to do with pretrial release and
detention. Fundamentally, the risk research dismantles many of our existing
assumptions underlying the charge-and-money-based system, and yet that
same research demonstrates that “risk” as measured by an actuarial pretrial
risk assessment instrument cannot wholly replace that system. Meanwhile,
courts are beginning to require jurisdictions to show rationality and nonarbitrariness at bail. This means that all jurisdictions will likely have to start
from scratch by articulating and adequately defining whom they intend to
release and whom they intend to detain pretrial. Then, using the pretrial
research to date, those jurisdictions can create rational, fair, and transparent
release/detain dichotomies that can survive judicial scrutiny for as long as
possible.
This entire paper has been leading to an answer to the question, “If we
change, to what do we change?” As evidenced by the length of the
discussion so far, the answer depends on knowing a variety of things about
bail. And those things, from the proper definition of bail to base rates and
false positives, naturally lead to a model pretrial release/detain dichotomy
designed to answer the underlying questions of, “whom do we release,
whom do we detain, and how do we do it?” But any model of line drawing
must be justified, and so this author proposes holding up whatever model a
state might create to three separate but overlapping analyses to help with that
justification. At the end of this paper, the reader will see this author’s model
release/detain dichotomy and process, which is then held up to these same
three analyses. It is advised that any state desiring to come up with its own
model – for example, one with a wider detention eligibility net or a slightly
different limiting process – hold that model up to the same three analyses so
154
that it can survive judicial scrutiny. It is not necessary to analyze any
particular model in the order presented, and indeed the overall analysis will
likely be a combination of all three together. The three are placed in this
particular order solely due to personal preference.
The first analysis is a somewhat general analysis based on the history, the
law, the research, and the national standards that requires us constantly to
consider narrowing detention to further fundamental American principles.
Thus, even when a detention scheme might pass muster under so-called
strict scrutiny analysis in the law, we must still consider whether there are
other factors that warrant further narrowing detention, thus embracing risk
and erring on the side of release.
The second analysis is a purely legal analysis, which can be achieved
primarily by holding up the detention scheme to United States v. Salerno.
This involves making sure the scheme survives not only a somewhat more
lenient analysis to determine whether it would be deemed punishment by the
courts, but also the “heightened” analysis required under general due process
principles in addition to concerns potentially leading to equal protection and
excessive bail claims.
The third analysis is based upon Andrew von Hirsch’s articulation of three
threshold requirements for any preventive detention scheme, which includes:
(1) the need for precise legal standards of dangerousness; (2) the need to
subject prediction methods to careful and continuous validation; and (3) the
need for certain minimal procedural safeguards.403 While there is some
overlap between this third analysis and the others, its importance lies
primarily in the discussion concerning precise standards and definitions, a
concept that has been lacking in American bail law for both dangerousness
and flight. Each of these analyses is discussed briefly below.
Analysis Based on General Narrowing Principles Gleaned from the
History, the Law, the Research, and the National Standards
Throughout this paper, this author has summarized what jurisdictions must
remember when re-drawing the line between release and detention. They
include the need to remember: that historically and legally speaking, bail is
release, and that the right to bail is technically the right to release; that the

403 See von Hirsch, supra note 281, at 725.
155
reason we have bail, or pretrial release, in America is due to the law and
certain fundamental legal traditions, such as using the moral deterrence of
the law to guide our actions and acknowledging the presumption of
innocence throughout the criminal process; that jurisdictions must not be
risk averse; that they must instead embrace the risk of release at bail, and
therefore accept some level of pretrial failure, just as we have “failure” to
the extent that people might not generally follow the law in a free society.
They include the need to remember: that the history of bail illustrates that
any interference with “bail” as release or “no bail” as detention leads to bail
reform; that because secured money bonds have been interfering with both
release and detention since the mid-1800s, dealing with secured money at
bail is likely a prerequisite to complete reform; that the law points to clearly
identifying the threats that we hope to address, to limit detention to a
justifiable eligibility net with a process designed to further limit detention of
persons within the net to those with identifiable and articulable risks of
either flight or serious danger; that jurisdictions must create detention
schemes that work within the fundamental balance of bail, which involves
maximizing release while maximizing public safety and court appearance;
that jurisdictions must remain mindful of fundamental legal principles that
have, until recently, been largely ignored at bail; that while it may be
admirable to aim for an appropriate release to detention ratio, such a ratio
will perhaps more appropriately evolve out of a fair and rational release and
detention process.
They include the need to remember: that American differences with English
bail were due to our fundamental notions of freedom and liberty, which led
to broad rights to bail (or release) in virtually every jurisdiction; that
America’s struggle with both unintentional and intentional detention was
exacerbated by fundamental flaws in our release and detention systems,
including allowing secured financial conditions of bonds to cause detention;
that when America began purposefully detaining noncapital defendants for
either flight or public safety, it was done only in “extreme and unusual
circumstances” shown by unique facts surrounding individual defendants
and potential victims; that America’s “big fix” involved a rational and
purposeful in-or-out process, with a broad right to release and a narrow
detention net, and that also eliminated money’s ability to detain; that this fix,
while admirable in theory, has grown unacceptably broad in the federal
system and has not been adequately adopted by the states; and that after
Salerno, most state detention provisions are constitutionally vulnerable.
156
They include the need to remember: that the research surrounding the
creation of actuarial pretrial risk assessment instruments tends to undermine
many of our old assumptions about risk, such as the assumption that high
charge equals high risk (indeed, a defendant charged with an extremely
serious crime might be “low” risk just as a defendant charged with a nonserious crime, like trespass, might be “high risk”); that the research on risk,
in fact, tells us that most defendants are simply not as risky as we think, that
vastly more defendants succeed pretrial than fail, and that, accordingly, there
are many more people in jail than necessary; that the research on risk also
points to different outcomes than we are used to when analyzing bail claims
based on due process, equal protection, and excessive bail; and, indeed, that
the research on risk likely makes it hard to justify anything but the narrowest
detention eligibility net.
They include the need to remember: that actuarial pretrial risk assessment
instruments are exceptionally good at what they tell us, and can be used for
99% of everything we care about in bail, but they do not tell us individual
risk, what to do with risk, and details concerning “risk of what” to the extent
necessary to justify pretrial detention by themselves; most importantly, that
those instruments label all defendants as risky, which means that we must
consider charge-based floors below which no detention – and possibly below
which no assessment – may occur; finally, that those instruments provide no
adequate definitions or explanations of the type of conduct necessary to
avoid pretrial detention.
Finally, they include the need to remember the lessons learned from the
evolution of the national standards on pretrial release and detention, which
provide a good framework for an in-or-out process, unhindered by secured
money conditions. Specifically, jurisdictions must remember that the ABA
Standards attempt to limit detention in a variety of ways, recommend a
charge-based detention eligibility net, and provide at least some justification
for recommending release of higher risk persons facing relatively minor
charges.
Each of these notions, and certainly all of them together, point to pretrial
release being very broad – likely broader than we are used to – and to
pretrial detention being extremely narrow – likely much narrower than the
minimum required by law. Accordingly, these notions suggest that any
detention provision should be assessed for whether it can be further
157
narrowed, beyond what is the minimum necessary to survive legal claims.
Thus, for example, if a state declares its detention eligibility net to consist of
persons charged with “all violent felonies,” the state should nonetheless ask
whether a narrower net might not work as well and might be better
supported by the history, law, and research. At the very least, jurisdictions
exploring changes should add narrower options into their lists for discussion
purposes when re-drawing the line between pretrial release and detention.
Analysis Based on the Current Law
As noted previously, jurisdictions re-drawing the line between release and
detention will need that line to survive legal scrutiny, especially under the
Equal Protection Clause, the Excessive Bail Clause, and the Due Process
Clause.404 The legal analyses under these three clauses are somewhat
overlapping, and thus certain government decisions might be dispositive of
elements under each theory. Nevertheless, jurisdictions should remember the
following overarching idea governing the need to justify any release or
detention model: “A state may constitutionally provide that bail be granted
in some cases as a matter of right and denied in others, provided that the
power is exercised rationally, reasonably, and without discrimination.”405
Reasonableness, in turn, will likely often be shown by the pretrial research.
For example, if the pretrial research shows that there is no link between a
certain charge and higher risk, it would not be reasonable to single that
charge out for potential detention without some further justification. Once
again, jurisdictions should not be so concerned with articulating a ratio of
released to detained defendants up front. If any proposed release and
detention model can be adequately justified, then the ratio will determine
itself.
Equal protection, excessive bail, and due process would all require the
government to adequately justify bail laws, and yet this justification has
been sorely lacking in previous decades. This is exacerbated by the fact that
many of our previous justifications for release and detention simply do not
hold up today. Where once we could create a detention eligibility net based
on the idea that defendant facing a felony was more risky than one facing a
misdemeanor, now we must face research showing that this assumption is

404 This paper focuses primarily on federal law. Any legal analysis would require scrutiny under state law
as well, which would include analysis under any right to bail provision.
405 United States ex rel Covington v. Coparo, 297 F. Supp. 203, 206 (S.D.N.Y. 1969) (quoted in Hunt v
Roth, 648 F.2d 1148, at 1161 (8th Cir. 1981), judgment vacated for mootness, 455 U.S. 478, (1982).
158
not necessarily true. Accordingly, new justifications may be necessary today
to survive scrutiny under these three clauses. And thus, in this generation of
bail reform, it is imperative that we continually force jurisdictions to
demonstrate why any particular person or groups of persons should be
detained pretrial. And if detention cannot be adequately justified for certain
defendants, as a society we must be ready for the inevitable conclusion that
those defendants must therefore be released. This may lead to somewhat of a
culture shock by moving from a society that is anathema to risk to one that
understands the need to embrace risk and the associated failures inherent in
bail. Any other conclusion, however, would be contrary to our fundamental
American principles of limited government and personal freedom.
In a comprehensive article on preventive detention written for the Harvard
National Security Journal, authors Adam Klein and Benjamin Wittes
analyzed this country’s history and practices surrounding preventive
detention and concluded that it is false to believe that the notion of
preventive detention is repugnant to America. Instead, the authors
concluded, “Congress and state legislatures create preventive detention
authorities without apology where they deem them necessary, and the courts
uphold them where judges find that the statutes, or their application, allow
only so much detention as is actually necessary to address a pressing public
danger.”406 After surveying the various types of preventive detention in
America, the authors correctly note that “the unifying theme is that the law
unsentimentally permits preventive detention where necessary but insists
upon adequate means . . . of insuring both the accuracy of individual
detention judgments and the necessity of those detentions.”407 In the law,
this sort of balancing of means and ends goes to the overall rationale of the
detention scheme, which requires states to provide justifications for who
should be detained and how that detention is done.
Narrowing (based on the first analysis) and justification (a prerequisite of the
second analysis) come together in the law through levels of scrutiny that
courts impose upon government action, and jurisdictions must keep these
levels in mind when constructing any new detention model. Generally
speaking, courts tend to use balancing tests for due process, equal protection,
and excessive bail (the three most likely theories to be used to assess release
and detention models), in which the courts weigh the means and ends behind

406 Klein & Wittes, supra note 230, at 186.
407 Id. at 187.
159
the laws. Balancing tests, in turn, fall into categories based on the levels of
scrutiny assigned to certain disputes. When the law being reviewed involves
a suspect classification or fundamental right such as liberty, courts typically
apply so-called “strict scrutiny,” which requires the government to show that
the law is necessary to achieve a compelling or overriding government
purpose. “Intermediate scrutiny,” which is typically used when a
classification is made along gender or legitimacy lines, requires the
government to show that the law is substantially related to an important
government purpose. “Minimum scrutiny” (often called “rational basis”) is
used whenever the other two levels are not triggered, and it requires the
government to show only that the law is rationally related to a legitimate
government interest. The first and third levels are particularly important.
Strict scrutiny is likely necessary for any provision affecting the
fundamental interest of liberty. Nevertheless, many bail provisions, policies,
and practices also lack the kind of rational justification to survive even
minimum scrutiny.
Excessive Bail
Any new or existing line drawn between release and detention will have to
survive excessive bail analysis. As noted previously, the current excessive
bail test is one of balance, as articulated by the United States Supreme Court
in Salerno:
The only arguable substantive limitation of the [Excessive] Bail
Clause is that the Government’s proposed conditions of release
or detention not be ‘excessive’ in light of the perceived evil. Of
course, to determine whether the Government’s response is
excessive, we must compare that response against the interest
the government seeks to protect by means of that response.408
Explaining this language, the Eleventh Circuit Court of Appeals noted that
“the test for excessiveness is whether the terms of release [or detention] are
designed to ensure a compelling interest of the government, and no more,”
thus implicating strict scrutiny.409 While other courts have merely re-stated
the Salerno test as one in which an appellate court should review conditions
of release to determine whether they were “excessive in light of the purpose

408 United States v. Salerno, 481 U.S. 739, 754 (1987).
409 Campbell v. Johnson, 586 F.3d 835, 842 (11th Cir. 2009).
160
for which it is set,”410 these restatements do not provide the guidance
necessary for deciding “excessiveness.” A test based on levels of scrutiny,
however, would provide that guidance. For example, the government might
argue that a condition was non-excessive, but if that condition was not
rationally related to a legitimate purpose, it would fail under even the most
lenient balancing test. Because Salerno itself described the defendant’s
liberty interest as “fundamental” and the government’s interest in preventing
crime as “compelling in that case,” the fact that the Court now uses a
balancing test for excessive bail likely means that the method for assessing
that balance should be one akin to strict scrutiny, as all conditions impinge
to some extent upon a defendant’s liberty interest.
411
Excessive bail analysis has been somewhat derailed in America, due largely
to an unfortunate line of cases declaring that persons do not necessarily have
a right to “bail” that they can afford.412 Nevertheless, this line of cases does
not mean that excessive bail analysis cannot be applied to a state’s
determination of where to re-draw its line between release and detention. In
addition to requiring adequate justification through articulating a compelling
interest, the test requires assessing the means of achieving that interest. This
provides a brake, of sorts, on states desiring to use pretrial detention as the
blunt instrument for all crime control. As noted by LaFave, et al., “[T]here
exists in Salerno at least the suggestion that under the Eighth Amendment
the risk of future crimes by certain types of arrestees could be so
insubstantial as to make preventive detention of such persons excessive.”413
Using excessive bail analysis in this way has obvious implications for a
purely risk-based system (states would have a difficult time justifying
detention for “low” or “medium” risk defendants when they have relatively
high statistical probabilities of success), but it should also affect a state’s
decision to create a more appropriate charge-based detention eligibility net.
Specifically, in addition to adequate justification, excessive bail analysis
should put limits on detaining certain defendants facing relatively low level
charges who are nonetheless deemed “high risk.” In short, some offenses –
perhaps misdemeanors or non-violent property offenses – are simply not
serious enough to trigger the blunt hammer of detention no matter how risky

410 See e.g., Galen v. County of Los Angeles, 477 F.3d 652, 661 (2007).
411 See Salerno, 481 U.S. at 749, 750 (“The government’s interest in preventing crime by arrestees is both
legitimate and compelling;” “On the other side of the scale, of course, is the individual’s strong interest in
liberty. We do not minimize the importance and fundamental nature of this right.”).
412 See NIC Money, supra note 30, at notes 73-82 and accompanying text.
413 LaFave et al., supra note 52, § 12.3(c), at 71-72.
161
some defendants may seem to be based on aggregate prediction. This is the
essence of excessive bail analysis.
For example, even if a state could provide a justification for including
defendants charged with trespass in its detention eligibility net, detaining
trespass defendants might be successfully challenged on Eighth Amendment
grounds because detention based on prediction is simply an overwhelming
and likely unnecessary response to such a minor crime. This is true
especially given our lack of knowledge about individual risk and details
concerning “risk of what,” combined with our ability to address virtually all
levels of risk by using less restrictive release conditions as well as bond
revocation for unmanageable defendants. As a society, we would likely
never condone detaining stop light violators, and so the question is simply
one of line drawing by creating a floor, below which we feel no detention
should be sought. In other words, at some point certain charges simply do
not warrant initial detention, no matter what the risk. This line drawing, in
turn, is shaped by excessive bail analysis.
Equal Protection
Equal protection, too, is relevant to the release/detain discussion even
though, like excessive bail, it is less likely than due process to provide an
overall guide to re-drawing lines. In Schilb v. Kuebel, the Supreme Court
wrote that, “[A] statutory classification based upon suspect criteria or
affecting ‘fundamental rights’ will encounter equal protection difficulties
unless justified by a compelling governmental interest.”414 Because liberty is
a fundamental right, traditional equal protection analysis will, once again,
require the government to show that its new law does not treat similar
persons dissimilarly and is necessary to achieve a compelling or overriding
government purpose.
Very recently, civil rights organizations have begun suing cities and counties
in federal court on the theory that local bail laws are treating similar persons
dissimilarly based on their wealth (which can be tied to race). These equal
protection suits are relatively novel – while highly relevant, they are
extremely rare in historic bail jurisprudence – but they are a reminder that
bail laws must be justified along equal protection lines.

414 Schilb v. Kuebel, 404 U.S. 357, 365 (1971).
162
And, indeed, historic (and broader) criminal justice examples illustrate how
a government detention scheme might violate the clause in either a chargebased or a risk-based system. For example, prior to the Fair Sentencing Act
of 2010, crack cocaine sentencing laws and policies had long been criticized
as violating equal protection by disproportionately impacting racial
minorities (who were perceived as using crack more than powder
cocaine).
415 Similarly, a detention eligibility net containing arrests for
possession of crack versus powder cocaine could lead to equal protection
claims based on the same underlying arguments and assumptions. Likewise,
under a risk-based or risk-informed system, jurisdictions must ensure that
detention provisions calling for either detention or increased supervision for
“dangerous” defendants are not based on instruments capable of racial bias
or provisions that would likely be subject to claims under the Equal
Protection Clause.
Due Process
By far, however, the most relevant legal analysis for re-drawing the line
between release and detention is due process analysis flowing from the U.S.
Supreme Court’s opinion in United States v. Salerno, in which the Court
reviewed the Bail Reform Act of 1984.
416 Pursuant to that analysis, there
would likely be two balancing tests: one designed to ensure the provision is
not punishment, and one designed to assess the provision under general due
process principles. Scrutiny under the first test – which combines a “rational
basis” element with an excessiveness element – appears far less exacting
than scrutiny under the second, but it still contains relevant criteria for
judging any release and detention model. And while the Court in Salerno did
not expressly label its analysis under the second test “strict scrutiny,” at least
one federal court of appeals has correctly concluded that Salerno’s due
process test for detention is one of “heightened” scrutiny due to its focus on
liberty as a fundamental right.417 As the Ninth Circuit noted in that case, “If
there was any doubt about the level of scrutiny applied in Salerno, it has
been resolved in subsequent Supreme Court decisions, which have
confirmed that Salerno involved a fundamental liberty interest and applied
heightened scrutiny.”418 Basic primers on bail reform now confidently and

415 See Paul Larkin, Jr., Crack Cocaine, Congressional Inaction, and Equal Protection, 37 Harv. J. of L. &
Pub. Pol’y, 241 (2013), found at http://www.harvard-jlpp.com/wpcontent/uploads/2014/01/37_1_241_Larkin.pdf.
416 See United States v. Salerno, 481 U.S. 789 (1987).
417 Lopez-Valenzuela v. Arpaio, 770 F.3d 772, 779 (2014).
418 Id. at 780 (citations omitted).
163
correctly state that,“[A]s a threshold requirement, any system providing for
pretrial detention must be narrowly tailored to the compelling government
interest put forward to justify detention.”419 We will look at each of these
two tests – punishment and general substantive due process – briefly.
Test for Punishment
To determine whether a detention provision is impermissible punishment
before trial, the Salerno Court used the two-part test articulated in Bell v.
Wolfish, decided in 1979.420 Under the first part, a reviewing court first looks
for government intent to punish. Finding none, under the second part the
reviewing court looks merely to “whether an alternative purpose to which
[the restriction] may rationally be connected is assignable for it, and whether
it appears excessive in relation to the alternative purpose assigned [to it].”421
Overall, the test is similar to a “rational basis” balancing test under due
process, combined with language similar to excessive bail analysis, above.
The Court in Salerno found no evidence of intent to punish under the Bail
Reform Act, but such intent is not beyond the realm of possibility. Indeed, in
Lopez-Valenzuela v. Arpaio, discussed previously, the Ninth Circuit
reviewed an Arizona “no bail” provision and noted its concern over “the
considerable evidence of punitive intent found in this record.”422 In his
concurrence in that case, Judge Nguyen wrote separately “to address the
extraordinary record of legislative intent, which I believe demonstrates that
[the detention provision] was intentionally drafted to punish . . .”423
Obviously, intent to punish some group of defendants regardless of their risk
for flight or public safety will invalidate any detention provision.
Instead, the Court in Salerno found that Congress had authorized detention
for “the legitimate regulatory goal” of protecting the community from
danger.424 Under the balancing test, the Court found that the incidents of
pretrial detention were not excessive to the articulated regulatory goal

419 Harvard Law School Primer, supra note 3, at 8.
420 441 U.S. 520, 535 (1979).
421 Salerno, 481 U.S. at 747 (quoting Kennedy v. Mendoza Martinez, 372 U.S. 144, 168-79 (1963)).
422 770 F.3d 772, 79 and n. 14.
423 Id. at 792.
424 481 U.S. at 747. Pursuant to the test in Bell, “if a particular condition or restriction of pretrial
detention is reasonably related to a legitimate governmental objective, it does not, without more, amount to
‘punishment.’ Conversely, if a restriction or condition is not reasonably related to a legitimate goal – if it is
arbitrary or purposeless – a court permissibly may infer that the purpose of the governmental action is
punishment that may not constitutionally be inflicted upon detainees qua detainees.” Bell v. Wolfish, 441
164
“because: (1) the Act ‘carefully limits the circumstances under which
detention may be sought to the most serious of crimes,’ including ‘crimes of
violence, offenses for which the sentence is life imprisonment or death,
serious drug offenses, or certain repeat offenders’; (2) ‘[t]he arrestee is
entitled to a prompt detention hearing’ at which the arrestee could seek bail;
and (3) ‘the maximum length of pretrial detention is limited by the stringent
time limitations of the Speedy Trial Act.’”425
Accordingly, the analysis under this test points to limiting through the use of
a charge-based detention eligibility net as well as to generally assuring that
the means of achieving public safety or court appearance are not excessive,
which obviously overlaps somewhat with excessive bail analysis, above.
Nevertheless, the rationality component also implicates even more basic
notions of due process as expressed by the Court’s opinion in in Jackson v.
Indiana, in which the Court wrote, “At the least, due process requires that
the nature and duration of commitment bear some reasonable relation to the
purpose for which the person is committed.”426 Jackson was a case dealing
with pretrial commitment of incompetent defendants, but the Court’s
“reasonable relation” requirement is highly relevant to ordinary pretrial
detention cases, and has been argued by noted law professor Christopher
Slobogin to mean that, “If a liberty deprivation pursuant to a prediction fails
to adhere to the logic of preventive detention . . . then it can become
punishment.”427 Previously, arguments against relying on Jackson in
detention scenarios were based on an assumption that “it is rational for a
legislative body to conclude that those charged with a particular type of
offense are likely to repeat their crimes and thus to authorize preventive
detention as to persons so charged.”428 Today, however, the research on risk
may make even that assumption potentially illogical and thus unreasonable.
To Slobogin, the general limitation articulated in Jackson “suggests three
specific restrictions on preventive detention,” including that: (1) its duration
must be reasonably related to the harm predicted, (2) its nature must bear a
reasonable relation to the harm feared (which, in turn, requires the
government to pursue the least restrictive means of achieving its goals), and

U.S. 520, 539 (internal footnote omitted). Creating a detention provision – a process that necessarily
involves articulating a proper purpose – is less likely to be arbitrary than imposing individual release
conditions, such as money, which often seem to have little legitimate purpose.
425 Lopez-Valenzuela , 770 F.3d 772, 779 (2014) (quoting Salerno, 381 U.S. at 748).
426 Jackson v Indiana, 406 U.S. 715, 738 (1972).
427 Slobogin Dangerousness, supra note 321, at 13.
428 See LaFave, et al., supra note 52, at 75.
165
(3) it is periodically reviewed to assure the need for continued
confinement.429 Slobogin’s second restriction most directly concerns
jurisdictions’ ability to lawfully re-draw their lines between release and
detention, and likely means that, in addition to a need for adequate
justification of a detention eligibility net, jurisdictions must further ensure
that alternatives to incarceration are considered to purposefully further
narrow that net to some smaller set of detained persons who represent
unmanageable defendants who present the risk they seek to address. In short,
it means that jurisdictions must constantly concern themselves with
reducing, if not eliminating the false positives, for when “the paucity of . . .
alternatives results in incarceration of those who don’t need to be confined,
the detention becomes punishment.”430
Test for General Substantive Due Process
While the test for punishment includes some substantive hurdles, it is far less
exacting than the test following general due process principles as articulated
in Salerno. As correctly summarized by the court in Lopez-Valenzuela, the
1984 Bail Reform Act’s detention provision survived heightened, or strict
scrutiny,
because it both served a ‘compelling’ and
‘overwhelming’ governmental interest ‘in preventing
crime by arrestees’ and was ‘carefully limited’ to achieve
that purpose. The Act was sufficiently tailored because it
‘careful[ly] delineat[ed] . . . the circumstances under
which detention will be permitted.’ It: (1) ‘narrowly
focuse[d] on a particularly acute problem in which the
Government interests are overwhelming,’ (2) ‘operate[d] only on individuals who have been arrested for a specific
category of extremely serious offenses’ – individuals that
‘Congress specifically found’ were ‘far more likely to be
responsible for dangerous acts in the community after
arrest,’; and (3) afforded arrestees ‘a full-blown
adversary hearing’ at which the government was required
to ‘convince a neutral decisionmaker by clear and
convincing evidence that no conditions of release can

429 Slobogin Dangerousness, supra note 321, at 14-16.
430 Id. at 14-16.
166
reasonably assure the safety of the community or any
person.’ It satisfied heightened scrutiny because it was a
‘carefully limited exception,’ not a ‘scattershot attempt’
at preventing crime by arrestees.431
These are the most important elements of a lawful detention provision to
survive substantive due process analysis, including justification for the
provision (an “acute problem” reflecting a compelling government interest,
which is required under any balancing test but arguably requiring a stronger
government showing here due to the heightened scrutiny), a charge-based
detention eligibility net (required under both this test as well as the test for
punishment), and a process designed to further limit that net to individuals
demonstrably unmanageable in the community. Importantly, while not
argued in Salerno, detention due to risk of flight should be reviewed under
the same analysis.
The need for a charge-based net is clear from current law, and the prior
analysis would suggest that it remain charge-based – no matter how good
our risk prediction becomes – due to other legal principles (such as
excessive bail and due process fair notice) as well as certain unavoidable
limitations surrounding prediction in general.
In addition to establishing any other justification for any particular detention
provision, showing an “acute problem” of pretrial crime or flight today,
while likely made more difficult by current research, is not impossible.
When the District of Columbia Court of Appeals ruled on the
constitutionality of the District of Columbia Court Reform and Criminal
Procedure Act of 1970, it summed up Congress’s showing of a need for
pretrial detention based on dangerousness:
Congress considered (1) the alarming increase in street crime in
the District of Columbia since 1966; (2) statistical studies
involving recidivism by persons while on pretrial release; (3)
recommendations by the President’s Commission on Crime in
the District of Columbia (1966), and the Judicial Council
Committee to Study the Operation of the Bail Reform Act in

431 Lopez-Valenzuela, 770 F.3d at 772, at 779-80 (quoting Salerno, 481 U.S. at 748-755, internal citations
omitted).
167
the District of Columbia (1969); and (4) pretrial release and
detention practices in England and other countries.432
Reading the congressional report accompanying the 1970 Act, however, one
can quickly see two things: (1) there was strong support to address rising
crime in the District in multiple ways, including the use of pretrial detention,
and yet (2) there was actually very little evidence to support Congress’s
declaration that a “significant percentage” of violent crime was caused by
persons on pretrial release.433 Indeed, in that report, Congress twice pointed
out the lack of precise statistics or other data on pretrial crime, relying
instead on relatively hyperbolic narrative and a handpicked list of ten case
histories, with each illustrating a defendant committing an additional crime
while on bail.434 Nevertheless, this justification was enough for the D.C.
Court of Appeals even in the face of conflicting evidence presented during
the appellate process. The court wrote: “Appellant attempts to litigate what
are essentially legislative findings, i.e., the extent of crime committed by
persons released pending trial and the predictability of criminal conduct,
citing studies which reached different statistical results than those relied
upon by Congress. These are matters properly committed to the legislative
process.”
435
Similarly, in United States v. Salerno, the Supreme Court noted that in
passing the Bail Reform Act of 1984, Congress perceived pretrial detention
as a potential solution to a pressing problem (“the alarming problem of
crimes committed by persons on release”) and based that solution on
findings that persons arrested for certain serious crimes were “far more
likely to be responsible for dangerous acts in the community after arrest.”436
Compared to the 1970 D.C. Act, the main committee report accompanying
the 1984 Act had a more robust body of empirical evidence in addition to
general governmental or organizational support, including a study of release
practices in eight jurisdictions as well as a similar study done in the District
of Columbia.
437 As noted previously, however, no empirical evidence was

432 United States v. Edwards, 430 A.2d 1321, 1341 (D.C. Ct. App. 1981) (citing H.R.Rep.No.91-907, 91st
Cong., 2d Sess. 87-94 (1970)).
433 See H. Rep. 91-907, at 82.
434 See id., at 79-104.
435 430 A.2d at 1341.
436 Salerno, 481 U.S. at 742, 750.
437 See S. Rep. 98-225 at 6 (citing Lazar Institute, Pretrial Release: An Evaluation of Defendant Outcomes
and Program Impact, 48 (Wash. D.C., Aug. 1981); Institute for Law and Social Research, Pretrial Release
and Misconduct in the District of Columbia, 41 (April, 1980)).
168
given for detention based on flight, as Congress appeared to consider that
authority to be inherent or implicit.
In this generation of bail reform, we have more empirical research than ever
before on pretrial misconduct, to the point where there is virtually no excuse
for using it to help justify release and detention provisions.438 Nevertheless,
during the legislative process (or any other process hoping to craft release
and detention provisions based on the research), drafters will be faced with
the issues raised previously in this paper – including issues of true versus
perceived defendant risk and with certain limitations of actuarial pretrial risk
assessment instruments – all of which likely point toward a much narrower
charge-based detention eligibility net and a more robust limiting process.
Overall, good research will lead to good legislative findings, which, in turn,
will lead to good laws.
Analysis Based on Threshold Requirements for Predictive Models
In addition to analyses based on general narrowing principles and the law to
shape and justify any particular element of a detention model, the model
itself should also satisfy certain pre-requisites to find legitimacy within the
larger sphere of criminal justice. In 1971, the well-known legal philosopher
and theorist Andrew von Hirsch published an article based upon a staff
paper written for the Committee for the Study of Incarceration, made up of
eminent policy makers and professors of law, criminology and criminal
justice, sociology, history, psychiatry, and economics.
439 In that paper, von
Hirsch took on broad philosophical questions concerning the appropriateness
of a model of detention based on prediction of dangerousness, discussing
many of the issues raised in this paper. While tempting to say that his
analysis has been largely ignored as shown by America’s increased use of
preventive detention despite its seemingly obvious conflict with American
laws and values,440 through the decades scholars have correctly relied on von
Hirsch’s articulation of the prerequisites for any preventive model whenever

438 In Lopez-Valenzuela v. Arpaio, the Ninth Circuit Court of Appeals chided the Arizona Legislature,
which created a detention provision with “no findings, studies, statistics, or other evidence (whether or not
part of the legislative record) showing” the problem that it sought to address. 770 F.3d 772, 783.
439 See von Hirsch, supra note 281.
440 See generally Klein & Wittes, supra note 230.
169
those scholars were required, as now, to question fundamental aspects of
pretrial release and detention.
441

Andrew von Hirsch wrote:
To have any possible merit, the model should satisfy three
important threshold requirements: (1) there must be reasonably
precise legal standards of dangerousness; (2) the prediction
methods used must be subjected to careful and continuous
validation; and (3) the procedure for [preventive] commitment
must provide the defendant with certain minimal procedural
safeguards. These requirements, however, are seldom met by
current practices of preventive confinement.442
It is precisely because these requirements continue to be seldom met that
they are included in this paper’s overall justification of a detention model
using three separate analyses.
Precise Definitions
As von Hirsch and others have correctly noted, a person should not be
detained preventively unless that person has a risk of sufficient likelihood
and gravity to warrant detention, and articulating how those two notions
should or should not lead to detention – that is, in the field of bail, how risky
a defendant needs to be and what, exactly, he or she is risky for – is a value
judgment reserved for the law. Moreover, unless the law defines these two
notions with precision, “the entire preventive model may well be
unconstitutional on grounds of vagueness.”443 In the past, scholars have
noted America’s utter failure to adequately define terms such as danger or
public safety. “Even when ‘danger’ or ‘public safety’ concerns are explicit,
most states fail to provide operational standards or definitions for these
constructs.”444 This has been a fundamental flaw with American detention
provisions crafted throughout the twentieth century,445 and so any
jurisdictions creating new detention provisions should seek to remedy it by

441 See, e.g., Goldkamp, supra note 3, at 16; Fagan & Guggenheim, supra note 3, at 419-20; Shima
Baradaran & Frank L. McIntyre, Predicting Violence, 90 Tex. L. Rev. 497, 506-07 (2012) [hereinafter
Baradaran & McIntyre].
442 von Hirsch, supra note 281, at 725.
443 Id. at 726.
444 Fagan & Guggenheim, supra note 3, at 420.
445 Goldkamp, supra note 3, at 16-29.
170
adequately defining “what kind of future criminal conduct, and what degree
of likelihood of that conduct, warrants preventive confinement.”
446
As noted previously, simply because the public discussion over preventive
detention in America began through attempts to detain for purposes of public
safety and danger does not mean that preventive detention does not now
apply to flight. Accordingly, jurisdictions should be equally concerned with
adequately defining the risk they are trying to address surrounding court
appearance.
Prediction Validation
Von Hirsch’s second threshold requirement – to subject the prediction
method to careful and continuous validation – is necessary to check the
accuracy of the predictions. “Adequate validation studies of the predictive
technique in the model are required, regardless of whether the predictive
method is purely statistical, purely clinical, or a mixture of the two.”447 This
involves not only making sure the instruments are accurate; it also involves
making sure they are unbiased and nondiscriminatory, that they include good
data to begin with, and that they measure what we want to know.448 In bail,
where we are arguably always leaning toward release, we should be careful
to continuously check our methods of prediction – looking under the hood of
various techniques and instruments, so to speak – so that we do not allow
those instruments to become instruments of detention based simply on our
own ignorance. Releasing so-called “high risk” defendants (as measured by
an actuarial pretrial risk assessment instrument) may seem anathema to some
in the criminal justice system, but given what we know about risk and bail, it
is likely warranted and can ultimately lead to a better understanding of risk
overall.

446 von Hirsch, supra note 281, at 726.
447 Id. at 728.
448 The need for caution in using statistically-derived risk assessment instruments should by no means be
interpreted as a call to keep the status quo – which uses a combination of charge and money as a proxy to
determine risk. The use of actuarial pretrial risk assessment instruments is considered to be a legal and
evidence-based practice, whereas the traditional money-based bail system is one that is utterly unsupported
by both the law and the research. Moreover, the primary call for caution within this paper is only when
jurisdictions attempt to use those tools solely to determine whether to detain in the first instance (i.e., based
on pure prediction using aggregate risk).
171
Procedural Safeguards
Andrew von Hirsch’s third threshold requirement – to include certain
minimum procedural safeguards – is likely met simply by following the
Supreme Court’s guidance in United States v. Salerno, which approved of
the federal preventive detention statute, in part, because that statute provided
adequate procedural due process protections to further narrow the detention
eligibility net.
These threshold requirements, coupled with legal requirements and the need
to constantly seek to narrow and justify elements in the preventive model,
become heightened in this generation of bail reform. With more exacting
judicial scrutiny, many of this country’s longstanding detention schemes will
likely fail when held up to these analyses. Accordingly, jurisdictions
attempting to re-draw the line between release and detention must take all
these things into consideration when crafting their own release and detention
models.
172
Part III – A “Model” Release and Detention Process
The following is this author’s model release and detention process.
Jurisdictions may disagree with the model, but they should nonetheless
subject any alternative models to the same scrutiny and analysis as outlined
in this paper.
Articulating Generally Whom to Release and Whom to Detain449
Based on the totality of legal, historical, and empirical evidence documented
throughout this paper, the model release and detention process is crafted to
release all defendants except for those who pose an extremely high and
unmanageable risk either to willfully fail to appear for court to avoid
prosecution or to commit serious or violent offenses against reasonably
identifiable persons while on pretrial release. “Extremely high and
unmanageable risk” and other terms will be further defined and
operationalized, but for now the fundamental point should be that as a
society, we should reserve secure detention only for defendants posing
unmanageable pretrial risks of an extremely rare and serious nature.
Articulating the Detention Eligibility Net
This paper has already explained why it would be wrong to base a detention
eligibility net on actuarial risk alone. Thus, any model release and detention
process will likely include a charge-based net, and the only question
becomes which charges or crimes to include in that net.
In this generation of bail reform, we must wrestle with the pretrial research
showing that persons committing all different kinds of crimes pose all
different kinds of risks. Knowing that everyone coming in the jail door is
potentially a “high risk” defendant may make it tempting to hold all
defendants, assess them, and potentially detain them, but this paper
illustrates why that cannot be the way that we administer bail in America.
And while it is understandable for states to be concerned with all failures to
appear and all pretrial crime, this concern over any and all failures cannot
form the basis for creating the eligibility net for detention in the first

449 As an initial matter, this author recommends eliminating the words “bail” and “sufficient sureties”
whenever possible from existing legal schemes. The confusion surrounding these terms and phrases is
largely avoidable by merely referring to “release” and “detention,” as has been done in the D.C. and federal
statutes as well as in many states.
173
instance based solely on prediction. American bail law requires jurisdictions
to embrace risk and thus to expect some failure. Requiring jurisdictions to
embrace risk and expect some failure, in turn, means that jurisdictions must
differentiate among cases. Because that differentiation must be done in
advance (in both the substantive criminal law and in pretrial release and
detention American law relies on the moral deterrence of rules used to guide
personal conduct), we have to use methods with which Americans are
comfortable. Fortunately, the law supports making these differentiations
based on seriousness of the criminal case, and people in America seem
comfortable with using seriousness as a guide.450
For example, as a society, we are concerned with all crime, but we are more
concerned with certain, more serious crimes to the extent that we can be
comfortable with imposing higher penalties, such as life imprisonment and
possibly even the death penalty, in extremely limited cases. Likewise, in
bail, we are concerned with all crime, but we are more concerned with
certain, more serious crimes to the extent that we can be comfortable with
using pretrial detention in extremely limited cases in the first instance based
solely on prediction. We simply would not be as comfortable with detaining
persons charged with traffic offenses while they wait for their trials as we
would with a defendant on trial for murder. Moreover, while we care about
all crime committed during pretrial release, we are simply more concerned
when a defendant commits a new crime while on pretrial release during a
trial for a more serious case.
This is not to say that we are not concerned when a defendant commits a
serious crime while released on a minor crime. We are, but we must
remember that we can never completely predict that crime, and that we are
differentiating in advance by using the law and our own comfort levels
concerning criminal justice policy. When differentiating in advance, we are
likely more comfortable with the possibility of detaining a person who poses
a risk to commit a crime while on release for a serious crime than we are a
person who poses a risk to commit a crime while on release for a less-thanserious crime (just as we are comfortable with the fact that persons not
accused of any crime might walk the streets even though they may pose
some risk to act in a criminal manner). Generally speaking, then, just as life
imprisonment or the death penalty are the last step on a continuum of

450 These two things are intertwined. Of course, the law (created to reflect American notions of justice) can
trump aberrations in personal belief, such as when a person is comfortable with using the death penalty for
minor crimes. Like many areas of criminal justice, bail laws must be crafted for aggregate use.
174
responses as punishment to crimes, pretrial detention in the first instance
based on prediction is the last step on a continuum of responses to risks
associated with more and more serious cases. Thus, overall, the less serious
the case, the less we should entertain the idea of pretrial detention based on
prediction of risk.
As another example, as a society, we are concerned with all failures to
appear for court, but we are more concerned with multiple, willful failures to
appear – the act of flight to avoid prosecution – to the extent that we can be
comfortable with using pretrial detention for court appearance in the first
instance based on prediction.451 Moreover, we are more concerned when
these failures to appear occur during prosecution for extremely serious
crimes; indeed, it is fairly safe to say that as a society, we are simply less
concerned with failure to appear for court generally than with pretrial crime,
but that those concerns grow closer together when we are looking at certain
extremely serious crimes. Thus, a charge-based net encompassing higher
levels of criminal cases is appropriate, too, when seeking to detain persons
in the first instance based on risk of failure to appear for court.
States should be able to articulate their own charge-based detention
eligibility nets, but based on the issues raised in this paper, this author would
suggest reserving detention in the first instance based on prediction only for
those defendants charged with “violent crimes” or “violent offenses,” which
would encompass both violent felonies and misdemeanors, which, in turn,
often include instances of domestic violence. As will be shown, this is due to
the fact that detention can be at least initially justified for this group as
showing a higher risk to do the things we seek to potentially address through
detention. Until we have much more nuanced research – and it is not clear
that we ever will have research to the extent necessary to completely
overcome certain risk limitations – the various arguments for creating the
narrowest possible net (which include arguments that criminal charge is only
a part of risk assessment, and that we can never completely predict
individual risk for danger or flight, and legal or even policy reasons
concerning the undesirability of detaining certain defendants charged with
“lower level” crimes based solely on prediction) likely preclude

451 An interesting turn in the national conversation on pretrial justice involves whether willful flight to
avoid prosecution should even be considered to be equal to concerns over public safety. Most defendants
are simply too poor to flee, and advances in policing in America have led to the ability to track defendants
even to distant countries (when we decide to do so). This conversation might lead, interestingly, to readopting America’s initial stance on pretrial detention based on flight, which was that virtually no
defendant should be initially held purposefully to avoid failure to appear for court.
175
consideration of non-violent crimes, including crimes involving high
amounts of property damage, such as motor vehicle theft,452 or even felony
drug offenses.
453 Moreover, while it appears that the commission of violent
crimes while on pretrial release even among persons charged with violent
crimes is relatively rare, it must be remembered that we are only creating the
net; a further limiting process will likely be the key to making the detention
provision constitutionally acceptable. Nets broader than “violent crimes or
offenses” are not necessarily justified by either the law or the research, and
will require a more robust use of the further limiting process (requiring
increased resources for hearings).454
Nevertheless, there is some empirical justification for a net based on violent
charges. In 2006, the Bureau of Justice Statistics released “Violent Felons in
Large Urban Counties,” in which it concluded that “an estimated 70% of
violent felons … had been arrested previously,” “sixty percent of violent
felons had multiple prior arrest charges,” “a majority (57%) of violent felons
had been arrested previously for a felony,” and “thirty-six percent of violent
felons had an active criminal justice status at the time of their arrest [which
included] 18% on probation, 12% on release pending disposition of a prior
case, and 7% on parole.” Specifically, the report found, “about 1 in 4
released violent felons committed one or more types of misconduct while in
a release status. This misconduct usually involved a re-arrest for a new
offense (14%) or a failure to appear for court.”455 “Violent” felonies in that
study was defined as murder, rape, robbery, and assault (all with certain
inclusions and limitations) as well as “other violent offenses,” which
included vehicular manslaughter, involuntary manslaughter, negligent or

452 This represents a policy choice based on the law and the research. As will be shown later in this paper,
while no justification might exist to use these crimes in the net, failures for these types of crimes will be
eligible for detention in a secondary net based on pretrial failure. Motor vehicle theft may actually pose
some slightly elevated risk of re-arrest, but because we do not know individual risk or “risk of what,” and
because that overall risk is still quite rare, the policy choice involves erring on the side of initial release.
Overall, the net is less important than the further limiting process, also discussed infra.
453 Historically, we have assumed that “drug dealers” pose a heightened risk for flight, but that has not been
borne out by the research; indeed, most people charged with “dealing” drugs are not the high profile dealers
we occasionally see on television fleeing justice, and as to public safety the research has shown that
“though defendants with drug felonies are presumed to be dangerous, they are among the least likely to be
arrested for a violent crime.” Baradaran et al., supra note 441, at 558. If states are concerned that certain
drug offenses should be in the net, they might define the term “violent crimes” to include them, but
jurisdictions must then justify that inclusion through the history, the law, research, or other reasonable
legislative findings. Even then, the limiting process might preclude individual detention by making a
normative decision that the risk posed by drug defendants is not necessarily the risk we seek to address
through detention.
454 Likewise, nets that define “violent crimes” broader than the research will also need to be justified.
455 Brian A. Reaves, Violent Felons in Large Urban Counties, at 1, 3-4, 6 (BJS, 2006).
176
reckless homicide, nonviolent or non-forcible sexual assault, kidnapping,
unlawful imprisonment, child or spouse abuse, cruelty to a child, reckless
endangerment, hit-and-run with bodily injury, intimidation, and extortion.
Similarly, in a comprehensive review and study of prediction of pretrial
crime in 2011, authors Shima Baradaran and Frank McIntyre found that
while those charged with violent crimes are not necessarily more likely to be
arrested pretrial,456 “those originally charged with violent crimes,
particularly murder, were much more likely to be rearrested pretrial for
violent crimes.”457 Moreover, the authors found, because defendants charged
with all crimes except murder, rape, and robbery were generally only about
1%-3% likely to commit a violent crime while on pretrial release, persons
“charged with violent crimes are more likely [at 3% to 6%] to be rearrested
for violent crimes than those charged with nonviolent crimes.”458 This study
used the Bureau of Justice Statistic’s State Court Processing Statistics
dataset and its definitions of violent crimes.
Additionally, in a research brief published by the New York City Criminal
Justice Agency, the author examined a large group of New York City
defendants arrested in 2001 and found (similar to the above study by
Baradaran, et al.) that while defendants facing violent felony charges were
less likely be re-arrested overall, “violent felony [defendants] – when they
were re-arrested – tended to be re-arrested for similar offenses.”
459
Moreover, in a publication discussing the re-validation of the federal pretrial
risk assessment instrument (PTRA), the authors concluded, among other
things, that the validation research refined earlier contradictory findings by
“showing violent defendants [failed] at higher rates than other defendant
offense categories.”
460
Finally, the Laura and John Arnold Foundation’s PSA Court risk assessment
instrument includes a “violence flag,” which was created to help

456 Baradaran & McIntyre, supra note 441, at 528 (“The highest pretrial arrest rates were for defendants
charged with drug sales or robbery [classified as both a violent and a property crime] (21%), followed by
motor vehicle theft (20%), and burglary (19%). Those released who were charged with the ‘more
dangerous crimes,’ such as murder, rape, and felony assault, had much lower overall rates of pretrial arrest
at 12%, 9%, and 12% respectively.”).
457 Id.
458 Id. at 528-29.
459 Qudsia Siddiqi, Pretrial Re-Arrest for Violent Felony Offenses, at 6 (NYCCJA, 2008).
460 Timothy P. Cadigan, James L. Johnson, & Christopher T. Lowenkamp, The Re-Validation of the
Federal Pretrial Risk Assessment (PTRA), 76 Fed. Prob. 3, at 6 (AOC, UMKC, 2012).
177
jurisdictions to distinguish among defendants (facing felonies and
misdemeanors) based on their likelihood to commit crimes of violence.
According to the Foundation, “the goal of most criminal justice
decisionmakers is to detain defendants who pose a risk to public safety –
particularly those who appear likely to commit crimes of violence – and to
release those who do not.”461 The violence flag is derived from looking at
five variables, including whether the current offense is violent and whether
the defendant has had a prior violent conviction, indicating, as well, that
there exist empirical data for drawing the line between release and detention
at “violent offenses.”462 Observations on using the instrument in the field to
identify defendants likely to commit violent crimes have illustrated the
tool’s utility. As Milgram and others have written, “[T]he tool helped judges
identify the small number of defendants – approximately 8 percent – who
were far more likely to commit violent crime than average defendants.
Defendants flagged by the tool as being at an elevated risk of violence who
were released pretrial were, in fact, re-arrested at a rate of 17 times higher
than that of other defendants.”
463
Each of these studies provides at least some empirical justification for a
detention eligibility net consisting of “violent offenses.” Jurisdictions must
remember, however, that in America we start with the notion that all
defendants should be released pretrial. A detention eligibility net simply
recognizes that the Supreme Court has said jurisdictions can, in advance,
articulate that defendants facing certain crimes have the potential for pretrial
detention. Until now, these nets have had very little justification, or they
have been justified based on false assumptions. To the extent that states
desire to carve out any exception to release, then, the fundamental point is
that exceptions may not be created arbitrarily or by whim, but must instead
be justified in some lawful manner.
Articulating the “Secondary Net”
There appears to be very little empirical evidence464 for expanding the
eligibility net for detention in the first instance based solely on prediction
beyond “violent crimes.” This notion is bolstered by American principles of
liberty and freedom, the risk that America has historically sought to address,

461 Developing a National Model For Pretrial Risk Assessment, at 1 (LJAF, 2013).
462 See Public Safety Assessment: Risk Factors and Formula, at 2 (LJAF, 2016).
463 Milgram, et al., supra note 231, at 220.
464 Nets might be justified on legislative findings using something other than empirical evidence, but those
findings will undoubtedly be scrutinized by the courts for rationality and reasonableness.
178
general legal standards and norms, and the inherent limitations of assessing
risk. Moreover, historically American jurisdictions have had less difficulty
in coming to near consensus when defining the term “violent” versus
“serious” or “dangerous,”
465 which also suggests using the term “violent”
over the more nebulous terms. Nevertheless, the net could be justifiably
expanded to include persons who are charged with jailable offenses or who
willfully fail to appear while on pretrial release, including release pending
sentencing or during appeal, for any offense.
466 This net could be more
limited – for example, only triggered by a defendant committing only a
“serious” crime while on pretrial release – but it does not have to be. In this
model, the willful failure to appear for court or the commission of any
jailable offense while on pretrial release may trigger the secondary net, and
the further limiting process for that net will serve to keep detention within
constitutional limits.
This author calls this net the “secondary net” rather than bond revocation for
one primary reason. Throughout the history of bail in America, courts have
not provided the sort of procedural due process that is constitutionally
required to detain in the first instance based solely on prediction, and bond
revocation hearings are often even more perfunctory. Accordingly, rather
than have courts see a pretrial violation as something that automatically
leads to a revocation of release, this author hopes jurisdictions will look
upon the new circumstances as creating a new, secondary eligibility net for
detention, which requires an equal amount of due process protection as is
required for initial detention. In short, rather than merely denying release
based on a new allegation, courts will be trying, once again, to determine if
the failure was willful, and whether the risk of what we fear in the future is
unmanageable.
The need for this sort of secondary net is acute in America. In a risk-based
system, it is the one instance where we have potential proof of individual

465 The United States Code defines “crime of violence” to mean “(A) an offense that has as an element of
the offense the use, attempted use, or threatened use of physical force against the person or property of
another; (B) any other offense that is a felony and that, by its nature, involves a substantial risk that
physical force against the person or property of another may be used in the course of committing the
offense; or (C) any felony under chapter 109A, 110, or 117.” 18 U.S.C. § 3156.
466 There is probably no need to include any provision dealing with crimes that somehow affect the
administration of justice, or protect witnesses or jurors, as some states have done. As noted previously, the
Court in Salerno largely did away with the distinction of danger in and out of the justice system. To the
extent that a defendant obstructs justice by committing a crime (such as threatening a witness or bribing a
juror), this model would apply. Moreover, the model does not eliminate traditional notions of incarceration
for contempt.
179
risk. The pretrial release and detain decision is, generally speaking, an
attempt by judges to predict who will fail out of some group of defendants.
When defendants actually fail after release, the decision is made simpler
because now there is evidence of the one thing we are unable to determine
for defendants before initial bail settings – that is, whether this particular
defendant, whether “low,” “medium,” or “high” risk, will be the defendant
who actually fails by doing the sort of thing we wish to avoid.467 The need is
acute, legally speaking, only because there is not always agreement among
the states on whether a bailable defendant can have his bail revoked and
release denied after he flees or is accused of another crime, or whether
courts must instead continually release him (or at least “set bail”) so long as
the underlying crime or the new crime is a “bailable” offense.468
Inclusion into the secondary net should not be based on so-called technical
violations of bond, unless violating those conditions is made unlawful by
statute.469 In a theoretically pure release and detention system, the only two
constitutionally valid purposes for limiting pretrial freedom are court
appearance and public safety. Accordingly, willful failure to appear and new
criminal activity – and not things such as missed check-in meetings – are the
only two things that should trigger eligibility for detention. Nevertheless,
this net allows jurisdictions to decide when violations of conditions of bond
might rise to the point of criminal prosecution.
Overall, this net is larger than the initial net; for example, any new crime and
willful failure to appear for court for any charge can trigger the detention
process. This net should be small (and no larger than any bond revocation
net), however, simply due to the relatively low numbers of failures pretrial.
Moreover, the further limiting process, described below, will keep from
over-detaining individuals who are not high risk to continue to willfully fail
to appear or to commit additional crimes in the future. While some persons
argue that courts should wait for several missed court dates before allowing
the initiation of a detention hearing, this author believes that because the

467 Occasionally, there have been arguments raised in the literature that committing a new crime while on
pretrial release should not lead to detention due to the fact that the second crime, like the first, is merely an
unproven allegation. However, concerns surrounding this issue are softened by knowing that the second
charge leads only to detention eligibility – not automatic detention – and will not eliminate the need for a
further limiting process with the various procedural due process safeguards.
468 See LaFave, et al., supra note 52, § 12.3 (b), at 63-67.
469 There is a trend in the pretrial field toward measuring technical violations as outcomes, but technical
violations should never trigger pretrial detention. Moreover, the crime of “violation of bond conditions,”
has been misused in many jurisdictions, making this author conclude that jurisdictions should use extreme
caution in using it as a basis for detention.
180
detention eligibility net for initial detention is narrow, this secondary net
should allow judges to make a more detailed inquiry for flight after a single
willful non-appearance. Finally, this proposed secondary net does not
include detention beyond temporary detention for failure while on postconviction release, relying instead on the existing processes to operate
separately.
Comparing these two nets with the current American Bar Association
Standards helps to understand the more-detailed mechanics. In doing so, it is
also helpful to remember that when new crimes occur, there is the potential
for detention under two separate cases: (1) the case involving the initial
alleged crime; and (2) the case involving the crime alleged to have occurred
while on pretrial release.
Recall that the ABA Standards’ net allowed detention in the first instance
for: (1) defendants facing violent or “dangerous” offenses; (2) defendants
charged with “serious” offenses while on pretrial release for a “serious”
offense, or on probation or parole for a violent or dangerous offense, or on
other post-conviction release; (3) defendants charged with serious offenses
posing a substantial risk of non-appearance; and (4) any offense with a
showing of substantial risk of obstructing justice. Under the Standards, bond
revocation [what this author calls the secondary net] may occur whenever a
defendant willfully violates any condition of release.
This author’s proposed model allows detention in the first instance based
solely on prediction only for defendants charged with violent offenses. This
is done for several reasons, which include the facts that: (1) the research, the
law, and all other concepts addressed in this paper do not support any initial
detention net beyond violent offenses for avoiding the harms we seek to
address; (2) the research, instead, shows that most defendants overall are
likely to succeed, and at rates better than expected if released on conditions;
(3) the term “serious” is too loose a concept on which to base detention; (4)
the fact that a defendant is currently on pretrial release for another charge is
not enough, by itself, to allow detention in the first instance for anything less
than a violent offense as the defendant is still un-convicted on both offenses,
and detention might be authorized on the underlying charge under the
secondary net; (5) if on probation, parole, or other post-conviction release,
the system allows for detention for the underlying charge. For all the reasons
outlined in this paper, unless a defendant is charged with a violent offense,
there is simply no trigger for detention in the first instance based solely on
181
prediction. In the author’s proposed model, the secondary net operates to
detain a person in the first case who has willfully failed to appear for court
or accused of committing a jailable offense while on pretrial release.470 It
will not allow detention based on willful violation of conditions that are
neither criminal nor a willful failure to appear for court (although it will
allow the defendant’s decision to refuse lawful conditions by not signing the
agreement to result in his or her detention). The proposed model has no
provision dealing with “obstructing justice;” as noted previously, historically
this concept has been folded into notions of public safety and court
appearance and can be dealt with through judicial contempt power for
serious cases in any event.
Like the American Bar Association Standards, the proposed model would
include a provision for the temporary detention of persons charged with
violent crimes (the primary net), persons charged with a new offense while
those persons are on pretrial release, including release pending sentencing or
during appeal (the secondary net), and persons currently on probation or
parole for any offense. This temporary detention provision allows time for
proper notifications, the implementation of revocation proceedings, and the
lodging of detainers. Like the ABA Standards, there should be strict time
limits and release on conditions if the hearing is not timely.
Finally, the use of actuarial risk assessment in the model can provide a
crucial function by continually leading jurisdictions toward release under the
model. While it should not be used solely to determine detention in the first
instance, it can be used to systematically weed out defendants from
detention eligibility based on aggregate risk, so long as courts understand the
various limitations of aggregate versus individual risk.
Can We Create a Different Net?
Jurisdictions can certainly create their own detention eligibility nets. The
fundamental point of this paper is that any net must be justified by the law or
the research in at least some rational way – beyond unfounded assumptions
– to survive court scrutiny.

470 Limiting charges to ones carrying a potential jail term seems intuitively beneficial, and detention should
never be ordered for a defendant who faces no possibility of jail, but many municipal codes have made a
wide variety of offenses jailable, and even contain catch-all provisions providing for fines and jail terms in
the event a specific provision is silent about punishment.
182
Articulating the “Further Narrowing Process” for Detention
Historically and legally, we require a process designed to individualize bailsetting and to further limit the detention eligibility net. In this proposed
model, we will articulate two processes: one for detaining defendants in the
first instance based solely on prediction (the primary net), and one for
detaining defendants who have already failed while on pretrial release (the
secondary net). Because of the concerns raised in this paper, the limiting
process applied to the primary net is more stringent than the one applied to
the secondary net.
For as long as America has been intentionally detaining defendants, the
limiting process has suffered from a lack of adequate definitions, and yet this
process should be considered the most important element of a release and
detention system. Indeed, a proper limiting process can “cure” an overbroad
net by fairly and adequately narrowing the numbers of actual pretrial
detainees.
Historically, jurisdictions seeking to detain based on “risk” have primarily
used (as noted previously, the states vary on the processes) a limiting
process that allows detention when there is clear and convincing evidence
that no condition or combination of conditions will suffice to provide
adequate assurance of court appearance or public safety. This process,
however, is subjective and resource driven. Moreover, without proper
definitions, it is inadequate in telling judges how high a risk and what type
of risk we seek to address. For all of the reasons outlined previously in this
paper, we must consider changing this historic process to better reflect the
law and the research. States that already have processes with elements
described in this model process (such as a good statement defining the sort
of risk they seek to address) will have less reason to change.
Accordingly, following general American notions of liberty and freedom,
what America has historically sought to address and how to address it,
general legal standards and norms, and the inherent limitations of assessing
risk, this initial process to be used for detention in the first instance should
require prosecutors to convince a neutral judicial official that there is clear
and convincing evidence471 as shown through specific facts and

471 Because it was not expressly articulated in the Bail Reform Act of 1984, the federal law, quite counterintuitively, has gradually settled on the “preponderance of the evidence” standard to show flight risk versus
“clear and convincing evidence” to show risk of harm. This author has found no adequate rationale for the
183
circumstances that the defendant will flee or attempt to flee to willfully
avoid prosecution or that the defendant will commit or attempt to commit a
serious or violent crime while on pretrial release against a reasonably
identifiable person, or groups of persons or their property, and that no
condition or combination of conditions will suffice to manage the extremely
high risk. An actuarial pretrial risk assessment instrument may be used as
one factor in this process, but may not be used as the sole justification or
basis for detention.472
At the very least, this narrowing process will be a vast improvement over
prior language, and by avoiding over-consideration of aggregate risk, it will
better follow the Supreme Court’s suggestion that jurisdictions attempt to
show “an identified and articulable threat to an individual or the
community,” while further requiring consideration of what, exactly, that
threat should be.
This process must include the various procedural due process elements
approved by the Supreme Court in Salerno, including: (1) the arrestee
should be entitled to a prompt hearing and the maximum length of pretrial
detention should be limited by stringent speedy trial time limitations; (2) the
arrestee should be housed separately from those serving sentences or
awaiting appeals; (4) after a finding of probable cause, the arrestee should be
given a full-blown adversary hearing through which the limiting process is
used, which would include a right to counsel and the right to testify or
present information by proffer and to cross-examine witnesses who appear at
the hearing; (5) objective statutory factors to guide judges in addition to the
statistically-derived risk assessment that go to the more adequately defined
risk, including factors such as the nature of the charge, other characteristics
and statements of the defendant, and other facts and circumstances that are
not necessarily addressed by an actuarial pretrial risk assessment instrument.
Additionally, judges must make written findings of fact for orders of
detention, which are to be given immediate appellate review.
For persons who fall within the secondary detention eligibility net by
willfully failing to appear for court or being charged with a new crime, the
process is the same except that the prosecutor must show clear and

difference, and so in an attempt, once again, to make detention carefully limited, this proposed model uses
“clear and convincing evidence” standard for both risk of flight and public safety.
472 The same is true for other risk tools for other defendant populations, such as the Ontario Domestic
Assault Risk Assessment.
184
convincing evidence as shown through specific facts and circumstances that
the defendant will flee or attempt to flee to avoid prosecution or that the
defendant will commit or attempt to commit any jailable offense while on
pretrial release, and that no condition or combination of conditions will
suffice to manage the extremely high risk. In this case, the evidence of a
willful failure to appear or commission of a new crime, along with assessed
risk from an actuarial pretrial risk assessment instrument, may be used as
substantial evidence of the potential for unmanageable risk.
No rebuttable presumptions should be used in this model, for two reasons.
First, the research (as well as various limitations of risk prediction) simply
does not support any rebuttable presumption toward detention, and because
of that, it is even more unfair to force defendants to attempt to prove they are
not dangerous, that they will not do some unknown but forbidden act, and
that they will not flee. Second, our country’s history of using rebuttable
presumptions has only led to their misuse, causing jurisdictions to treat them
more like un-rebuttable presumptions. The only presumption should be a
general presumption of release in all cases or more specific presumptions
similarly guiding courts toward release that must be overcome by the
government.
185
Part IV – Holding Up the Model to the Three Analyses
The Model Held Up to General Narrowing Principles
This particular model for pretrial detention largely succeeds when held up to
the three analyses, discussed above. Based on the general narrowing
principles gleaned from the history, the law, the research, the national
standards, and the limitations of actuarial pretrial risk assessment, this model
provides a good compromise between releasing virtually everyone – a
position also supported from the concepts addressed in this paper – and the
need for jurisdictions to have some ability to protect the public from rational
and articulable risks. The model makes sense of the history of bail by taking
into account historical notions of both risk for flight and danger. It makes
sense of the law by creating a purposeful in-or-out release and detention
process with nothing hindering the decision. It makes sense of the pretrial
research by correcting various faulty assumptions and by “looking under the
hood” of actuarial pretrial risk assessment instruments. And, finally, it
makes sense of the national standards by largely following those
recommendations (albeit with some modifications), which were, in turn,
adopted in America’s “big fix.”
Potentially detaining only persons charged with violent offenses in the first
instance based solely on prediction or after failing to appear or committing a
crime while on pretrial release might lead to more purposeful detentions than
America is currently used to, but overall the model should lead to far fewer
detentions generally than those occurring through the use of money. Most
importantly, the model appears to solve many of the problems that have
historically led to bail reform while maintaining the overall American notion
of a presumption of freedom and liberty, recognizing that most risk can be
managed through conditions less restrictive than secured confinement, and
allowing pretrial detention as an exception reserved for “extreme and
unusual” cases.
Under this analysis, the model serves as a justifiable way to release and
detain pretrial, but, like bail reform generally, it only re-emphasizes
America’s need to better embrace the risk of purposeful release at bail
instead of relying on the somewhat random method of deciding release and
detention based on wealth. Nevertheless, this is the way American bail is
supposed to be – rational and purposeful. Moreover, by using a charge-based
186
detention eligibility net, articulating better definitions, and attempting to
answer the question, “risk of what,” the model is also superior to most
current detention models and avoids many of the problems associated with
basing detention solely on actuarial pretrial risk assessment.
This analysis also requires that jurisdictions constantly assess whether
detention may be further narrowed, and indeed it could. For example, a state
that chooses to leave its detention eligibility net at “capital offenses”
(narrower than “violent offenses”) would also be justified by the law and the
research. Likewise, a jurisdiction making a policy decision to simply narrow
the net from all violent crimes to only violent felonies would find ample
justification for such a change. A much narrower net, however, might strain
our current tolerance for risk, which appears lower than when America was
founded. Likewise, jurisdictions could make various elements of the “further
limiting process” narrower, and this narrowing, too, is justified by the
history, the law, and the research. The fundamental point is that while
narrower detention processes are justifiable, any broader process than that
outlined in this paper does not appear today to be justifiable by the history,
the law, or the pretrial research.
The Model Held Up to the Law
The model also largely succeeds when held up to the law. It is justifiable,
rational, and seemingly fair, and would likely survive even strict scrutiny by
the courts. First, by using a charge-based net, the model avoids problems
with vagueness, which still theoretically plague any detention model based
on risk or other subjective notions. Second, because it includes a floor,
below which no detention would be allowed, it appears to satisfy even the
most lenient view of excessive bail analysis. Third, it avoids equal protection
problems largely by avoiding the use of money. Fourth, while the model
likely easily surpasses the test to avoid punishment prior to trial, it also
appears to hold up to Salerno’s discussion of elements necessary to avoid
due process violations. Specifically, it narrowly focuses on an acute problem
that is identifiable and justifiable from the research, it only operates on
defendants accused of “extremely serious offenses,” and it contains a due
process hearing elemental to any liberty deprivation.473

473 It is clear that the law will be the primary guide to shaping America’s detention provisions. Moreover,
certain elements of the law – even elements not primarily discussed in this paper – may be persuasive, if
not deciding of the issue. For example, in an unpublished manuscript by noted legal researcher Sandra G.
Mayson, the author challenges the assumption that the government has constitutional, moral, or practical
187
The Model Held Up to von Hirsch’s Threshold Requirements
Finally, the model succeeds when held up to von Hirsch’s threshold
requirements. First, as in the discussion under vagueness, above, the model
better defines the risk that we are trying to address. Second, it uses certain
advantages of statistical risk prediction while recognizing that the risk we
seek to avoid is often something different from that assessed by an actuarial
tool. Moreover, the further narrowing process, along with a procedure for
dealing with failures while on release, should lead to a better understanding
of risk, which is hindered and masked by our tendency to over-detain
defendants today. Finally, as mentioned under the legal analysis, it has the
sort of due process protections approved by the Court in Salerno, but largely
missing from many state detention processes today.
The Model Applied to Difficult Hypotheticals
This particular model also works for some of the more common vexing
hypotheticals that currently exist at bail. For example, within the net, it
would allow for judges to detain persons charged with violent offenses who
score extremely “low risk” on an actuarial pretrial risk assessment
instrument, when there are facts and circumstances indicating the need for
detention. Persons who are extremely high risk but outside of the net due to
being arrested on a less serious charge are not initially detained as a
purposeful choice based on American principles rooted in the history, the
law, and the research, but would be eligible for detention if our initial
prediction of risk was faulty. In short, it would eliminate many false
positives, but would allow courts to deal with false negatives in a fair and
transparent manner.
Other complex hypotheticals are also accounted for under the model, even if
those hypotheticals raise problems with solutions ultimately found outside of
bail. For example, a homeless person who has a long history of trespasses
would never be detained under this model in the first instance based solely
on prediction; instead, he or she might be detained after failing on release
(something we would not have been able to predict initially in any event) if
the requisite process is used. This outcome simply represents the model

bases to exercise preventive restraint over defendants versus equally dangerous persons not accused of
crimes. Among other things, the author concludes (as in this paper) that preventive detention requires far
more justification than that given to date.
188
making a purposeful decision to limit detention in the first instance based on
prediction alone to some identifiable group of extremely dangerous persons,
the homeless trespasser not being one. Nevertheless, if that homeless person
fails on release and poses a significant risk for either flight or public safety,
he or she may be detained. Ultimately, though, the problem of a homeless
person committing multiple trespasses is likely solved not through bail, but
through other means, such as general sentencing considerations and
innovative sentencing designs, evidence-based criminal prevention
programs, programs to reduce homelessness, etc. The same is true for
mentally ill persons, illegal immigrants, and other so-called “difficult
groups” of persons, with whom jurisdictions have traditionally struggled.
This is an important point worth repeating. Jurisdictions must recognize that
not every problem can be solved at bail. Accordingly, in addition to
prevention and sentencing, various scholars, including LaFave, have offered
other solutions to address some of these problems, including: a widening of
the revocation procedure; innovative policing methods in addition to police
simply releasing more persons rather than booking them into jail; triage
procedures for addicts, alcoholics, and the seriously mentally ill into
behavioral health pathways; still other triage procedures for minor offenses,
and the overall reduction of delays in normal trials; using special dockets for
even speedier trials for higher risk persons; and adding resources, such as
pretrial services functions, to help with the release and detention process.474
Many of the vexing hypotheticals involve what are, in fact, aberrational
cases. A fundamental principle underlying the model, however, is that just as
we can never fix every conceivable problem through bail, we can also never
base release and detention law or policy on aberrational cases. Occasionally,
a person charged with drunk driving will drink and drive while on pretrial
release, putting many persons at danger. But we can never know who that
person will be, and the risk may not necessarily lead to actual harm – indeed,
under many risk assessment instruments, the driver might be deemed high
risk and fail by merely committing another, non-drinking infraction. To
capture the single person who drinks, drives, and hurts someone on pretrial
release, we would have to include all persons charged with drunk driving in
the net, and, frankly, to detain every one of them to eliminate the risk. In

474 See LaFave et al., supra note 52, § 12.3(g), at 88-89; see also John Jay, supra note 307, at 20 (Statement
by Dr. Faye Taxman, George Mason Univ.), 23 (statement of Dr. Saurabh Bhargava, Carnegie Mellon
Univ.), and 26 (statement by Dr. Mike Jones, Pretrial Just. Inst.). Many of these scholars articulate a need
for criminal justice stakeholders to look for nontraditional solutions, or to “think outside of the box.”
189
bail, we must work in aggregates, but use rational processes to individualize
conditions and to manage risk. The law requires reasonable, and not
complete assurance of public safety and court appearance, and this proposed
model provides a mechanism to obtain that reasonable assurance.
Overall, the model attempts to answer the three big questions we have asked
ever since a thing called the pretrial phase of a criminal case has been in
existence – whom do we release, whom do we detain, and how do we do it?
It does so by following the history, the law, and the research to adequately
define the level of risk and the kind of risk we seek to address to justify
secure detention prior to trial.
190
Part V – The Language of the Model
If adopted, the particular language of any model like the one expressed here
will largely depend on a given state’s current release and detention scheme
as well as philosophical notions surrounding articulating constitutional or
statutory rights of defendants. Nevertheless, it must be recognized that
statistical risk assessment may one day be able to fully predict the kind of
risk we fear, and might lessen the subjective and political aspects causing
reluctance for use within (or to help determine) the detention eligibility net.
Moreover, in the future, the risk research may show that we may be fully
justified in detaining defendants for certain categories of crimes that we
simply are not justified in detaining today. For these and other reasons, if
language is inserted into a constitution to allow for detention based on this
model, it should be broad enough to describe an overall release and
detention process, but narrow enough to keep legislatures and courts from
expanding detention beyond constitutional limits.
Every current constitutional right to bail provision has at least two parts: (1)
a section, line, or clause concerning release; and (2) a section, line, or clause
concerning detention. The most basic of these provisions are found in the socalled “broad right to bail” states, and those provisions read something like,
“All persons have a right to bail, except for capital offenses (or other
offenses), when the proof is evident and the presumption is great.” In this
example, the two clauses of the single sentence provide for the release/detain
dichotomy. As mentioned previously, a jurisdiction operating under the
above language need not change this language to implement the model, but
in a moneyless (or otherwise intentional release/detain scheme), the
jurisdiction would likely be required to release all “bailable” defendants, and
thus the jurisdiction would need to be content with capital offenses as its
detention eligibility net, and would need to be similarly content with
releasing everyone – at least in the first instance – who is not charged with a
capital offense. Other states with different nets (i.e., capital offenses,
treason, persons facing life imprisonment) will likewise need to be
comfortable with the existing nets or be prepared to change. States with
broader preventive detention provisions likewise may not have to change
(unless a court finds the net to be too broad), but they must be content with
their expanded nets in an intentional system and be prepared to justify those
nets pursuant to the law and the research today. States with no right to bail
191
provisions will need to work with these issues within their current statutory
(or court rule) release and detention framework.475
In addition to these two parts, a third part to a model constitutional provision
is contained in many states with preventive detention – that is, a provision
giving direction to the legislature or the courts for carrying out the basic
release and detention policy. While technically part of the release provision,
many persons believe that a fourth part is also necessary, which is a line
expressly articulating that no condition of release should lead to the
detention of an otherwise bailable or releasable defendant.
Based on the fact that there are perhaps four main concepts to be contained
in any release/detain provision (release, detention, possibly enabling
language, and a “no condition shall detain” clause), and that these four
concepts can have a variety of linguistic formulations, it is hard to anticipate
the sort of language states will wish to adopt. Nevertheless, because this
author believes that most states will want to change their basic in-or-out
structure to take advantage of what we know in this century about risk,
release, and detention, the following example templates are provided for
guidance:
1. Relatively Brief Model Provision
All persons charged with a criminal offense shall be released on
the conditions that they return to court and abide by all laws. If
needed, a court may impose such further and least restrictive
conditions necessary to provide reasonable assurance of court
appearance and public safety, except that a court may confine a
person who is eligible for pretrial detention [the detention
eligibility net, which could be articulated here or elsewhere] when the court finds, in addition: (1) clear and convincing
evidence as shown through relevant facts and circumstances
that the person poses an extremely high risk of intentional flight
to avoid prosecution; or (2) clear and convincing evidence as
shown through relevant facts and circumstances that the person
poses an extremely high risk to commit or attempt to commit a
serious or violent crime while on pretrial release against a
reasonably identifiable person or groups or persons or their

475 Of course, a state with no current right to bail provision in its constitution could add one.
192
property. In addition, a court may confine a person when the
court finds: (a) probable cause that a person already on pretrial
release for any jailable offense willfully failed to appear for
court to avoid prosecution or has committed a new jailable
offense; and (b) clear and convincing evidence as shown
through relevant facts and circumstances that the person poses
either an extremely high risk to willfully fail to appear for court
to avoid prosecution or to commit or attempt to commit any
new jailable offense against persons or their property.
For all persons eligible for pretrial detention under this
provision, the court must also find clear and convincing
evidence that no condition or combination of conditions will
suffice to manage the person’s extremely high level of risk. In
considering the facts and circumstances to detain persons under
(1) or (2), above, the court may consider the risk assessed
through an actuarial pretrial risk assessment instrument,
however the court may not detain based solely on the results of
that instrument. In considering the facts and circumstances to
detain persons who have willfully failed to appear for court or
committed a new crime while on pretrial release, the court may
rely substantially on the assessed risk from an actuarial pretrial
risk assessment instrument. In all cases, the court may not
impose a condition of release that results in the pretrial
detention of the person. However, a person’s willful refusal to
agree to lawful conditions of release may result in the detention
of that person.
The legislature shall enact laws designed to carry out these
provisions, including specific and reasonable laws allowing for
the temporary detention of defendants not to exceed three days,
excluding Saturdays, Sundays, and holidays, with an additional
two calendar days granted to the state upon a motion showing
good cause and no fault for the delay by the state, and
additional periods of time granted to the arrested person upon
motion and for good cause shown. The legislature shall also
enact laws to define the terms “serious” and “violent” crimes, to
create a due process hearing required for pretrial detention, and
to provide for speedy trials and immediate and expedited
appeals for detained defendants. At a minimum, the detention
193
hearing shall contain [insert procedural due process
requirements from Salerno here].
Of course, jurisdictions could – and, indeed, may want to – include their
detention eligibility nets in their version of this language. This model does
not include it only to allow for changes in research that might affect the
primary detention eligibility net in the future, but with certain limits
designed to make sure only a small group of unmanageable defendants
(extreme and unusual cases) may be detained no matter how broad the net.
By stating the type of risk jurisdictions seek to address, the language subtly
urges the risk research to design ways to better assess that particular risk. Of
course, because a release and detention process will be reviewed by the
appellate courts as to how all parts of it work together – constitution,
statutes, rules, etc. – an even broader constitutional provision could be
enacted to simply allow judges to detain extremely high risk persons so long
as it was narrowed through implementing statutes or rules. Without some
constitutional protection, however, this sort of extremely broad language
could easily slip toward abuse if not significantly narrowed through the
implementing laws. The more protections a jurisdiction wants, such as
limiting the definition of violent offenses, etc., the more it can include into
its constitutional provision. Of course, this language could be altered to
accommodate states without constitutional right to bail provisions, although
the reasons for including various terms and phrases are the same.
2. A More Detailed Provision
Release Provision
Following fundamental American notions of due process and
individual liberty, all persons arrested in the State of [insert
state name here] shall be presumed eligible for release from jail
or other confinement unless the State demonstrates
extraordinary reasons for pretrial detention, subject to the
guidelines contained herein. The people of the State of [insert
state name here] find and declare that the right to release prior
to trial is one that must be vigorously protected, that liberty
during the pretrial phase of a criminal case is the American
norm, that allowing conditions of release (including money) to
cause pretrial detention violates fundamental legal principles
and rights that are given to citizens under the United States and
194
____ Constitutions, and that defendants who are presumed
innocent in a free society may be detained prior to trial only
through a fair and transparent procedure with numerous
procedural due process safeguards designed to detain only those
persons absolutely necessary due to the risk associated with
their release.
Detention Provision
Notwithstanding any particular defendant’s presumptive right
to release pretrial, the people of the state of [insert name of state
here] also find and declare that to adequately protect the public
safety and the integrity of the judicial system through assuring
the appearance of the accused at court, a system for pretrial
detention that is designed to measure and respond to pretrial
risk shall be established. Accordingly, a court may confine a
person who is charged with a violent crime when the court
finds, in addition: (1) clear and convincing evidence as shown
through relevant facts and circumstances that the person poses
an extremely high risk of intentional flight to avoid prosecution;
or (2) clear and convincing evidence as shown through relevant
facts and circumstances that the person poses an extremely high
risk that he or she will commit or attempt to commit a serious
or violent crime while on pretrial release against a reasonably
identifiable person or groups or persons or their property. In
addition, a court may confine a person when the court finds: (a)
probable cause that a person already on pretrial release for any
jailable offense willfully failed to appear for court to avoid
prosecution or has committed a new jailable offense; and (b)
clear and convincing evidence as shown through relevant facts
and circumstances that the person poses either an extremely
high risk to willfully fail to appear for court to avoid
prosecution or to commit or attempt to commit any new jailable
offense against persons or their property. For all persons
eligible for pretrial detention under these provisions, the court
must also find clear and convincing evidence that no condition
or combination of conditions will suffice to manage the
person’s extremely high level of risk. In considering the facts
and circumstances to detain persons under (1) or (2), above, the
court may consider the risk assessed through an actuarial
195
pretrial risk assessment instrument, however the court may not
detain based solely on the results of that instrument. In
considering the facts and circumstances to detain persons who
have willfully failed to appear for court or committed a new
crime while on pretrial release, the court may rely substantially
on the assessed risk from an actuarial pretrial risk assessment
instrument.
Enabling Provision
The legislature shall enact laws designed to carry out this
provision, including specific and reasonable laws allowing for
the temporary detention of defendants not to exceed three days,
excluding Saturdays, Sundays, and holidays, with an additional
two calendar days granted to the state upon a motion showing
good cause and no fault for the delay by the state, and
additional periods of time granted to the arrested person upon
motion and for good cause shown. The legislature shall also
enact laws to define the terms “serious” and “violent” crimes, to
provide for a due process hearing required for pretrial detention,
to provide for the use of an actuarial pretrial risk assessment
instrument, to provide for speedy trials and immediate and
expedited appeals for detained defendants, to create sanctions
that are less restrictive than detention for violations of
conditions of release, including for failure to appear for court or
for the commission or new crimes while on pretrial release, to
provide requirements for release and detention orders, and to
allow credit for pre-sentence detention and provide for the
periodic re-examination or review of the need for continued
detention during the criminal case.
The legislature shall enact procedures for the release or
detention hearing that uphold and protect the defendant’s rights
to procedural due process and that ensure that pretrial detention
remains a carefully limited exception to the law favoring
pretrial release. The detention hearing, at a minimum, shall be
held before a neutral judge or magistrate, shall allow defendants
the right to counsel, to testify, to cross-examine witnesses, and
to present evidence. The rules governing admissibility of
evidence shall not apply, the hearing shall be recorded, and the
196
testimony of the defendant shall not be admissible in the case in
chief, except for a prosecution for perjury based on that
testimony or for the purposes of impeachment in any
subsequent proceedings. If the hearing results in a decision to
detain, the judicial official shall provide the reasons for the
decision either orally or in writing within three days.
Those defendants not detained pursuant to the process
authorized by this amendment shall be released on those
reasonable, least restrictive, and individually tailored conditions
necessary to provide reasonable assurance of court appearance
and public safety. Financial conditions may not be set to
address issues of public safety, and no condition, including a
financial condition, may result in the pretrial detention of an
otherwise releasable defendant. However, a person’s willful
refusal to agree to lawful conditions of release may result in the
detention of that person. The legislature shall enact such
additional provisions as are necessary to effectuate a statewide
pretrial release scheme, using the tools and resources of the
various state courts, that maximizes court appearance and
public safety rates for those defendants deemed eligible for
release.
From these two examples, one can see that the various permutations are
seemingly endless. Nevertheless, even now the various state constitutional
provisions similarly range from relatively short to relatively long and more
detailed. Of course, and again, states without a constitutional right to bail
provision will be working on these issues in their statutes or court rules.
Under This Model, What Will Be Our Ratio of Released to
Detained Defendants?
The answer to that question is unknown. As noted previously, however, it is
better to enact a fair and rational process based on the law and the research
first and let the ratio determine itself, than it is to come up with a ratio first
and then attempt to design a process to reach that goal. Jurisdictions are
reminded, however, that most actuarial pretrial risk assessment instruments
only currently label approximately 10% of defendants “high” risk for failing
to appear for any reason and for committing any new crime. Moreover, they
197
are reminded that the District of Columbia currently only detains
approximately 5% of arrested defendants, and the entire justice system in
that jurisdiction is relatively pleased with this number. Thus, jurisdictions
should not be surprised if the ultimate number of detained defendants is 10%
or even significantly lower.
What Will My “Failures” Be Under This Model?
The author cautions jurisdictions not to define “failures” pretrial as we have
been defining them in America’s recent history. By re-defining the risk that
we hope to address – for example, by redefining failure for court appearance
from any FTA to only willful FTAs with a purpose to avoid prosecution, our
“failures” in the future should be extremely low.
198
Part VI – Essential Elements of Bail Statutes
or Court Rules
As mentioned at the beginning of this paper, once a jurisdiction has done the
difficult work of articulating its bail/no bail, or release/detain dichotomy
based on the history, the law, the research, and the standards, the rest
includes merely creating an in-or-out framework so that nothing interferes
with either intentional release or detention. In short, once a jurisdiction has
decided whom to release and whom to detain, model laws simply make this
happen by using legal and evidence-based practices to achieve the
constitutionally valid purposes of bail and no bail.
Nevertheless, there are certain fundamental themes or principles that likely
should be included in any comprehensive bail scheme. The following list is
derived from many sources, including: the Pretrial Justice Institute’s Key
Features of Holistic Pretrial Justice Statutes and Court Rules;
476 Harvard
Law School’s Moving Beyond Money: A Primer on Bail Reform;
477 NIC’s
Fundamentals of Bail and Money as a Criminal Justice Stakeholder
papers;
478 the American Bar Association and National Association of Pretrial
Services Agencies Standards; the D.C. and federal release and detention
statutes; and conversations primarily with Alec Karakatsanis of Civil Rights
Corps, John Clark of the Pretrial Justice Institute, Mike Jones of the Pretrial
Justice Institute, Larry Schwartztol, of the Harvard Law School Criminal
Justice Policy Program, Claire Brooker, independent pretrial consultant, and
the Honorable Truman Morrison, III, Senior Judge on the District of
Columbia Superior Court.
1. Provisions articulating the state’s purposes and goals behind pretrial
release and detention, and definitions of key terms and phrases.
2. As a part of those goals, provisions expressly articulating a strong
presumption of release for all defendants and that no condition of
release – particularly a financial condition – shall cause detention.
3. Provisions favoring release on citation and summons over arrest and
arrest warrants, and expressing preferences of release through citation
for all misdemeanors and nonviolent felony offenses.

476 Key Features of Holistic Pretrial Justice Statutes and Court Rules (PJI, 2016), found at
https://university.pretrial.org/viewdocument/key-features-of-holi.
477 See Harvard Law School Primer, supra note 3.
478 See NIC Fundamentals, supra note 6; NIC Money, supra note 30.
199
4. Provisions allowing for evidence-based pretrial diversion of
appropriate defendants.
5. Provisions eliminating all financial conditions at bail, including
amounts on warrants.
6. Provisions allowing or mandating pretrial services agency functions
(assessment, recommendations, and supervision) based on the law and
the research.
7. Provisions articulating prompt first appearances.
8. Provisions giving defendants a meaningful right to counsel at first
appearance.
9. If not already in a constitution, release provisions, including
presumptions of release on a promise to appear; the use of least
restrictive and individualized conditions designed to provide
reasonable assurance of court appearance and public safety; various
factors to be used by judges relevant to the release decision; contents
of the release order; provisions articulating the procedure for dealing
with violations of conditions, including those violations that result in
the defendant being considered for pretrial detention; provisions
expressly encouraging or mandating the use of actuarial pretrial risk
assessment instruments for released defendants by favoring the
assessment over pure clinical assessment, but by balancing the tool
with other elements of risk relevant to flight and the danger we seek to
address; provisions encouraging or mandating the use of researchbased least restrictive conditions of release.
10. If not already in a constitution, detention provisions, including
provisions articulating the detention eligibility net, further limiting
process, and procedural due process hearing for detention; various
factors judges should use in making the detention determination using
principles articulated in this paper; other details made necessary by
the enabling language from the main right to release provision.
11. If not already in the constitution, the requirement that judges provide
written records of the reasons for imposing any and all limitations on
pretrial freedom, up to and including detention.
12. If not already in the constitution, provisions dealing with speedy trial,
periodic review of detained defendants, and with physically separating
defendants from sentenced offenders.
200
13. If not already in the constitution, provisions dealing with victims and
victim’s rights, so long as they do not interfere with defendant
rights.479
14. Provisions mandating data collection and performance measures by
all persons in the justice system to help assure that the underlying
purposes of bail are met as well as fostering conversations over the
proper context for pretrial release and detention within a state.
Conclusion
While this generation of bail reform is leading to change, even jurisdictions
desiring not to change nonetheless have an obligation to continually justify
their current release and detention schemes based on the law and the
research or face the probability of having them declared unlawful by the
courts. This paper provides a detailed justification for a proposed model for
release and detention in America, which is designed to follow the history,
law, research and national standards at bail while attempting to fix certain
longstanding problems that have plagued America’s desire to create a
rational and fair release and detention system. Importantly, it answers the
three fundamental questions associated with the pretrial phase of the
criminal case: (1) whom should we release?; (2) whom should we detain?;
and (3) how should we do it?
As demonstrated in this paper, the desire to move from a “charge-based”
system to one that is “risk-based” or “risk-informed,” while understandable,
is simply more complex than it outwardly appears. This paper attempts to
illuminate those complexities and to find justifiable solutions that balance
the variables found in such an undertaking. Jurisdictions might reasonably
disagree with the proposed model, but they cannot disagree with the need to
provide the same or similar justification for whatever model they ultimately
adopt or retain.

479 Because the person is un-convicted and at the center of a government prosecution on behalf of both
individual victims as well as all persons within the state, victim’s rights must never interfere with defendant
rights.
201
Appendix One
“For instance, now,” she went on, sticking a large piece of plaster on her
fingers as she spoke, “there’s the King’s Messenger. He’s in prison now,
being punished; and the trial doesn’t even begin till next Wednesday; and of
course the crime comes last of all.”
“Suppose he never commits the crime?” said Alice.
“That would be all the better, wouldn’t it?” the Queen said as she bound the
plaster round her finger with a bit of ribbon.
Lewis Carroll, Through the Looking Glass, at 88 (Harper & Bros. ed. 1902).
202
Appendix Two – A Hypothetical
John Quizzacious Public was driving down the street when he ran a red
light. He was stopped and the officer quickly realized that he was driving
without insurance. After using an actuarial pretrial risk assessment
instrument designed to help officers in the field, the officer informed John
that he scored as a “high risk,” and that, therefore, he would be taken to jail
and detained until his trial.
“Wait a minute,” John said. “I’m going to jail for a traffic violation?”
“Yes,” responded the officer. Our constitution allows detention of high risk
defendants and this tool indicates that you are high risk.”
“What does ‘high’ risk mean?”
“It means that you look like a group of similar defendants who failed to
show up for court or committed a crime while on pretrial release.”
“Well, I’m not one of those people,” John said. “I’ll come to court and stay
out of trouble.”
“The instrument doesn’t measure individual risk,” replied the officer, “so we
understand that you might come to court and stay out of trouble, but we can’t
really take that chance.”
“I’ve heard about these things,” John said. “But I heard they only use risk
factors; they don’t use any protective factors – like, having insurance or a
car that works.”
“Look, we use the tool they give us. Most don’t measure those things.”
“Who created this risk tool?” John asked.
“Researchers,” said the officer. “They decided who was risky and not risky
based on a quartile method, which divided up failures into quarters. Then a
group appointed by the Governor decided that the ‘high’ risk category didn’t
have enough people in it, so they changed that cutoff to include more
people.”
203
“So people can change these things whenever they want?”
“Oh sure. In fact, we’re thinking of meeting next week to shrink the ‘low
risk’ category a bit more. I think people are getting riskier, don’t you?”
“Does everyone have one of these risk tools?” John asked.
“Well, not everyone,” the officer replied, “but they use them around this
state, and they’re different wherever you go. Over in Kansas they still only
look at your risk if you commit some serious or violent crime. They call the
right to bail a right to release. Can you believe that?”
“Well, I remember seeing a case once, called Salerno, and it specifically said
that detention was okay because it was limited to a certain class of extremely
serious offenses.”
“That old case? Look, Mr. Big Brain, that case happened way back in 1987,
and nobody has really taken it seriously. Besides, if the Court saw that we
were detaining only high risk people, it would definitely find that to be a
much more rational way to do things than by looking at charge. Get with the
times, dude.”
Exasperated, John asked, “What, exactly, does this tool say I’m risky for?”
“Well, that’s an interesting question,” said the officer. “You could be risky
to commit murder, or risky to get another traffic ticket. You could be risky
to miss a court date, but we’ll never know if it’s willful or not. We don’t
really distinguish between types of risk.”
“That seems crazy. Isn’t that what you should be doing?”
“Well, yeah, over in Kentucky they have a violence ‘flag’ that tells them that
you are at an elevated risk for violence if they release you. That’s because
about 1% of high risk defendants will commit a violent offense if we let
them out. But Kentucky is still concerned about every crime, so just because
you don’t have a flag in Kentucky doesn’t mean you aren’t high risk. The
flag is more like a guarantee that you won’t get out. Plus, that flag is only
based on just a few cases. You see, it’s hard to create a tool to predict
something that simply doesn’t happen that much. Anyway, you aren’t in
Kentucky.”
204
“Look,” John said, “if 99% of defendants don’t commit a violent crime on
pretrial release, and if violent crime is what you really care about, by
releasing everyone you’ll be right 99% of the time. Does your risk
instrument do what well?”
“Don’t be a smart-ass, Professor Cranium. We’ve always had base rate
problems and nobody but you seems to care.”
“I’ve never, ever, skipped a court date in my life,” John said. “Does that tool
tell you that?”
“Well, no, this particular tool doesn’t even include prior FTAs on it. That’s a
long story, but these things are only as good as the data we put into them.
Oh, and remember, they only have risk factors in them.”
“What do you do with low risk people?”
“Well, the other day, I stopped as guy from strangling his wife. She was
really messed up – went the hospital in a coma. But I ran the assessment and
the guy was ‘low’ risk. He kept telling me he was going to kill her, but you
can’t ignore the risk assessment. So I gave him a summons. I don’t know
what happened after that.”
“Do high risk people always fail?”
“Oh, no. In fact, about 60% of high risk people will succeed if we let them
out. Also, if we let out high risk people and put a bit of supervision on them,
they’ll perform like medium risk people. It’s all kind of confusing. But it’s
like your snarky base rate comment. Nobody cares about the details.”
“So, if I’m more likely to succeed than fail, why do you lock me up?”
“Well, we can’t be sure you won’t be the one who fails. Everyone we’ve
locked up so far hasn’t failed, though, so we must be doing something
right.”
“So,” John asked, “when you release somebody and they fail, what do you
do?”
205
“We lock them up.”
“And when you detain people, they don’t fail, right?”
“Right.”
“That’s crazy. If someone fails on release, you say release didn’t work. And
you assume release won’t work for everyone you detain. Doesn’t that just
lead to more and more detention?”
“Oh, I don’t know. I saw a federal district recently that had an 80%
detention rate. But they had hardly any failures to appear or new crimes, so I
guess it evens out. Anyway, that reminds me that we’re building a new jail.
While you’re there, you can donate.”
“So if our constitution says I can be locked up for being high risk,” John
asked, “what in the world am I supposed to do to keep from being locked
up?”
“You need to try not to be high risk,” the officer replied. “You know, stop
being dangerous, and don’t do anything wrong while you’re high risk.”
“I never felt risky before,” said John. “What’s changed?”
“Well, I never thought people were all that risky either, until I saw that
everyone was risky on the tool. I mean, everybody is risky! Who knew?
Good thing we know now.”
“I noticed some of these other police keep stopping people and letting them
go,” John said. “What’s with that?”
“Well,” the officer replied, “it’s just like the old days with broken tail lights.
We’ve learned that if we stop a lot of cars and assess risk, we can take a lot
of dangerous people off the street. Once they’re in jail, it takes a while
before they get out. And then – here’s the beauty part – taking them to jail
actually makes them higher risk for the next time I get them. The whole
thing makes everyone safer.”
“I want to see my lawyer,” said John.
206
“Oh, you want your lawyer? Well, okay, we’ll get your lawyer, but he’ll just
tell you that the Supreme Court just recently said that ‘detention is the norm,
and pretrial release is the carefully limited exception.’ Now get in the car.”