Defining Flight Risk

Introduction

The number of low-risk defendants who spend time in pretrial detention in this country is staggering: “Every year, more than 11 million people move through America’s 3,100 local jails, many on low-level, non-violent misdemeanors.”1 According to a Department of Justice estimate, up to two-thirds of the members of this population pose “no significant risk to . . . the community” and “a low risk of flight.”2 In a study in New York City, Human Rights Watch documented similarly high rates of overuse of pretrial detention: more than 20 percent of pretrial detainees charged with misdemeanors were not ultimately convicted and over half of those who were convicted were not sentenced to incarceration.3 Studies show that nine out of ten pretrial felony detainees remain incarcerated because they cannot afford their bail,4 demonstrating the system’s inequities and inefficiencies.

The tragic death of Kalief Browder highlights layers of systemic dysfunction. After being arrested as a sixteen-year-old, Browder spent three years in pretrial detention at Rikers Island, mostly in solitary confinement, and was released only when the charges against him were eventually dropped.5 Browder’s pretrial detention was not justified by any public-safety claim,6 and given his family’s limited resources, he was not likely to flee the jurisdiction. Nevertheless, the judge set bail at a figure that Browder and his family could not afford ($3,000), and Browder was detained.7 He remained in jail because he refused to plead guilty (even to a deal for time served) to a crime he said he had not committed.8 Browder’s detention story is just one example of the routine phenomenon of wealth-based pretrial detention that cannot be justified as managing any cognizable or legitimate risk.9 When Browder’s story was first reported in 2014, it sparked public outrage and spurred efforts to reform pretrial detention policies.10 Tragically, Browder never recovered from his experience, and he committed suicide in 2015.11 His death has renewed and amplified calls for a range of pretrial reforms.12

Significant bail reforms are underway. Across the country, constitutional and statutory amendments, impact litigation, and community-based initiatives are forcing jurisdictions to change their pretrial practices, sometimes dramatically.13 Much of the focus in this active “third generation” of bail reform efforts is on judicial decisionmaking.14 Judges deciding whether to release or detain defendants before trial have broad discretion.15 They tend to overestimate the risks posed by defendants on release,16 and because judges are notoriously risk averse, they err on the side of detaining or otherwise overmanaging defendants.17

The proponents of risk-assessment tools assert that these tools address flaws in judicial decisionmaking by supplementing judicial decisionmaking, and countering judicial risk aversion, with objective science and data.18 On June 30, 2016, the Obama administration announced the Data-Driven Justice Initiative—a multifaceted, nationwide effort to reduce unnecessary pretrial detention.19 One of the initiative’s principal strategies is to “[u]se data-driven, validated, pre-trial risk assessment tools” to ensure that low-risk defendants are not detained before trial.20 The initiative endorsed a risk-assessment reform movement that has already gained significant momentum. There is widespread enthusiasm for the prospect of “moneyballing” pretrial decisionmaking.21 By the time the Data-Driven Justice Initiative was announced, jurisdictions across the country were already using a range of different risk-assessment tools: a federal tool; state-specific tools used in Colorado, Florida, Indiana, Ohio, and Virginia; and a free tool developed by the Laura and John Arnold Foundation that has been adopted in about thirty-eight jurisdictions across the country, including three states (New Jersey, Kentucky, and Arizona).22 The foundation has pledged to continue the initiative’s work even if the Trump administration discontinues it.23

An elusive issue, unsolved by past generations of bail reformers, threatens the new reform efforts’ success: ambiguity regarding the risks that judges who set money bail or order pretrial detention are trying to mitigate or avoid. Even as scholars, reformers, and practitioners are showing renewed interest in pretrial detention and bail,24 there is little focus on one central question: the appropriate meaning and role of what is often called “flight risk.”25 What judges, attorneys, and scholars frequently describe in shorthand terms as “flight risk” is defined in older statutes and in newer risk-assessment tools in significantly broader terms: the risk that a defendant will fail to appear for a future court date.

Scholars,26 judges,27 and legislative drafters28 often use flight and nonappearance interchangeably. But these terms are not coextensive. Flight risk is properly assigned to defendants who are expected to flee a jurisdiction. This is a small, and arguably shrinking,29 subcategory of a much larger group of defendants who pose risks of nonappearance.30

This Article calls for nuance in the definition of nonappearance and flight, both in actuarial risk-assessment tools and in bail reform efforts more broadly. Constitutional and statutory requirements demand precision about these distinctions. In other work, I have asserted that disentangling flight risk and nonappearance from dangerousness will be essential to successful bail reform.31 This Article builds on that contention by isolating the appearance-related pretrial risks that judges seek to measure and manage and then highlighting fundamental problems with the definition of those risks.

Clarifying these muddy risk descriptions requires taking a step back to delineate the harms posed by various forms of nonappearance. This Article proposes dividing the broad category of “nonappearing defendants” (or defendants who fail to appear) into three subcategories. The first subcategory of true flight comprises defendants who flee the jurisdiction. The other two subcategories are both local or nonflight forms of nonappearance: defendants who remain in the jurisdiction but actively and persistently avoid court, described as local absconders, and defendants who remain in the jurisdiction but whose failures to appear are more preventable in advance and less costly after the fact, termed low-cost nonappearances. These subcategories differ in nature and not merely by degree.32 Defendants in these subcategories impose distinct systemic costs and call for different types of supervision and management. The distinctions between these groups turn on the intent of the actor, the persistence of the nonappearance, the difficulty (for the jurisdiction) of locating him or her, and the specific types of pretrial interventions that might be appropriate to ensure appearance.33

Ideally, risk-assessment tools would help judges identify which defendants are likely to fall into these subcategories. Unfortunately, the tools currently available do not define, measure, or guide judges about how to manage these separate risks. Instead, current risk-assessment tools treat all nonappearances equally and produce risk numbers that do not adequately account for a court’s ability to manage and mitigate pretrial flight and nonappearance risk.34 Although current reform efforts are focused on identifying less intrusive and lower-cost interventions that can effectively manage pretrial risks, those efforts (including, in particular, the risk-assessment tools that have been developed) are hobbled by vague and overly general descriptions of nonappearance risk. A more nuanced understanding of these differences will be a key piece of broader efforts to reduce judicial reliance on pretrial detention and to mitigate the risks posed by defendants on release. Ensuring that practitioners and judges have a greater appreciation for these distinctions may also improve outcomes in individual cases.

This Article proceeds as follows: Part I outlines the problem of pretrial nonappearance, detailing the inconsistent descriptions of the problem, highlighting data deficits, and outlining costs. This Part also highlights the connection between pretrial nonappearance and the overwhelming backlogs of outstanding warrants (many of which are bench warrants for failures to appear for court) in jurisdictions across the country. Part II analyzes the governing legal framework, including constitutional protections against excessive pretrial restraint, federal and state statutory requirements that govern pretrial assessments of flight and other nonappearance risks, and judicial decisions applying those statutes. Part III evaluates how modern risk-assessment tools fit into broader reform efforts, how those tools predict nonappearance, and why they define and predict nonappearance only in vague and overly general terms.

Finally, Part IV outlines a new nonappearance taxonomy that divides the broad category of nonappearance into the three separate subcategories mentioned above: those who flee the jurisdiction, local absconders, and low-cost local nonappearances. At least some judges making pretrial-detention decisions have implicitly drawn some of these distinctions between true flight risk and other nonappearance risks, but an explicit and detailed account of these distinctions is overdue. Part IV closes by summarizing how this taxonomy advances the larger objective: identifying less intrusive and lower-cost interventions than those currently used in most jurisdictions, which can more efficiently and fairly manage the full range of nonappearance and flight risks.

I. Fugitives, Flight, and Nonappearance: Defining Problems and Risks

For centuries, judges have understood that—in order to guarantee that justice is administered and to promote efficiency in the process—they must ensure that criminal defendants will appear for future court appearances.35 Despite this long history, little attention has been devoted to defining flight risk and to distinguishing flight from other types of nonappearance.36

More-rigorous thinking about nonappearance is essential for several reasons. First, although concerns about predicting future offending behavior understandably dominate the current reform conversation, stakeholders also clearly care about the costs of nonappearance: a defendant’s history of nonappearance weighs heavily in actuarial risk calculations.37 Second, as release rates rise with the implementation of reform, rates of nonappearance will also rise. The sustainability of reform depends on maintaining acceptable appearance rates. Finally, as outlined in more detail in Part IV.C, efforts to better define the nature and causes of various forms of nonappearance force engagement with other fundamental questions about the criminal justice system, including questions about what we criminalize and who we arrest.

This Part contributes to the pretrial-release literature by identifying: (i) the mix of terms that are currently used (inconsistently) to describe various sorts of nonappearance problems; (ii) the data that are available about the scope of the problem of nonappearance; and (iii) the general costs of various forms of nonappearance, both for the system and for nonappearing defendants, their families, and their communities.

A. Misdescribing the Problem(s)

Legislators, judges, scholars, and reformers define and describe the risk that a defendant will not appear for future court appearances in different and inconsistent ways. To some extent, the different terminologies acknowledge that some nonappearances are more problematic than others; for example, a defendant who flees the jurisdiction is more costly to recover than one who fails to appear but stays local. Put in risk terms, then, a flight risk should be more serious than a nonflight risk.

Statutes frequently direct judges to predict whether a defendant will “appear” and to set release conditions that will ensure appearance.38 The federal statute and some state statutes refer to both flight risk and other forms of nonappearance risk, drawing at least an implicit distinction.39 Although several statutes highlight flight as a more “serious” problem, they provide no definitions.40 Most state statutes, like the federal statute, set out factors for judges to consider in evaluating nonappearance risk but do not identify which factors are specifically relevant to flight and which are relevant to other local (or nonflight) types of nonappearance.41

Yet judges making bail determinations use the term “flight risk” to refer to all nonappearance risks, whether or not the individual is actually likely to flee the jurisdiction.42 Scholars and reformers do the same.43 In contrast, the new risk-assessment tools that aid judges making risk calculations use “appearance” language but do not mention or measure flight risk.44 But using either “flight” or “failure to appear” (FTA) so expansively elides significant distinctions between true flight, local absconding, and low-cost forms of nonappearance that are discussed in Part IV.

Empirical studies of nonappearance introduce still more inconsistent terminologies, including, for example, frequent references to “fugitives.”45 Sometimes, the term “fugitive” is used, in keeping perhaps with more colloquial understandings, to identify an individual who has left the jurisdiction.46 At other times, being a fugitive is not contingent on any spatial or geographical movement. Instead, it turns on the passage of time: for instance, a fugitive may be someone who has failed to appear for more than one year.47 Although an imperfect proxy, this temporal distinction may reflect the intention or willfulness one might associate with a fugitive. Intention or willfulness, however, is not a universally accepted aspect of the definition of a “fugitive”—there are references elsewhere in the literature to “inadvertent” and “unintentional” fugitives.48 Nor is the label “fugitive” reserved for those who are actively sought by the government.49

Terminology, however, is only part of the problem. Even when labels are agreed on, there are major issues with the accuracy and significance of the data that are collected about nonappearances. What does it mean to have an FTA on your record? Different jurisdictions apply different standards for measuring and recording what qualifies as an FTA or nonappearance. In a 2011 Pretrial Justice Institute report, Cynthia Mamalian documented the inconsistencies:

In some jurisdictions, if the defendant fails to appear for court, he is immediately assigned a[n] FTA status. In other jurisdictions, if the defendant fails to appear, the family is called, the defendant is given another chance, and the defendant’s case is only considered a[n] FTA if a warrant is ultimately issued.50

If a defendant has an FTA on his record, that will have a significant impact on whether he will be released before trial and what conditions may be imposed if he is released.51 Prior FTAs have always been significant to judges as part of a mix of factors that might predict the likelihood of flight or other forms of nonappearance.52 In the risk-assessment tools that are increasingly being used across the country, prior FTAs are displacing other factors, becoming the primary determinant of a defendant’s nonappearance risk score.53

B. Data: How Many Missing Defendants?

Notwithstanding the ambiguity in terminology, some facts are evident from the available data. Estimates for the numbers of felony defendants who fail to appear for scheduled court appearances vary, but data from 2009 indicate that the vast majority (83 percent) of felony defendants who are released before trial appear for all scheduled court appearances.54 The remaining 17 percent missed at least one scheduled court appearance, with 13 percent (of the total number) returning to court within one year.55 Only 3 percent of all released felony defendants remained a “fugitive” after a year.56 These numbers represent a significant decline compared to earlier studies; in 1996, nearly one-fourth of all felony defendants had at least one nonappearance.57

Nonappearance rates for those charged with lower-level felonies and misdemeanors are typically higher than for defendants charged with higher-level felonies.58 In 2009, for example, “failure-to-appear rates were lowest for murder (5%) and rape (7%) defendants, and highest for those released after being charged with motor vehicle theft (28%).”59

Over time and in jurisdictions across the country, pretrial nonappearance rates have created significant backlogs of outstanding warrants.60 Bench warrants for defendants who fail to appear for court “often represent a large proportion of a jurisdiction’s open warrants.”61

Although data about open warrants are imperfect,62 they provide some perspective on the phenomenon of nonappearance. Information from the National Crime Information Center (NCIC) Wanted Person file indicates that on any given day in the United States, there are “over two million active criminal warrants.”63 In her 2016 dissent in Utah v Strieff,64 in which she observed that “[o]utstanding warrants are surprisingly common,”65 Justice Sonia Sotomayor cited state warrant data putting the total number of outstanding warrants much higher, at over 7.8 million.66 The discrepancy is partially driven by the fact that the state data include more misdemeanors and other types of warrants (including civil, traffic, and juvenile warrants) that are not included in the NCIC Wanted Person file.67

The bulk of the warrant backlog is for low-level offenses. Only 725,000 of the 7.8 million records cited by Sotomayor are for felonies68 and only 100,000 of those involve “serious violent crime.”69 Researchers have established that in some jurisdictions, up to 75 percent of warrants for FTA are for traffic offenses.70 Close examination of warrant backlogs in jurisdictions like Ferguson, Missouri, raise real questions about the incentives to generate warrants and how overuse of warrants affects communities.71 In Ferguson, the Department of Justice found that sixteen thousand people in a population of twenty-one thousand—more than 76 percent—had outstanding warrants, most for traffic offenses or other municipal-ordinance violations.72

Disproportionately large numbers of outstanding warrants for low-level offenses and infractions clearly reflect systemic dysfunction. Without more information about the front-end processes for generating bench warrants or the back-end processes for resolving them, warrant backlog data provide little in the way of illumination about whether nonappearance is the problem or the product of other problems, such as overcriminalization, the poverty of arrestees, or the difficulty of navigating the cumbersome pretrial process.73 The problems with existing data reinforce this Article’s central claim: putting all types of nonappearance and all bench warrants in the same bucket muddies these waters, making solutions harder to find and potentially obscuring the government’s contribution to the problem.

Concerns about the numbers of outstanding warrants and at-large fugitives are not new. The US Senate held hearings and issued a report on the national fugitive problem in 2000.74 But little has been done to measure and document the problem, let alone to resolve it, since.75 As Professor David Bierie has explained, “[m]ore than a decade later,” the problem is “essentially unchanged, with scholars lamenting ‘how little is known about the fugitive phenomenon—including attributes as simple as the actual volume of fugitives either per year or currently active in all justice systems in the United States.’”76 By bringing attention to the critical—but previously unexamined—definitional problems that must be addressed as prerequisites for successful analysis and reform, this Article provides the necessary foundation for successful reforms and includes a preliminary research agenda to address these significant data deficits.

C. The Costs of Nonappearance and Flight

Before moving into the legal framework that governs the pretrial process, we should translate some of these observations about the scale of the pretrial nonappearance problem into thoughts about costs.77 This Section addresses systemic costs, then shifts to the costs and impacts on nonappearing defendants, and finally identifies some ways that current practices are counterproductive.

Defendants who fail to appear for court impose costs on the system that are traditionally described in broad, sweeping terms. Across jurisdictions, the government has “a substantial interest in ensuring that persons accused of crimes are available for trials.”78 For more serious crimes, the interest is more compelling.79 Evading justice is described as an affront to every purpose of punishment:

A fugitive’s flight erodes the deterrent effect of sanctions by lowering the probability of catching the offender and by reducing the present value of punishment. Fugitive flight delays incapacitation, because criminals “on the lam” enjoy the opportunity to commit additional offenses. Their flight also denies society the opportunity to exact retribution for the offender’s crime.80

The judiciary has a related institutional interest in “safeguarding the integrity of the adjudication process.”81 High volumes of nonappearances may suggest to the public

that the courts are dysfunctional and unable to deliver justice—to victims or defendants. In short, a court system that cannot ensure that defendants will refrain from crime, attend court as required, and protect victims, witnesses, or jurors has difficulty protecting the integrity of the judicial process and ultimately its legitimacy.82

The costs imposed on the system by nonappearing defendants, however, vary with the nonappearance’s nature and persistence. Those who flee the jurisdiction impose different costs than those who remain in the jurisdiction.83 And persistent nonappearances are more costly than short-term nonappearances.84 The costs of managing nonappearance risks (that is, of preventing nonappearance) also vary. These differences drive this Article’s proposed division of nonappearance into subcategories, as discussed below.85

The final set of costs is the impacts on nonappearing defendants and their families and communities.86 Defendants with outstanding warrants may avoid “secur[ing] legitimate and stable employment” because of the fear of detection and arrest.87 They will also “have difficulty obtaining a driver’s license, [and] cannot legally obtain public benefits.”88

Criminologists also describe the direct impacts of outstanding bench warrants on a defendant’s physical and mental well being.89 In On the Run, her ethnography of life with the “Sixth Street Boys” in Philadelphia, Professor Alice Goffman details the cumulative impacts on the young men she observes of “dipping and dodging” to avoid police and outstanding warrants.90 As Goffman explains, in addition to closing off opportunities for education and legitimate employment, this is a lifestyle that severely strains relationships with friends and family.91

In one study of a years-long project (Fugitive Safe Surrender) to try to clear outstanding warrants and bring defendants back into the justice system, researchers surveyed defendants about why they turned themselves in voluntarily. The responses highlight many of the points made above:

The most common reason cited for why individuals surrendered voluntarily was because they “want to get their driver’s license,” as noted by nearly half of all respondents (47.1%), followed by “want to start over” (41.8%) and “fear of arrest” (39.4%). The next most common reasons for surrendering were “for my kids” (33.6%), “want to get a job” (33%), and “tired of running” (29.1%). Asked why they had not surrendered before today, nearly 60% said they did not have money to pay bail or fines, but others noted being afraid: “I was afraid of what would happen to me” (36.5%) and “I didn’t want to go to jail” (37.2%) or “I didn’t want to get arrested” (28%).92

These impacts on defendants translate into new community problems. Fear of additional punishment for failing to appear, including fees and fines, reinforces a defendant’s desire or need to avoid court.93 Even initially inadvertent nonappearances can quickly become a persistent phenomenon. When nonappearances become persistent, those defendants risk becoming modern “outlaws,” persons who have been “put outside the sphere of legal protection,” and for whom “crime becomes a natural source of income.”94 This, in turn, creates a criminogenic environment in the community “through a sort of cyclical regeneration of noncompliant individuals at the core of a criminal culture of resistance and disrespect.”95 Professor Alexander Tabarrok explains that “the logic that encourages asociality and discourages work, school, routine, and any interaction with the justice system [ ] pushes outlaws away from civil society when we most want logic to push them toward civil society.”96

Taken together, the data about nonappearances and bench warrant backlogs, and the discussion of the costs of nonappearances, reveal serious pathologies in the current system. These problems relate to how FTAs are logged, what they mean, and how they impact the future behavior of defendants. As outlined below, reforms that build on the existing system without addressing the cracks in this foundation are doomed to repeat current mistakes.

II. The Legal Framework

Ambiguity in the description and definition of nonappearance risks violates both constitutional and statutory provisions, which require a match between the means and ends of any pretrial restraint on liberty. The Constitution prohibits judges from imposing “excessive” restraints on liberty before trial,97 and federal and state statutes similarly limit judges to imposing the “least restrictive” set of conditions that are necessary to manage risks of nonappearance.98 Employing vague, overly general risk descriptions significantly increases the likelihood of unconstitutional and costly overregulation. The government needs to know the risks it is regulating in order to abide by constitutional and statutory requirements.

A. The Constitutional Mandate

A judge’s pretrial evaluation of nonappearance risk includes at least two steps. First, the judge evaluates whether a defendant poses risks of flight or other forms of nonappearance. The judge is then empowered by federal and state bail statutes to impose conditions of release (if necessary) to manage these risks. In extreme cases, judges can detain defendants who pose otherwise unmanageable risks.99

Judges traditionally managed flight and other forms of nonappearance risk by imposing financial conditions of release, also called money bail.100 When money bail is imposed, a defendant or his sureties are required to put up money or property to obtain the defendant’s release from custody; the money or property is returned if the defendant successfully appears for subsequent court proceedings.101 This use of bail as an incentive to appear traces its roots to the colonial era.102

Problems with judicial discretion in setting bail—and specifically with the imposition of “excessive bail”—are centuries old. The British Habeas Corpus Act of 1679103 established some protections around bail but left the amount of bail to the “discretions” of the magistrate.104 This created immediate problems: judges could ensure defendants’ detention “by deliberately setting bail so high that the defendants could not pay.”105 The English Bill of Rights of 1689106 addressed this pretextual use of bail by prohibiting “excessive bail.”107 The language from the English Bill of Rights became a model for the Eighth Amendment to the US Constitution, drafted a century later, which states: “Excessive bail shall not be required, nor excessive fines imposed, nor cruel and unusual punishments inflicted.”108

As Professor Samuel Wiseman has explained, we do not know much about the Framers’ intent regarding the term “excessive”:

The only known remark addressing the proposed Excessive Bail Clause came from Mr. Livermore in the House of Representatives as part of a comment on the Eighth Amendment as a whole . . . . “The clause seems to express a great deal of humanity, on which account I have no objection to it; but as it seems to have no meaning in it, I do not think it necessary. What is meant by the terms excessive bail? Who are to be the judges?”109

Unfortunately, the Supreme Court has given the Excessive Bail Clause very little attention; to date, it has provided only modest protection against pretrial detention.110 The Supreme Court first addressed the definition of “excessive” in Stack v Boyle,111 decided in 1951. As the Stack court explained:

Like the ancient practice of securing the oaths of responsible persons to stand as sureties for the accused, the modern practice of requiring a bail bond or the deposit of a sum of money subject to forfeiture serves as additional assurance of the presence of an accused. Bail set at a figure higher than an amount reasonably calculated to fulfill this purpose is “excessive” under the Eighth Amendment.112

Over thirty years later, in its 1987 decision in United States v Salerno,113 the Supreme Court upheld the constitutionality of the federal Bail Reform Act of 1984114 against a facial challenge to its novel preventive (danger-based) detention provisions.115 Although the Salerno decision focused principally on these provisions, its interpretation of “excessive” under the Eighth Amendment is broadly applicable to nonappearance risk as well. The Salerno Court addressed the spare text of the Bail Clause and explained that the Clause’s “only arguable substantive limitation . . . is that the Government’s proposed conditions of release or detention not be ‘excessive’ in light of the perceived evil.”116 The Court elaborated on how excessiveness ought to be calculated:

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. Thus, when the Government has admitted that its only interest is in preventing flight, bail must be set by a court at a sum designed to ensure that goal, and no more.117

The Court’s prior discussion in Stack indicated that this calculation had to be an individualized one: “Since the function of bail is limited, the fixing of bail for any individual defendant must be based upon standards relevant to the purpose of assuring the presence of that defendant.”118 Read together, these decisions support the claim that pretrial restrictions on liberty that are not tailored to the specific risk an arrestee presents are unconstitutionally “excessive.”119

This Article asserts that the inquiry into “excessiveness” requires more finely drawn distinctions between flight risk and other forms of nonappearance risk. The government must also be precise in proposing restrictions (detention or conditions of release) to manage those risks.120 Although lower courts quote these passages from Stack and Salerno,121 their embrace of the broad concept of excessiveness has not translated into meaningful limits on pretrial detention or money bail.122 This may be due, in part, to the “amorphous” nature of Salerno’s conception of excessiveness.123 The advent of tools that promise a shift away from intuitive risk estimates and toward purportedly more precise risk calculations may offer opportunities to argue for more finely drawn measures of excessiveness in ways that were previously unsuccessful.

B. Statutory Requirements

Federal and state statutes also constrain pretrial decisionmaking. This Section details relevant federal statutory law and summarizes state approaches, focusing on statutory definitions of nonappearance risks and statutory limits on judicially imposed conditions of release.

The Bail Reform Act of 1984 begins by directing a judge to release a defendant prior to trial unless the judge “determines that such release will not reasonably assure the appearance of the person as required or will endanger the safety of any other person or the community.”124 Language similar to the italicized text appears throughout the statute, which repeatedly tasks judges with evaluating and managing pretrial nonappearance risks.125 Many state statutes contain similar language.126

The Bail Reform Act’s detention provisions seem to draw an important distinction between nonappearance risk broadly and risk of flight more specifically, although it is unclear precisely what the drafters envisioned. Under the statute, if the court determines that there is a “serious risk that [the defendant] will flee,”127 the court may be able to detain the defendant until trial.128 Before ordering detention, the court is required to hold a detention hearing to determine whether any conditions of release can reasonably manage the risks presented.129 The Act does not appear to permit a detention hearing based solely on general nonappearance risk,130 and the rule is clear: a detention hearing must be held in order to detain a defendant until trial.131 The statute draws the distinction between flight and nonappearance only with respect to detention; it does not otherwise distinguish or define the terms.132

The Act limits judicial authority by generally requiring reasonable assessments of risks and by permitting only the “least restrictive” conditions of release that are necessary to assure that defendants will appear.133 Specifically, the Act authorizes judges to impose conditions on released defendants that those judges view as necessary to “reasonably assure the appearance” of defendants at future court proceedings,134 but limits judicial discretion in that judges are required to choose “the least restrictive further condition, or combination of conditions, that such judicial officer determines will reasonably assure the appearance of the person as required and the safety of any other person and the community.”135 Many state statutes echo this least-restrictive-conditions provision of the federal statute.136

C. Judicial Interpretations of Statutory Risk Factors

Judges have traditionally relied on a series of statutorily prescribed factors to predict pretrial risks.137 The statutes typically do not specify whether a particular factor is relevant to flight risk or to public-safety risk.138 Nor do they meaningfully or consistently distinguish between risks of nonappearance or flight as proposed in Part IV. Judges applying these statutory factors describe their task in both ways: as ensuring a defendant’s appearance and as preventing a defendant’s flight.139 Indeed, judges talk in terms of “flight risk” even in states where the statutes discuss only nonappearance.140 Because these statutory factors provide limited guidance and restraint, they give significant discretion to judges, who are imperfect agents.141 As jurisdictions contemplate legislative reforms and adopt risk-assessment tools, they must clarify the continuing role of these factors in guiding judicial decisionmaking. More specific and careful consideration of the factors’ predictive utility is overdue.

The following sections briefly review the traditional factors that have been relied on by judges as predictive of nonappearance or flight, organizing them into four broad categories: (i) factors that suggest incentives to flee the jurisdiction or to avoid court deliberately; (ii) those that suggest an ability to flee the jurisdiction; (iii) those that show a defendant’s connections or anchors to the jurisdiction (and thus discount flight risk); and (iv) those factors that, without suggesting flight from the jurisdiction, nevertheless raise concerns about the defendant’s reliability or trustworthiness (to return to court).

1. Incentives to flee the jurisdiction or avoid court.

Courts frequently cite the first two statutory factors listed in the federal bail statute—the seriousness of the offense and the weight of the evidence—as giving defendants incentives to flee the jurisdiction or otherwise to avoid court.142 In the taxonomy that follows in Part IV, these factors might be expected to predict flight or local absconding. This analysis assumes that defendants’ actual behavior matches their incentives; that is, that defendants with greater incentives actually flee or fail to appear at higher rates. As outlined below, several studies question the validity of those assumptions.

1. Offense seriousness.  The seriousness of the offense of arrest is the factor that prosecutors and judges most frequently cite when claiming that a defendant poses a serious flight risk. The argument is straightforward: because more serious charges carry heavier penalties, defendants have increased incentives to flee. Both federal143 and state144 courts frequently repeat this claim. Federal and state statutory presumptions in favor of detention for certain types of offenses also rely on this argument.145 Indeed, the seriousness of the alleged offense is the sole factor that determines money bail in jurisdictions with preset bail schedules.146

   Assuming that offense seriousness correlates to flight risk has intuitive appeal, but decades of bail studies challenge that claim.147 As noted earlier, defendants charged with more serious offenses like murder or rape do not, in fact, fail to appear at higher rates than those with lesser charges.148

   These studies conclude that other factors, such as employment, family ties, community reputation, and prior record of appearances, are better predictors of nonappearance.149 Although reformers have been successful in getting judges to pay attention to other factors over time,150 most judges still rely heavily on the charge.151

   In some cases, judges have also described a second link between FTA and the seriousness of the charged offense. Some judges, in addition to viewing offense seriousness as predictive of flight, believe that the seriousness of the offense represents the costs of an FTA.152 In other words, for more serious crimes, some judges are likely to impose higher bail or detain defendants at higher rates because they perceive an increased “justice” cost to the community if a defendant who has committed a serious crime flees the jurisdiction, avoiding possible conviction and punishment.153 While this argument is compelling, it does not alter the predictive value of offense seriousness. Instead, it suggests that for more serious offenses, judges will be more risk averse.

2. Weight of the evidence.  As the argument goes, the stronger the prosecutor’s case against the defendant (that is, the weight of the evidence), the more likely it is that the defendant will be convicted, either because she will plead guilty in the face of a strong case or a jury will be more likely to find her guilty. The greater the chance of conviction, the greater the incentive to flee. Here, again, this increased incentive to flee is viewed as increasing flight risk. As with offense seriousness, the perceived increased likelihood of guilt may also influence judges who will see the loss of the opportunity for justice as a factor if a guilty defendant flees.154

   Like the seriousness of the offense, the strength of the prosecution’s case has a long historical pedigree155 and is embedded in many federal and state statutes.156 Judges denying bail also frequently discuss the weight of the evidence. Although this factor could be relevant to any risk of nonappearance, judges who discuss this factor in setting high bail amounts (or who deny bail on this ground) frequently describe their concerns in terms of “flight.”157

   Scholars have long recognized the due process concerns that are implicated by setting conditions of release or denying release based on a pretrial assessment of the prosecution’s case.158 Professor Shima Baradaran Baughman explains that the effects of pretrial detention, both on the likelihood of conviction and the severity of the sentence that will be imposed, can erode the presumption of innocence.159

2. Ability to flee the jurisdiction.

Many statutes include factors that seek to predict which defendants are true flight risks. Ideally, judges would focus on those defendants who are likely to flee and not simply those who are able to flee.160 Because it is more difficult to ascertain or predict a defendant’s inclinations, however, judges making flight risk predictions often focus on a defendant’s ability to flee.161 The distinction is subtle. A defendant’s intent to flee might be inferred from prior flight or a lack of cooperation with authorities, which are discussed below as reliability or trustworthiness factors.162 Judges may also look for evidence of preparation to leave the jurisdiction.163 Predicting a defendant’s ability to flee, however, generally involves analyzing her ties to other jurisdictions and her resources, both of which are discussed briefly below.

In a 1978 decision granting bail pending appeal to Truong Dinh Hung (who was already convicted of espionage), Justice William Brennan emphasized this difference between ability to flee and intention to flee.164 The district court revoked Truong’s bail after his conviction based on his Vietnamese citizenship and family ties, his contacts with the Vietnamese ambassador in Paris (with whom he had exchanged classified information), and his lack of permanent residence in the United States.165 The Fourth Circuit affirmed.166 Writing as the circuit justice for the Fourth Circuit, Brennan reversed, citing Truong’s domestic ties and numerous character references. He explained that even if the considerations cited by the lower courts “suggest opportunities for flight, they hardly establish any inclination on the part of applicant to flee.”167 As we work to develop and refine our flight risk predictions, we should evaluate how well a defendant’s ability to flee meaningfully predicts his intention to flee.

1. Ties outside the jurisdiction.  Defendants with significant ties outside the jurisdiction are traditionally viewed as more likely, or at least more able, to flee.168 Courts evaluating ties to another jurisdiction consider: (i) both family and financial connections;169 (ii) a defendant’s travel history;170 and (iii) the jurisdiction’s extradition practices.171

   This factor, of course, is entirely flight focused. Foreign ties may be predictive of whether a defendant remains in the jurisdiction or flees, but for defendants who remain in the jurisdiction, foreign ties are irrelevant to the likelihood that the defendants will appear for court, except to the extent that they may correlate with language or cultural barriers to information about court processes or schedules.

2. Resources.  Courts frequently analyze a defendant’s resources in evaluating flight risk.172 In short, courts view wealthier defendants as more able to flee the jurisdiction and to sustain their flight.

   In a number of controversial, high-profile cases, courts have ordered wealthy defendants to be confined in their homes prior to trial (as opposed to ordering them to be detained in jail facilities) on the condition that these defendants (or their families) pay for twenty-four-hour private security.173 In a 2016 news article describing these self-funded “gilded cage” arrangements, Judge Jed Rakoff (who nevertheless ordered a defendant’s release on these conditions) was described as having “acknowledged concerns that such an arrangement . . . gave people of means ‘an opportunity for release that poorer people could never obtain.’”174

   The vast majority of defendants, whose resources fall far short of those of the defendants just described, are on the other end of the spectrum. While these defendants may pose risks of nonappearance,175 their socioeconomic status makes it unlikely that they could flee from the jurisdiction. Successful flight from the jurisdiction suggests access to networks and resources that are not part of the equation for the vast majority of nonappearing defendants.

   A federal district-court judge in Detroit recently summarized this distinction succinctly. In rejecting the Government’s assertion that the defendant was a flight risk, the judge explained:

   The Court disagrees. Defendant has lived in the community his entire life, was employed for over a year and living with his girlfriend and their infant daughter until his arrest. . . . Contrary to the government’s view, defendant is not a risk of flight. Offenders like defendant almost never flee; they have nowhere to go.176

3. Anchors to the jurisdiction.

Federal and state statutes also include a series of factors that function as anchors to the jurisdiction, reducing the risk that a defendant will flee the jurisdiction. These include a defendant’s family responsibilities, community ties, employment or educational commitments, citizenship, and length of residency.177 While these factors have traditionally been discussed by both federal and state courts in terms of flight,178 they are also relevant to predicting and managing the categories of local nonappearances that are included in the taxonomy.

Connections to the jurisdiction operate in two ways. First, courts view them as anchors to the jurisdiction, creating incentives for released defendants to stay. Second, they make defendants remaining in the jurisdiction easier to locate, either before (for reminders) or after a scheduled appearance.179

4. Reliability and trustworthiness.

The final set of statutory factors reflects a court’s assessment of the defendant’s reliability or trustworthiness to show up for court. These factors are relevant to all of the categories of nonappearance proposed in Part IV.

1. Prior record (defaults, nonappearances, or cooperation).  Most federal and state statutes direct courts to consider evidence of a defendant’s past “record concerning appearance at court proceedings” when weighing pretrial risks.180 Nearly seventy years ago in Stack, Justice Robert Jackson described this information as relevant to a defendant’s “trustworthiness to appear for trial.”181 As Jackson explained, this evaluation should include positive and negative evidence: “Each accused is entitled to any benefits due to his good record, and misdeeds or a bad record should prejudice only those who are guilty of them.”182 In appropriate cases, courts have also considered a related measure: a defendant’s prior cooperation with authorities.183

   Federal and state courts applying this factor describe it as predictive of “flight”184 and “nonappearance risk.”185 Particularly when defendants have multiple FTAs, courts and risk-assessment tools give this factor substantial weight.186

2. Substance abuse.  In some cases, judges have also used a defendant’s drug or alcohol abuse, another statutory factor, as evidence that the defendant is a greater risk of nonappearance.187 Occasionally, courts note the lack of such problems as reducing the risk of nonappearance.188

* * *

Analyzed collectively, the statutory factors that guide judges’ discretion pose several problems. First, as noted above, further work must be done to confirm the factors’ predictive utility. Second, the statutes themselves provide little direction to judges about which factors are relevant to which pretrial risks and how to weigh the presence of multiple factors. As a result, judges have broad discretion, which they use to overmanage pretrial risks.189 For these reasons, jurisdictions across the country have been anxious to modernize and refine the process of pretrial risk prediction. The products of those efforts are the focus of Part III.

III. Predicting Nonappearance: Modern Risk-Assessment Tools

Risk-assessment tools that employ “rigorous, scientific, data-driven” analyses promise a move away from unbridled judicial discretion and from problematic judicial reliance on “gut and intuition.”190 But these tools have serious issues relating to the calculation of nonappearance risk.191

First, as I have argued previously, many tools merge nonappearance and public-safety risks into one “pretrial failure” risk measurement despite constitutional, statutory, and policy arguments for measuring those risks separately.192 Second, as detailed here and in Part IV.B, the tools predict nonappearance risk in only its broadest form. Thus, these tools ignore both long-standing doctrinal and statutory emphasis on concerns about flight risk and clear, practical policy needs for a more nuanced understanding of the problems of nonappearance. Finally, these risk-assessment tools measure only this very broadly defined nonappearance risk in the absence of any court intervention. Judges, however, need guidance and feedback about how to manage flight risk using existing tools.193 Most statutes can be interpreted as requiring this sort of inquiry. A static nonappearance calculation that does not incorporate various conditions of release, then, is of limited utility.

In the following sections, this Article (i) briefly reviews the connection between risk assessment and broader bail reform efforts; (ii) examines how the tools define risks of nonappearance and validate outcomes; and (iii) evaluates why the tools have been inattentive to distinctions between flight and other forms of nonappearance.

A. Bail Reform, Generally

In a 2013 report describing the impetus for its efforts to develop a national pretrial risk-assessment tool, the Laura and John Arnold Foundation explained that, in jurisdictions that do not rely on risk assessment, “too many high-risk defendants go free, and too many low-risk defendants remain locked up for long periods.”194 In June 2016, the Obama administration announced the Data-Driven Justice Initiative—with increased use of these tools as one of its key strategies—reflecting a growing consensus that risk-assessment tools should more effectively identify low-risk defendants who should be released before trial.195 As the tools become more economical, more jurisdictions are adopting them.196

The rise of risk assessment has occurred alongside another pretrial reform agenda: the effort to end reliance on money bail as a means of managing pretrial risk. For decades, and despite legislative reforms, judges have transformed bail from a condition of release to a predicate for detention.

Reformers have focused on eliminating money bail or, alternatively, radically changing how bail amounts are set. New Jersey and New Mexico have enacted comprehensive constitutional and legislative changes, shifting away from money bail.197 Connecticut passed legislation in June 2017 that significantly limits the use of money bail in misdemeanor cases,198 and, at the time of this writing, Governor Andrew Cuomo is advocating for comparable changes in New York.199 In February 2017, the Maryland Court of Appeals approved changes to the Maryland Rules of Procedure intended to reduce (but not eliminate) reliance on money bail by requiring judges to consider defendants’ ability to pay before setting bail.200 Class-action plaintiffs have, with support from the Obama administration’s Department of Justice, forced reforms in jurisdictions across the country by bringing successful challenges to existing money-bail systems.201 Reformers have also sought to change who pays bail, advocating for the end of commercial bail202 and promoting the use of community bail funds.203

The significant investment of time, resources, and political capital in both of these efforts—increasing the use of risk-assessment tools and reducing reliance on money bail—has been cause for optimism that what is being described as a “third generation,” or “third wave,” of bail reform might achieve meaningful change.204 In most places, these reform efforts are paired, with money bail schemes giving way to risk assessment.

B. Defining and Describing Risks of Nonappearance

Although the tools being used in jurisdictions across the country vary from each other in interesting ways, they have a few important common features. In addition to gauging a particular defendant’s “danger to the community,” each tool endeavors “to identify the likelihood of failure to appear in court.”205 No tool mentions or measures “flight risk.”206

1. Risk factors.

The tools use between seven and fifteen factors,207 but most tools include the factors found in the federal variant:

1. “charges pending against the defendant at the time of arrest”
2. “number of prior misdemeanor arrests”
3. “number of prior felony arrests”
4. “number of prior failures to appear”
5. “whether the defendant was employed at the time of the arrest”
6. “defendant’s residency status”
7. “whether the defendant suffered from substance abuse problems”
8. “nature of the primary charge”
9. “whether the primary charge was a misdemeanor or a felony”208

Other tools incorporate additional risk factors, including any prior violent convictions, whether the defendant was previously incarcerated, whether a defendant has a working phone, and demographic indicators like the defendant’s age, mental health, marital status, citizenship, and education.209

While some risk factors overlap with the statutory factors discussed in Part II.C, there is one important difference. Legislators incorporating risk factors into bail statutes attempted to identify factors with a causal relationship to nonappearance—either factors that provided incentives to flee or factors anchoring defendants to the jurisdiction. Risk-assessment tools, however, generally do not endeavor to identify what causes nonappearance (or recidivism). Instead they seek to identify “what other [correlative] factors tend to be present” when nonappearance (or recidivism) occurs.210

Studies suggest that the most commonly used tools have similar “predictive validity.”211 The factors that most current tools use also “resemble” those highlighted in earlier generations’ risk-prediction studies.212 Many of the same factors appear in federal and state statutes.213

Risk-assessment tools give a defendant a risk score based on the defendant’s risk-factor scores (which are weighted differently according to their predictive value) and assign the defendant to a particular risk category. Most tools do not specify which factors predict nonappearance. The Public Safety Assessment (PSA) tool (discussed in the next Section) is an exception to that rule.

2. Shifting away from interviews.

By 2015, less than 10 percent of jurisdictions used actuarial risk-assessment tools, but that number is rising.214 The cost of administering some tools is a barrier to widespread adoption and, as a result, developing less expensive tools is a priority for those who see risk assessment as the key to bail reform.215 The PSA tool developed by the Laura and John Arnold Foundation was designed to respond to those cost concerns.216 The PSA does not rely on an interview with a defendant, which makes it much less expensive for jurisdictions to administer (and thus to adopt).217 The PSA also shifts entirely to static risk factors (that is, fixed attributes) and away from dynamic risk factors (that can change and can be the focus of judicial intervention).218

The PSA improves on other tools in several ways. First, it separately predicts nonappearance and future criminal activity, avoiding a problem that plagues other tools.219 Second, it predicts future violence separately (“new violent criminal activity”) from the much broader category of “new criminal activity.”220 Third, the foundation is more transparent than some proprietary vendors about how its tool operates.221 Finally, the PSA is free for jurisdictions to obtain (and as noted above, it is less expensive to administer).222 As of June 2015, the PSA was being used in thirty-eight US jurisdictions, including three states: Arizona, Kentucky, and New Jersey.223

Because it does not rely on a defendant interview, the PSA does not include many of the traditional statutory factors that have been incorporated into other risk-assessment tools, including factors like family and community ties, employment, and residency status.224 Instead, the PSA analyzes factors that can be ascertained from the defendant’s record, including:

1. the defendant’s age
2. whether the current offense is violent
3. whether the person has a pending charge at the time of arrest
4. whether the person has a prior misdemeanor conviction
5. whether the person has a prior felony conviction
6. whether the person has a prior conviction for a violent crime
7. whether the person failed to appear at a pretrial hearing in the last two years
8. whether the person failed to appear at a pretrial hearing more than two years ago
9. whether the person has previously been sentenced to incarceration225

Four of these factors are used to predict a defendant’s nonappearance risk:226

1. a pending charge (misdemeanor or felony) earns a defendant one point
2. a prior misdemeanor or felony conviction earns a defendant one point
3. a prior FTA that is more than two years old earns a defendant one point
4. an FTA within the last two years earns two points; for two or more FTAs in the last two years, the defendant earns four points

These points are tallied to determine where the defendant will be placed on the PSA’s six-point FTA scale. The scale ranges from one, the lowest risk level, to six, the highest. As the foundation has explained, “the likelihood of a negative pretrial outcome increases with each successive point on the scale.”227

Unfortunately, the PSA predicts future nonappearance in only its broadest form. There is no discussion (or prediction) of flight risk and no distinction made between types of nonappearance.228

All the risk-assessment tools currently being used rely heavily on prior FTAs to generate predictions of future nonappearance.229 In general, a defendant’s FTA at a court appearance—even if it does not involve willful nonappearance or flight from the jurisdiction—will give the defendant a recorded FTA.230 Of all the factors, this will significantly elevate the defendant’s risk of a future FTA, particularly if the prior FTA is more recent.231 The validation of the tools also overemphasizes nonappearance, broadly defined: in order for a defendant to be deemed a pretrial success, he must appear for all court dates.232 As outlined in more detail in Part IV, merely evaluating broad nonappearance risk reinforces and exacerbates existing overdetention problems.

C. Explaining the Omission of Flight

What explains the omission of true flight risk from pretrial risk-assessment tools? Why have the creators of the tools failed to distinguish between flight and local (nonflight) forms of nonappearance, or between various types of local nonappearance?233

Historical inattention to these distinctions, and the heritage of confused, inconsistent terminology may be part of the problem.234 Although some judges and some statutes have tried to draw meaningful distinctions, the general historical picture is one of a lack of attention to this pretrial detail. To be fair, conceptions of nonappearance have become more nuanced as the number of pretrial court appearances and the overall delay in the pretrial process have increased over time.

Reformers’ preoccupation with perfecting the definition and prediction of dangerousness may have led them to neglect flight and other forms of nonappearance risk.235 Pretrial risk assessments have been adapted from tools used in other criminal justice contexts like corrections, in which the risks of a particular defendant’s future violence and recidivism are primary concerns, and nonappearance is not part of the risk calculus.236

Another explanation is that the tools may not be as objective and unbiased as they claim. The creators of these risk-assessment tools (including some for-profit companies) are concerned with reducing administrative costs so that the tools can be widely employed. As a result, they have an obvious appetite for and bias in favor of cheap data.237 Logging an FTA, in some jurisdictions, is the equivalent of taking attendance, and that bit of data is readily available.

In addition, broadly defined risk categories are easier to predict, allowing creators to claim higher success rates. In other words, because the tools promise only to predict nonappearance broadly, they can claim greater success than if the tools purported to predict the narrower and more serious categories of risk (flight risks and local absconding risks) defined in the next Section. Certainly narrower categories are more difficult to accurately predict, but the greater utility of those predictions would justify some increase in costs.238

The problem with using readily available data, however, is that it may simply import existing problems into new contexts.239 Even worse, it may create new problems by falsely inflating the seriousness of mere nonappearance. As the tools become cemented in judicial decisionmaking and pretrial policymaking, these problems may become entrenched features. Scholars are increasingly recognizing the need for actuarial tools and for the next generation of algorithms to incorporate, from the outset, accountability for “legal and policy objectives.”240

Finally, as noted above, this flight-focused critique is nested within a broader literature that challenges the current use of risk-assessment tools.241 Scholars have raised important concerns about the use of actuarial tools, or in some contexts more sophisticated machine-learning algorithms, to predict risk in the criminal justice system.242 The literature includes thorough engagement with the basic underlying questions whether and when it is appropriate to rely on “statistically sound but nonuniversal generalizations” to draw conclusions about the future behavior of members of particular groups.243 Others have focused on the due process and equal protection problems that risk-assessment tools present.244

Even though proponents of the PSA (and other tools) claim to have avoided the selection of discriminatory factors,245 heavy reliance on prior nonappearance has a direct and clear impact on poorer defendants who may lack transportation or childcare or cannot miss employment.246 In addition, because these tools heavily weigh prior criminal records, they will disadvantage those who live in heavily policed communities and who will be picked up and prosecuted for misdemeanors that fly under the radar in more affluent neighborhoods.247

IV. A New Taxonomy: Defining Flight, Distinguishing Nonappearance

Before we can identify what risks we seek to predict and prevent, we must specify the categories of harms we seek to avoid. We still know far too little about who fails to appear, why they fail to appear, and what can be done to remedy that.248 This Article outlines what ought to be measured so that appropriate data can be gathered and employed to refine actuarial risk-assessment tools, to improve judicial management of pretrial risks, and to highlight other bail reform priorities.

It bears emphasizing here that the task for judges at the pretrial stage is one of risk management. Their objective is to ensure a defendant’s appearance at future court dates. They must calibrate risks of nonappearance and employ available conditions of release (or in extreme cases, deny release) to manage and mitigate those risks. Precision about risk definitions, then, is required so that judges can make appropriate choices from a range of interventions that will manage the risks presented.

The following sections propose a taxonomy of nonappearance before outlining research and reform priorities. First, the category of “true flight,” which includes defendants who leave the jurisdiction, is treated separately from “local” nonappearances. Within the remaining category of “local” nonappearances, further distinctions are suggested to isolate a more serious and costly category of “local absconding” from what is termed “low-cost nonappearing.”

A. True Flight

The first subcategory of nonappearances—those defendants who flee the jurisdiction of arrest—is likely the narrowest of the subcategories. Because it is based simply on geographic movements, this subcategory is also the easiest to define.249 This Section details why this group of defendants merits its own subcategory (despite important similarities to the local absconders discussed below), the issues that presently exist with respect to collecting data about flight, and the best ways to remedy those data deficits to make useful and reliable flight predictions.

There are several reasons why we must identify “true flight risks”—that is, those defendants who pose a high risk of leaving the jurisdiction—and isolate them from other individuals who pose risks of what will be termed “local nonappearances” in the sections that follow. As outlined above, treating all nonappearances the same ignores the long-standing statutory and doctrinal focus on “flight risk.”250

The judicial, legislative, and scholarly emphasis on flight risk is not a semantic quirk. It reflects awareness that flight from the jurisdiction imposes special costs. Although technology makes it increasingly easy to locate defendants who flee the jurisdiction, administrative headaches and financial realities often make it difficult to return defendants to the jurisdiction. Notations in the NCIC database indicate whether a state or local jurisdiction is willing to pay for extradition and under what circumstances. Two factors, among others, determine the jurisdiction’s willingness: the distance a defendant travels from the jurisdiction and the seriousness of the offense.251 Many jurisdictions stop trying to retrieve defendants who have fled the jurisdiction, particularly if they are not found in a border state or if they have been charged with a less serious offense.252 For defendants accused of lower-level offenses, it is plausible that the community may not view flight as particularly problematic and, as noted below, the community may elect to spend few resources attempting to locate such defendants. Here again, judges should be encouraged to adjust their risk tolerance, as detaining low-level defendants to prevent flight is a poor allocation of resources.253

Tools should also measure flight risk apart from other nonappearance risks because the possibility of flight suggests different judicial interventions. For example, judges may impose travel restrictions or confiscate travel documents from defendants who pose flight risks. Judges faced with defendants who are flight risks may (particularly for defendants charged with more serious offenses) be able to justify using more aggressive forms of community supervision, GPS monitoring, house arrest, or, in extreme cases, detention. Given its high costs (both for the individuals being jailed and for the community), detention should be reserved for those who cannot be prevented or dissuaded from leaving the jurisdiction using less intrusive interventions.254 Judges use a range of tools to prevent released defendants from fleeing the jurisdiction and to incentivize them to return to court, including imposing financial conditions, supervision conditions, travel restrictions, and, in some cases, the use of electronic monitoring.255

Typically, in order to flee, defendants need resources.256 This category of true flight, then, will likely encompass wealthier defendants than other nonappearing defendants (who remain in the jurisdiction). That resource divide may be useful both for predictive sorting purposes and for developing means of preventing flight. Although broad, indiscriminate use of money bail often leads to detention for indigent defendants, scholars should study and evaluate whether selectively applying financial conditions can effectively discourage flight for released defendants who have resources.257

Over time, technology and improved interjurisdictional coordination have diminished the prospect of successful flight.258 The FBI’s NCIC database facilitates rapid information sharing.259 For high-profile suspects, shows like America’s Most Wanted significantly increase the likelihood of apprehension.260 For other defendants, internet mug shots make it difficult to hide.261 Professor Wiseman explains that “technological advances in tracking and monitoring defendants” have meant that “there is no longer as high a likelihood of avoiding conviction by escaping across a state or county line, as the police will eventually detect and track down the defendant.”262

Unfortunately, although courts and commentators frequently describe nonappearance risk in terms of “flight,” the data rarely segregate flight from other FTAs. There are reports and studies that purport to isolate data about “fugitives” from other nonappearance data, but those reports and studies generally do not use the term “fugitives” to refer to fleeing defendants.263

Some studies of flight behavior show how we can gather more data about defendants who flee the jurisdiction. Although limited, these studies offer preliminary insights into where fleeing defendants may be most likely to go and how they behave. For example, based on a recent study of alleged sex offenders who unlawfully crossed federal, state, or tribal lines, for example, Professor Bierie and Paul Detar found that “fugitives who live alone choose familiar areas, and those who go to an unfamiliar location tend to reside with friends, partners, or family.”264 Bierie and Detar concluded that “fugitives often need some connection to the life they had prior to their warrant—few can truly disappear to seek a new life wholly unconnected with their prior one.”265 Bierie and Detar’s study is one of the few to study “flight behavior,” including “the distance offenders traveled, whether they lived alone or with others at capture, and whether they were arrested in a community they were familiar with (e.g., a city they had lived in before).”266

B. The Nuances of “Local” Nonappearance

Even after isolating “local” nonappearances from flight, further distinctions ought to be drawn between two types of local nonappearance: what this Article terms “low-cost nonappearance” and “local absconding.” The distinctions between these categories turn on the intent of defendants,267 the persistence or duration of their nonappearances, the preventability of their FTAs, and the costs imposed by their nonappearances—that is, the costs of returning these defendants to court through rearrest or other means.268

1. Low-cost nonappearance.

The problem with an overly broad definition of nonappearance is that it lumps comparatively minor forms of nonappearance together with much more serious and costly nonappearance problems. This conflation raises the question: How do these purportedly minor nonappearances differ from what is termed “local absconding” below? The key differences, outlined here, turn on: (i) why defendants who fall into this category fail to appear; (ii) what interventions might improve their appearance rates; and (iii) the cost of returning these defendants to court if they do fail to appear. All of these distinctions focus on the differences in the costs that these nonappearances impose on the system.

1. Explanations/purpose.  These low-cost nonappearances include defendants who fail to appear for a range of different reasons, including: being unaware of or forgetting the date of the court appearance (which might reflect either ineffective notice by the court or poor calendar management by the defendant); illness or other unforeseen personal emergencies; external logistical challenges including employment conflicts, childcare issues, or lack of transportation; confusion or ignorance about the process or a general lack of capacity to navigate the process (this may reflect the complexity of the system and/or the defendant’s cognitive limitations); fear of punishment relating to the pending charge; or lacking the funds to pay fines and fees that are owed at the courthouse.269 Judges have different means of addressing each of these underlying causes, and it is easy to anticipate that a judge trying to manage this type of nonappearance risk might approach the endeavor with a checklist to determine what interventions to employ.

   Criminologists are increasingly drawing similar distinctions between who they describe as “active flaunters” and “inadvertent absconders.”270 As Professor John Goldkamp explains, defendants who fail to appear for court include the “active flaunters as well as inadvertent absconders who did not miss their court requirements through intentional actions—rather, they may have been confused or lost in the courts.”271

   Professors Daniel Flannery and Jeff Kretschmar elaborate on the same idea, explaining that at least some nonappearing defendants “may just not have the capacity or competence to show up at court hearings at the required place and time.”272 When individuals who fail to appear inadvertently or for other nonwillful reasons are lumped in with other “fugitives” and “flight risks,” there is clear potential for mismanagement of the risks that are actually present. The risk-assessment tools described above make that precise mistake.

   While the “inadvertent” nonappearances envisioned by Goldkamp are certainly part of the category proposed here, that label is too narrow. As the examples provided above make clear, some of the defendants who fall into the low-cost category proposed here are deliberate (and not inadvertent) nonappearances.

   One might be tempted to label the entire category “excusable” nonappearances, but that term is problematic because it may not align with current court practices regarding excusing FTAs. As noted in Part IV.C below, however, adjusting current court practices is a key part of this Article’s proposed reform agenda.

2. Preventability.  Appearance rates are distinctively “malleable” for defendants in this subcategory.273 For example, studies show that reminding defendants or their families of court dates can significantly reduce FTAs.274 In a 2011 study, researchers reviewed studies that evaluated various court reminder systems for low-level defendants who received citations.275 Another study found that, by providing reminders to those defendants, the court system was able to reduce the FTA rate “from 25 percent in the control group to six percent in the reminder group when the caller spoke directly to the defendant, 15 percent when a message was left with another person, and 21 percent when a message was left on an answering service.”276

   Other studies have also reported “immediate and dramatic improvements” with implementing similar reminder and notification procedures.277 Some jurisdictions also “take advantage of each and every contact with [released] defendants to remind them of their obligations,” by ensuring, for example, that staff at drug testing facilities also remind defendants when required court appearances are looming.278 Administrators of one notification program in San Mateo, California, report that most FTAs are not willful and that logistical and practical issues underlie the majority of FTAs (for example, lost paperwork, lack of contact information, and fear to ask questions, among others).279

   Jurisdictions also report improved appearance rates when community organizations and networks are engaged in creative and novel ways to help ensure that defendants are aware of appearance dates and able to appear when required.280 A startup company called Uptrust is currently marketing to county governments a “behavioral science-driven SMS reminder service” that will increase court appearance rates and therefore lead to reductions in the use of pretrial detention to manage nonappearance risk.281

   There are other tools that judges can use to discourage or prevent FTAs, including improving access to high-quality substance abuse treatment and improving pretrial services support. In Part IV.C, various systemic changes are proposed, all of which will improve appearance rates among this group of defendants.

   Pretrial detention and more aggressive forms of community supervision are unnecessary for defendants who could be nudged back to court on the appointed day with a simple and inexpensive reminder.

3. Low costs.  Even if some of these nonappearances cannot be prevented, they can be viewed as low-cost events for at least two reasons. First, the injury or harm that these individuals inflict on the system is less serious. While those who flee or abscond “harm the public by preventing the operation of its criminal justice system,” defendants who fall into the category of low-cost nonappearances impose shorter-term administrative burdens.282 Thinking in terms of the “justice costs” described above, these defendants, while clearly “annoying” contributors to an already inefficient system, are not likely to evade justice permanently.283

   In addition, returning a nonappearing defendant to court is likely to be a relatively low-cost proposition if that person is not actively and persistently avoiding the courthouse, like the true flight examples described above and the local absconders described below. First, courts should attempt to resolve FTAs without penalties by simply advising the suspect of the missed court date and urging a prompt return to court.284 As noted above, some jurisdictions already take this approach by not formally recording an FTA until steps have been taken to alert a defendant of the missed appearance and the no-cost opportunity to correct it.285

   Even if a phone call does not work, if a nonappearing defendant can be readily located and approached, then pretrial detention is an unnecessary, expensive, and oppressive way to prevent that type of nonappearance.286 Policymakers looking to reduce pretrial incarceration should weigh the costs and availability of other mechanisms for ensuring that these defendants return to court.287

   Of course, even short-term nonappearances trigger nuisance costs. These include the administrative costs of rescheduling court dates and the wasted time of court personnel and attorneys when court dates are rescheduled.288 Once a defendant fails to appear, the government may also incur some expense locating and rearresting defendants.289 Time spent locating and rearresting defendants who fail to appear on their own volition imposes opportunity costs, as well. Those officers could be spending that time on other investigations and endeavors.

   Although jurisdictions with arrest warrant backlogs complain of an inability to serve outstanding warrants due to a “lack of manpower,”290 that complaint seems flawed. It is likely that it is still more costly to detain defendants in this category than to locate them, so the manpower issue is one of resource allocation (and not truly a claim that the costs of rearrest are higher than the costs of detention). The resource allocation problem could reflect either red tape and bureaucracy problems or, perhaps more likely, the fact that many outstanding warrants are issued for offenses that are simply not a high enough priority to justify any additional resource expenditure. Professor Rachel Harmon asserts that “much of the time, no one bothers to hunt for suspects who fail to appear, though that is often because they were not worth charging with a crime in the first place.”291

   These distinctions are useful as interventions are considered. As Goldkamp explains, the distinctions suggest that “sanctioning and threat may not serve effectively as the all-purpose response relevant in all cases.”292 As Professor Tabarrok has explained, adopting a “behavioral perspective” would permit courts to view “crime control as less about punishing rational actors and more about helping criminogenic people to overcome their behavioral biases, thereby avoiding crime and the sequence of choices and events that inexorably leads to imprisonment and downfall.”293

2. Local absconding.

Of course, not all local FTAs are cheaply or easily prevented or remedied. Although imperfect, the lines between the low-cost nonappearances described above and the “local absconders” category described below reflect a combination of the willfulness, persistence, and higher costs associated with preventing and managing these nonappearances.

1. Willfulness.  Some nonappearing defendants actively and willfully avoid court and hide from law enforcement. Simple reminders are unlikely to bring them back to court. Even if they lack the resources to leave the jurisdiction, some defendants will actively work to “evade capture,” making “specific and strategic” choices to avoid detection, including, for example, “avoiding some forms of employment, choosing not to apply for public benefits, or otherwise limiting their interaction with formal institutions that could signal their location to authorities.”294 As this description immediately makes clear, the conduct of these local absconders creates a host of cascading problems for them and for their communities.295

   The true flight risks described above are similarly willful—they act intentionally and with a purpose to thwart the judicial process to avoid prosecution and punishment. The key distinction between these two groups is geographic. As noted above, fleeing the jurisdiction imposes costs and problems that distinguish it from local absconding. Members of the latter group are easier and less expensive to find than defendants who have fled the jurisdiction.

   Relying on willfulness is not a new premise. Some jurisdictions use willfulness to determine whether an FTA will be excused and whether a defendant who fails to appear forfeits his bail.296 For these reasons, there is a temptation to call this category “willful nonappearances.” But willfulness of an initial nonappearance is sometimes defined broadly by courts (and might capture some of the low-cost nonappearances described above). And deliberately missing a court date does not quite capture the problems presented by this category. For example, a defendant who deliberately misses court in order to keep his job belongs in the low-cost nonappearance bucket. A defendant who deliberately misses court to evade justice is properly categorized as a local absconder. Considering persistence and cost in combination with willfulness helps to narrow this category appropriately.

2. Persistence.  Nonappearances with a longer duration (that is, more persistent nonappearances) impose greater costs on the community than short-term nonappearances. Why does persistence matter? Certainly, as time passes, the more intangible costs of failing to secure justice for the underlying offense might be viewed as increasing.297 The direct costs of a nonappearing defendant increase over time, as well.298

   We have not adequately invested in identifying the defendants who frequently become persistent nonappearances, but the data regarding persistence are available, so this work could be done with some ease. For example, in their analysis of data about pretrial release of state felony defendants, Thomas Cohen and Brian Reaves used the term “fugitive” to refer to “anyone who missed a court appearance but could not be found (e.g., brought back to court) within the one year study coverage period.”299

   Although persistence is relevant and seems easy to track, it poses some problems and, like willfulness, should not be the sole criterion for prioritizing risks among local nonappearances. Principally, there is nothing in Cohen and Reaves’ data that indicates the efforts made to notify, locate, and bring back to court the particular defendants.300 The passage of time may be some indication of the evasiveness of a defendant, which would be helpful for a risk manager to know. But it may also reflect the disinterest of the jurisdiction. That, in turn, may be a function of several issues: (i) how difficult it is (or would be) to find the defendant; (ii) that the charged offense is not serious enough to justify the expenditure of effort to recover the defendant; or (iii) the weakness of the state’s case (even for more serious charges). Here again, these different explanations for the persistence of the nonappearance would suggest different interventions (or perhaps no intervention at all). Time, while relevant, cannot be the only categorical determinant. Combined with willfulness (described above) and higher costs (described next), it narrows our category of local absconders.

3. Higher costs.  Defendants who willfully fail to appear and who do so for extended periods of time (or indefinitely) have been described as the “Achilles heel” of law enforcement.301 Particularly when “local absconders” have been charged with more serious crimes, their nonappearances impose greater justice costs. The costs of locating defendants who are actively hiding is higher. Locating these defendants is possible but likely more “resource intensive” than locating defendants in the low-cost category.302 For this group, then, more aggressive conditions of release might be warranted.

C. A Research and Reform Agenda

This Article’s new taxonomy is a preliminary step. Because this taxonomy has not been formally recognized, gathering data about these new categories is an immediate priority. In addition, the definition of categories highlights systemic reforms that will immediately reduce nonappearance rates. The final Section of this Article briefly sketches a research and reform agenda.

Reformers must gather and analyze data about the number of defendants who might fall into each category and the interventions that best manage these distinct risks.303 This task will include: (i) collecting more data about true flight and local absconding to develop risk predictions for these more serious categories of nonappearances; (ii) collecting more data about which risk-management tools most effectively—and least intrusively—manage and prevent all forms of nonappearance; and (iii) better managing the creation of FTA data at the front end of the process, including, for example, suggesting best practices across jurisdictions about when FTAs are logged. Prior studies drawing some of the distinctions proposed here demonstrate that this research and analysis is feasible.304 This Article illustrates that the work is also required by federal and state constitutions, statutes, and as a matter of effective policymaking.

A shift to risk management (and beyond risk measurement) also highlights the need to change aspects of the system to reduce nonappearances. Differentiating between risks will force self-reflection for a criminal justice system that is complicit in the nonappearance problem. While it is certainly appropriate to focus principally on alleged offenders in developing a new nonappearance risk taxonomy (as the previous parts have done), there are also opportunities to reduce rates of nonappearance—principally among the broadest category of low-cost nonappearances—by pursuing systemic changes. Those changes should occur along at least five fronts.

First, the system’s complexity makes it difficult to navigate, particularly for defendants who may have “lower levels of education and IQ than the general population.”305 Judges within the system can work to simplify the process for the defendants who appear before them,306 or they can adjust their expectations and make appropriate accommodations.

Second, studies suggest that nonappearance rates increase as court delays increase.307 Reducing unnecessary delays must be a priority. In jurisdictions with bloated criminal court dockets and lengthy backlogs,308 high rates of nonappearance may not be surprising. Defendants may be expected to return to court frequently, “spend[ing] all day waiting for their cases to be called, only to be told that the proceedings are being put off for another month.”309 Addressing these dysfunctional court backlogs and reducing the time between an arrest and the resolution of a criminal case may help reduce rates of nonappearance.310

The third proposed system intervention allows for increased flexibility in scheduling court appearances. In prior studies, courts that permit appearances on weekends and evenings have seen increases in appearance rates.311 Even without opening the court at those times, courts might adopt more accommodating approaches to scheduling and rescheduling future appearances if necessary.312 Criminal courts might also consider permitting defendants to appear remotely.313

Courts must also adjust their responses to nonappearance. For the reasons outlined above, immediately logging an FTA and issuing a bench warrant is not an effective strategy for addressing nonappearance, particularly for low-level offenders.314

In addition, courts must eliminate the imposition of fines and fees on those who cannot afford to pay them, and should consider taking these steps as part of broader amnesty efforts for defendants who fail to appear.315 Too many defendants indicate that they cannot afford to return to court due to steep court fees and fines to clear warrants.316 The Department of Justice’s 2015 report on Ferguson described the Ferguson municipal court’s “focus on revenue generation” as leading to “court practices that violate the Fourteenth Amendment’s due process and equal protection requirements.”317 In particular, the report singled out the issuing of municipal arrest warrants “as a routine response to missed court appearances” intended to generate financial benefits for the court, in part because each additional missed appearance triggered more fines and fees.318 Former President Barack Obama described the role that these sorts of “user fees” play in “consigning those who cannot afford to pay to a cycle of debt, incarceration, and prolonged poverty.”319

The final system-focused intervention is less an independent proposal and more a byproduct of using the taxonomy and implementing the other interventions proposed here. The proposed taxonomy requires considering the different circumstances that lead to nonappearance. That inquiry may foster greater judicial and system-wide awareness of the competency, capacity, and resource limitations that inhibit compliance with court orders. Ideally, greater awareness will increase investment in flexible schedules or transportation and childcare accommodations that might increase appearance rates. In this way, requiring judges to inquire about—and strategize to address—the reasons for nonappearance could humanize the process and improve system outcomes.

Conclusion

This country’s uniquely swollen jail population and new economic reality have created a fiscal appetite for reform that is almost as great as the country’s moral obligation to reassess its pretrial detention practices. We have a system in which detention is the default choice. Overestimating pretrial risks has driven the jail population’s growth.

Judges imposing conditions of release or ordering pretrial detention must be clear-eyed about the precise risks they are trying to avoid or mitigate. The existing ambiguity is symptomatic of broader pretrial risk-measurement and risk-management problems. These persistent problems threaten new bail reform efforts. Greater precision is needed in the statutes, in the research being done to study these problems, in the judges’ attempts to manage these risks, and in the risk-assessment tools being developed to aid that endeavor.

While new risk-assessment tools promise improvements for pretrial decisionmaking, they also pose special risks. The tools lay bare obligations to address fundamental flaws in the existing doctrine, and, indeed, they make those obligations more urgent. If new risk-assessment tools map onto overbroad definitions of risk, it is likely that detention rates will not decrease significantly. Worse yet, weak definitions of the risks being addressed during the pretrial phase may gain legitimacy if supported by “scientific” estimates.

If risk-assessment tools are to fulfill the promise of reducing detention rates, they must identify and isolate those defendants who pose the most serious and costly pretrial risks. We have constitutional and statutory obligations to be clear about the outcomes we are trying to prevent and to be sparing in our use of liberty-restricting tools to avoid and manage those pretrial risks.

• 1FACT SHEET: Launching the Data-Driven Justice Initiative: Disrupting the Cycle of Incarceration (White House Office of the Press Secretary, June 30, 2016), archived at http://perma.cc/QKE5-PJVH (“On any given day, across the country more than 450,000 people are held in jail before trial.”).
• 2ABA Criminal Justice Section, State Policy Implementation Project *2, archived at http://perma.cc/K78U-HJ7Q.
• 3Human Rights Watch, The Price of Freedom: Bail and Pretrial Detention of Low Income Nonfelony Defendants in New York City *29–30 (Dec 2010), archived at http://perma.cc/RC9D-H8TJ (noting that nonfelony defendants were serving time in jail pretrial “only because they were unable to post bail”), quoting Mary T. Phillips, Pretrial Detention and Case Outcomes, Part 1: Nonfelony Cases *59 (New York City Criminal Justice Agency, Nov 2007), archived at http://perma.cc/P5FT-JWPZ.
• 4Ram Subramanian, et al, Incarceration’s Front Door: The Misuse of Jails in America *32 (Vera Institute of Justice, July 29, 2015), archived at http://perma.cc/R8PT-3PB7 (explaining that, of felony defendants who spend the full pretrial period in jail, “only one in ten” is incarcerated because a judge denied bail and ordered him or her to be detained and “[t]he rest simply cannot afford . . . bail”).
• 5Tina Rosenberg, Putting Fewer Innocents behind Bars (NY Times, July 3, 2015), archived at http://perma.cc/QVD9-7JJ2.
• 6Id. Judges need not consider public-safety concerns in pretrial detention and bail decisions. See NY Crim Proc Law § 510.30(2)(a).
• 7Rosenberg, Putting Fewer Innocents behind Bars (cited in note 5). By the time Browder’s family eventually raised the money, his status as an alleged probation violator precluded release. Jennifer Gonnerman, Before the Law (New Yorker, Oct 6, 2014), archived at http://perma.cc/GZ4Z-45E3 (first reporting Browder’s story).
• 8Gonnerman, Before the Law (cited in note 7).
• 9See Alec Karakatsanis, Policing, Mass Imprisonment, and the Failure of American Lawyers, 128 Harv L Rev F 253, 262 (2015) (describing pretrial hearings as an “assembly line” moving impoverished people into “cages”). The fact that Browder remained in jail for three years is far from routine. Most defendants in his shoes accede to the pressure to plead guilty—even if innocent—if it means immediate release from Rikers. See Gonnerman, Before the Law (cited in note 7) (“In 2011, in the Bronx, only a hundred and sixty-five felony cases went to trial; in three thousand nine hundred and ninety-one cases, the defendant pleaded guilty.”).
• 10See Randal John Meyer, Despite Public Outrage, Bail Reform Still Needed in New York City (Huffington Post, Feb 22, 2016), archived at http://perma.cc/ZM8H-M73Q; Alysia Santo, No Bail, Less Hope: The Death of Kalief Browder (Marshall Project, June 9, 2015), archived at http://perma.cc/H79F-ZA4P.
• 11Rosenberg, Putting Fewer Innocents behind Bars (cited in note 5).
• 12Jim Dwyer, A Life That Frayed as Bail Reform Withered (NY Times, June 9, 2015), online at http://www.nytimes.com/2015/06/10/nyregion/after-a-shocking-death-a-renewed-plea-for-bail-reform-in-new-york-state.html (visited Nov 9, 2017) (Perma archive unavailable) (explaining that Browder’s story is an indictment of the criminal justice system broadly and not simply a story of failures at Rikers Island). In January 2016, President Barack Obama retold Browder’s story while announcing a federal solitary-confinement ban for juveniles in federal facilities. See Barack Obama, Barack Obama: Why We Must Rethink Solitary Confinement (Wash Post, Jan 25, 2016), archived at http://perma.cc/7GQR-8C5J.
• 13See Part III.A (detailing current reforms).
• 14See Kristin Bechtel, et al, Dispelling the Myths: What Policy Makers Need to Know about Pretrial Research *2 n 1 (Pretrial Justice Institute, Nov 2012), archived at http://perma.cc/C9RE-N9GY (explaining that the first wave of bail reform began in the 1960s, the second in the 1980s, and the third in the early 2000s); Timothy R. Schnacke, Claire M.B. Brooker, and Michael R. Jones, The Third Generation of Bail Reform (Denver U L Rev Online, Mar 14, 2011), archived at http://perma.cc/F6MR-DZTU (discussing the interrelated factors that form the genesis of the third generation of bail reform). See also Sandra G. Mayson, Dangerous Defendants, 127 Yale L J 490, 515 (2018) (explaining that the third-generation bail reform model “requires that judges have authority to order pretrial preventive detention”).
• 15See Samuel R. Wiseman, Fixing Bail, 84 Geo Wash L Rev 417, 424 (2016) (explaining that, even in jurisdictions where judges employ actuarial tools to assess a defendant’s risk of flight and/or dangerousness, most often judges “retain full discretion to detain the defendant”); John S. Goldkamp and E. Rely Vîlcică, Judicial Discretion and the Unfinished Agenda of American Bail Reform: Lessons from Philadelphia’s Evidence-Based Judicial Strategy, in Austin Sarat, ed, Studies in Law, Politics, and Society: Special Issue New Perspectives on Crime and Criminal Justice 115, 127 (Emerald 2009).
• 16Lauryn P. Gouldin, Disentangling Flight Risk from Dangerousness, 2016 BYU L Rev 837, 887–88 (describing judges’ tendency to overestimate pretrial risks). See also Chris Guthrie, Jeffrey J. Rachlinski, and Andrew J. Wistrich, Blinking on the Bench: How Judges Decide Cases, 93 Cornell L Rev 1, 27–28 (2007) (explaining that judges “rely heavily on their intuitive faculties” and that they often cannot override their intuitions to clear the way for deliberative decisionmaking on the bench); Daniel Kahneman, Thinking, Fast and Slow 300–02 (Farrar, Straus & Giroux 2013) (explaining that, generally, downside risks have much higher salience than upside gains).
• 17See Laura and John Arnold Foundation, Developing a National Model for Pretrial Risk Assessment *1 (Nov 2013), archived at http://perma.cc/YB6U-3KEY (noting that judges’ subjective evaluations are not reliable risk assessments).
• 18See, for example, id at *2 (describing the tools as facilitating a shift away “from a system based solely on instinct and experience to one in which judges have access to scientific, objective risk assessment tools”); Gouldin, 2016 BYU L Rev at 888 (cited in note 16) (“Risk assessment tools address some of these concerns [of unconscious bias and overestimation] by replacing reliance on subjective and intuitive judicial measures of risk with more objective data that is insulated from cognitive bias.”).
• 19See Launching the Data-Driven Justice Initiative (cited in note 1). The initiative is a “bipartisan coalition,” with support from many organizations. Id.
• 20Id.
• 21See Anne Milgram, Moneyballing Criminal Justice (The Atlantic, June 20, 2012), archived at http://perma.cc/YU9K-LRGV. “Moneyballing” refers to the practice of using statistical analysis to achieve better and more efficient results. Id. The term is originally derived from a 2003 book about the Oakland Athletics’ use of an innovative method of data and statistical analysis to assemble a winning team on a limited budget. See generally Michael Lewis, Moneyball: The Art of Winning an Unfair Game (Norton 2003).
• 22Public Safety Assessment (Laura and John Arnold Foundation, Feb 5, 2017), archived at http://perma.cc/TW2H-NN5Y. The Arnold Foundation estimated in 2015 that approximately 10 percent of courts in the United States had adopted a risk-assessment tool. Id. See also Gouldin, 2016 BYU L Rev at 867 (cited in note 16) (noting that the use of risk-assessment tools is “rapidly growing”).
• 23Laura and John Arnold Foundation to Continue Data-Driven Criminal Justice Effort Launched under the Obama Administration (Laura and John Arnold Foundation, Jan 23, 2017), archived at http://perma.cc/Q887-2LXV. The foundation pledged to “dramatically expand[ ] its efforts to use data and analytics in order to address challenges in the criminal justice system” and explained that two former White House advisors who played key roles in Obama’s initiative had joined the foundation to continue their work. Id.
• 24See Wiseman, 84 Geo Wash L Rev at 420 (cited in note 15) (documenting a drop-off in bail-related scholarship after United States v Salerno but noting a recent resurgence of interest). See also, for example, Shima Baradaran and Frank L. McIntyre, Predicting Violence, 90 Tex L Rev 497, 507–13 (2012) (analyzing factors relevant to pretrial “dangerousness” predictions); Shima Baradaran, Restoring the Presumption of Innocence, 72 Ohio St L J 723, 738–39 (2011) (tracing the erosion of the pretrial presumption of innocence).
• 25See Wiseman, 84 Geo Wash L Rev at 421 (cited in note 15) (“Flight risk—the other chief factor in bail decisionmaking—has been largely ignored.”).
• 26See, for example, id at 442–46 (using flight risk and nonappearance interchangeably); Baradaran and McIntyre, 90 Tex L Rev at 545–48 (cited in note 24) (generally describing all nonappearance risks as “flight risk”). Criminologists supply additional ambiguity, using the term “fugitive” in similarly inconsistent ways. See Part I.A.
• 27See Part II.C (analyzing cases and highlighting references to both flight and nonappearance).
• 28See Part II.B (reviewing discussions of flight and nonappearance in federal and state statutes).
• 29Samuel R. Wiseman, Pretrial Detention and the Right to Be Monitored, 123 Yale L J 1344, 1352–53 (2014) (explaining that technological advances make fleeing less effective in avoiding trial).
• 30In this Article, “nonappearance risk” is used interchangeably with the risk that a defendant will “fail to appear.”
• 31See generally Gouldin, 2016 BYU L Rev 837 (cited in note 16).
• 32Risk-assessment tools place defendants into tiers of nonappearance risk (generally ranking them), but scaling nonappearance risk does not account for the differences between the categories. See Part III.B (describing the operation of risk-assessment tools). For example, someone who poses a very high risk of nonappearance is not the same as someone who poses a true flight risk.
• 33See Part IV.
• 34See Part III.B.
• 35See Stack v Boyle, 342 US 1, 5 n 3 (1951) (explaining that bail should be set to “insure the presence of the defendant”). See also Laura I. Appleman, Justice in the Shadowlands: Pretrial Detention, Punishment, & the Sixth Amendment, 69 Wash & Lee L Rev 1297, 1335 (2012) (“[A]lthough the specific intent of the Framers regarding bail cannot be conclusively determined, all the available evidence points to the fact that pretrial detention, both under English common law and at the time the Constitution was written, was limited to flight risks.”).
• 36See notes 24–25 and accompanying text.
• 37See, for example, Wiseman, 84 Geo Wash L Rev at 451 (cited in note 15) (discussing Virginia’s pretrial risk-assessment tool, which considers whether the defendant has two or more FTAs, among other factors).
• 38See, for example, 18 USC § 3142(c) (instructing federal judges to impose conditions of release that will “reasonably assure the appearance” of a defendant released pretrial); Cal Penal Code § 1275(a)(1):

  In setting, reducing, or denying bail, a judge or magistrate shall take into consideration the protection of the public, the seriousness of the offense charged, the previous criminal record of the defendant, and the probability of his or her appearing at trial or at a hearing of the case. The public safety shall be the primary consideration.

  See also, for example, Tenn Code Ann § 40-11-115(b) (“In determining whether or not a person shall be released . . . the magistrate shall take into account: . . . (8) Any other factors indicating the defendant’s ties to the community or bearing on the risk of willful failure to appear.”); Ind Code § 35-33-8-3.2(a):

  [A] court may admit a defendant to bail and impose any of the following conditions to assure the defendant’s appearance at any stage of the legal proceedings, or . . . (7) [r]elease the defendant on personal recognizance unless: (A) the state presents evidence relevant to a risk by the defendant: (i) of nonappearance . . . and (B) the court finds by a preponderance of the evidence that the risk exists.

• 39See 18 USC § 3142(g)(3)(A) (instructing judicial officers to take into account the defendant’s “family ties, employment, financial resources, length of residence in the community, [and] community ties” as well as the defendant’s “record concerning appearance at court proceedings,” among other characteristics). See also, for example, NY Crim Proc Law § 510.30(2)(a) (outlining that a court evaluating “the kind and degree of control or restriction that is necessary to secure [a defendant’s] court attendance when required . . . must . . . consider and take into account” a defendant’s “previous record if any in responding to court appearances when required or with respect to flight to avoid criminal prosecution”); Fla Stat § 903.046(2) (explaining that in setting bail or conditions of release courts “shall consider,” among other things, “[t]he defendant’s past and present conduct, including any record of convictions, previous flight to avoid prosecution, or failure to appear at court proceedings”).
• 40See, for example, 18 USC § 3142(f)(2)(A); 18 USC § 3142(b) (failing to define what it means that a release would not “reasonably assure the appearance of the person as required”).
• 41See 18 USC § 3142(g). See also note 39.
• 42See Part II.C.
• 43See note 26.
• 44See Part III.B.
• 45See, for example, Thomas H. Cohen and Brian A. Reaves, Pretrial Release of Felony Defendants in State Courts *8–10 (Bureau of Justice Statistics, Nov 2007), archived at http://perma.cc/6C8R-WSVU; John S. Goldkamp, Fugitive Safe Surrender: An Important Beginning, 11 Crimin & Pub Pol 429, 429–30 (2012).
• 46See, for example, 18 USC § 1073. The federal fugitive statute targets a fugitive that “moves or travels in interstate or foreign commerce” with the intent to avoid prosecution or arrest. 18 USC § 1073.
• 47In Bureau of Justice Statistics reports that use data from the State Court Processing Statistics program, defendants are deemed “fugitives” if they do not return to court within a year “irrespective if they remained in the local jurisdiction or fled to a different state [or] county.” Thomas H. Cohen, Administrative Office of the US Courts, Office of Probation & Pretrial Services, Email to Lauryn P. Gouldin, Syracuse University College of Law (Jan 24, 2017) (“Cohen Email”) (on file with author). See also Cohen and Reaves, Pretrial Release of Felony Defendants in State Courts at *8–10 (cited in note 45) (discussing fugitive status for purposes of pretrial release).
• 48Goldkamp, 11 Crimin & Pub Pol at 430 (cited in note 45) (discussing the phenomenon of “inadvertent fugitives”); David M. Bierie, National Public Registry of Active-Warrants: A Policy Proposal, 79 Fed Probation 27, 27–28 (2015) (describing defendants who are unaware of outstanding warrants as “unintentional fugitives”).
• 49Professor Alice Goffman notes that, at least in some communities, residents tend to “draw . . . distinctions between those likely to be taken into custody if the authorities do a general sweep, and those for whom the authorities are aggressively searching.” Alice Goffman, On the Run: Fugitive Life in an American City 7 (Chicago 2014).
• 50Cynthia A. Mamalian, State of the Science of Pretrial Risk Assessment *27 (Pretrial Justice Institute, Mar 2011), archived at http://perma.cc/Y9HF-QCZZ, citing Pretrial Justice Institute, Pretrial Justice in America: A Survey of County Pretrial Release Policies, Practices and Outcomes *10 (2010), archived at http://perma.cc/37AB-68ZN.
• 51See Part II.C.4.a (highlighting the significance of prior nonappearance in both statutes and risk-assessment tools); Part III.B (same).
• 52See Part II.C (reviewing factors specified in state and federal statutes that judges use to predict nonappearance risk).
• 53See Part III.B.2 (describing the emphasis in modern tools on prior FTAs).
• 54Brian A. Reaves, Felony Defendants in Large Urban Counties, 2009—Statistical Tables *21 (Bureau of Justice Statistics, Dec 2013), archived at http://perma.cc/U4RC-4WER. Reaves indicates that “[d]ata on the court appearance record and fugitive status were available for 95.1% of cases involving a defendant released prior to disposition.” Id. Other, older studies estimate rates of nonappearance as high as one-third for felony defendants. Daniel J. Flannery and Jeff M. Kretschmar, Fugitive Safe Surrender: Program Description, Initial Findings, and Policy Implications, 11 Crimin & Pub Pol 437, 453 (2012) (citing a 1999 study of New York City bench warrants). FTA rates also vary by “jurisdiction and offense type.” Brian H. Bornstein, et al, Reducing Courts’ Failure-to-Appear Rate by Written Reminders, 19 Psychology, Pub Pol & L 70, 70 (2013).
• 55Reaves, Felony Defendants in Large Urban Counties at *21 (cited in note 54).
• 56Id (explaining that the “detail” (13 and 3 percent) may not add up to the total (17 percent) because of rounding). See also id (“All defendants who failed to appear in court and were not returned to the court during the 1-year study period were counted as fugitives.”).
• 57Id at *15 (observing that the 2009 data was roughly equivalent to 2006 percentages “but lower than the 24% rate observed prior to 1996”). See also Eric Helland and Alexander Tabarrok, The Fugitive: Evidence on Public versus Private Law Enforcement from Bail Jumping, 47 J L & Econ 93, 109 (2004).
• 58Bornstein, et al, 19 Psychology, Pub Pol & L at 70 (cited in note 54); Meagan Cahill, Focusing on the Individual in Warrant-Clearing Efforts, 11 Crimin & Pub Pol 473, 478 (2012).
• 59Reaves, Felony Defendants in Large Urban Counties at *15 (cited in note 54).
• 60See Cahill, 11 Crimin & Pub Pol at 478 (cited in note 58) (“The failure-to-appear problem cannot be overlooked in any effort to address a jurisdiction’s warrant backlog.”). See also Kenneth Howe and Erin Hallissy, When Justice Goes Unserved: Thousands Wanted on Outstanding Warrants—but Law Enforcement Largely Ignores Them (SF Gate, June 22, 1999), archived at http://perma.cc/PTA7-R5GS (describing the backlog in California as of 1999 of “more than 2.5 million unserved warrants” and explaining that bench warrants for suspects who “failed to appear for their court dates” have “caused the number of warrants to balloon”). Warrant backlogs include warrants for other “wanted persons” including complaint and indictment warrants for suspects who have not yet been arrested, parole violation warrants, warrants for other individuals who have escaped from government custody, and criminal summonses, among others. Greg Hager, et al, Improved Coordination and Information Could Reduce the Backlog of Unserved Warrants *4–7 (Kentucky Legislative Research Commission Research Report No 326, July 14, 2005), archived at http://perma.cc/STE4-U2HJ.
• 61Cahill, 11 Crimin & Pub Pol at 478 (cited in note 58).
• 62Flannery and Kretschmar, 11 Crimin & Pub Pol at 437 (cited in note 54) (“The exact number of outstanding felony and misdemeanor warrants in local jurisdictions is unknown.”).
• 63Bierie, 79 Fed Probation at 27 (cited in note 48). See also David M. Bierie, Fugitives in the United States, 42 J Crim Just 327, 330 (2014) (“The data showed there were a total of 1.95 million active warrants in NCIC [in April 2011].”).
• 64136 S Ct 2056 (2016).
• 65Id at 2068 (Sotomayor dissenting).
• 66Id at 2066 (Sotomayor dissenting) (rejecting the majority’s holding that the discovery of an outstanding warrant served as an “intervening circumstance” between the clearly unlawful stop of Edward Strieff and the search that followed his subsequent arrest), citing US Department of Justice, Bureau of Justice Statistics, Survey of State Criminal History Information Systems, 2014 *38 (Dec 2015), archived at http://perma.cc/C45D-W83Y (Table 5a).
• 67Survey of State Criminal History Information Systems, 2014 at *38 (cited in note 66) (Table 5a).
• 68Id.
• 69Bierie, 79 Fed Probation at 27 (cited in note 48). See also Rachel A. Harmon, Why Arrest?, 115 Mich L Rev 307, 348 (2016) (arguing that summonses could replace arrest warrants even for those accused of committing serious violent felonies).
• 70Bierie, 42 J Crim Just at 328 (cited in note 63):

  [A prior study] examined one year’s worth of warrants in local crime databases of two counties and found that court violations (e.g., failure to appear) accounted for just over half of warrants present. They also found that approximately 75% of those court violations were for traffic offenses. However, there is no larger research to date describing warrants across crime types or other categorizations.

• 71US Department of Justice, Investigation of the Ferguson Police Department *55 (Mar 4, 2015), archived at http://perma.cc/U367-SFSQ (“Ferguson’s municipal court issues arrest warrants at a rate that police officials have called . . . ‘staggering.’”).
• 72Id at *6, 55.
• 73Recent efforts to reform the summons process in New York City reflect awareness that both the overall volume of summonses and the complexity of the summons process contribute to high rates of nonappearance. See Summons Reform (Mayor’s Office of Criminal Justice), archived at http://perma.cc/JF4H-VC2B (“Nearly four in ten summonses issued resulted in a warrant for failure to appear in court, which may affect the likelihood of future detention.”).
• 74See generally Fugitives: The Chronic Threat to Safety, Law, and Order, Hearing before the Subcommittee on Criminal Justice Oversight of the Committee on the Judiciary of the Senate, 106th Cong, 2d Sess (2000) (“Senate Hearings on Fugitives”).
• 75Bierie, 42 J Crim Just at 328 (cited in note 63).
• 76Id, quoting Goldkamp, 11 Crimin & Pub Pol at 430 (cited in note 45).
• 77See Parts IV.A–B (proposing a new taxonomy of nonappearance and flight and considering the costs of various types of nonappearance).
• 78Bell v Wolfish, 441 US 520, 534 (1979) (“[T]he Government has a substantial interest in ensuring that persons accused of crimes are available for trials and, ultimately, for service of their sentences, [and] confinement of such persons pending trial is a legitimate means of furthering that interest.”).
• 79See Goldkamp and Vîlcică, Judicial Discretion and the Unfinished Agenda of American Bail Reform at 131–32 (cited in note 15) (describing judges’ views that the seriousness of the charge is tied to the cost of “mistakes that could be made in making release decisions”).
• 80Thomas J. Miles, Estimating the Effect of America’s Most Wanted: A Duration Analysis of Wanted Fugitives, 48 J L & Econ 281, 281 (2005). Premised on the evasion of justice, Professor Thomas J. Miles’s analysis here assumes the guilt of the fugitive. See id.
• 81E. Rely Vîlcică and John S. Goldkamp, Bail Prediction: Exploring the Role of Neighborhood Context in Philadelphia, 42 Crim Just & Behav 1159, 1161 (2015).
• 82Id. See also Goldkamp, 11 Crimin & Pub Pol at 430 (cited in note 45) (“[T]he fugitive problem . . . has great significance for deterrence and the courts. The numbers of those intentionally avoiding court demonstrate the weakness in any intended deterrent message from the courts.”).
• 83See Part IV.A (defining “true flight risks”).
• 84See Part IV.B.2.b (analyzing the significance of the persistence or duration of nonappearance).
• 85See Part IV.
• 86Of course, from the defendant’s perspective, these costs may pale in comparison to the costs of showing up, which may include getting convicted, being sentenced, and being deported.
• 87Cahill, 11 Crimin & Pub Pol at 476 (cited in note 58) (“Outstanding warrants . . . [may] lead to additional illegal actions by cutting off individuals from legitimate sources of income or activity or by making them potential victims who are less likely to report their victimization to authorities for fear of the risk to themselves.”). See also Goffman, On the Run at 52 (cited in note 49) (“[A man on the run] doesn’t show up at the hospital when his child is born, nor does he seek medical help when he is badly beaten. He doesn’t seek formal employment. . . . He avoids calling the police when harmed or using the courts to settle disputes.”).
• 88Flannery and Kretschmar, 11 Crimin & Pub Pol at 439 (cited in note 54).
• 89Cahill, 11 Crimin & Pub Pol at 476 (cited in note 58) (“Having an outstanding warrant (or even simply believing that one has a warrant) can cause significant levels of individual- and family-level stress.”).
• 90Goffman, On the Run at 6, 186 (cited in note 49).
• 91Id at 52–53. See also Flannery and Kretschmar, 11 Crimin & Pub Pol at 439 (cited in note 54) (“The threat of going to jail can affect their close relationships and weaken already tenuous attachments to family, work, and community.”); Alexander Tabarrok, Fugitives, Outlaws, and the Lessons of Safe Surrender, 11 Crimin & Pub Pol 461, 464 (2012) (“Asociality, secrecy, and unpredictability become strategies to maintain liberty.”).
• 92Flannery and Kretschmar, 11 Crimin & Pub Pol at 449 (cited in note 54).
• 93Goldkamp, 11 Crimin & Pub Pol at 429–30 (cited in note 45) (“Some [defendants] . . . simply cannot afford financially to turn themselves in.”). See also Alicia Bannon, Mitali Nagrecha, and Rebekah Diller, Criminal Justice Debt: A Barrier to Reentry *13 (Brennan Center for Justice, 2010), archived at http://perma.cc/44VV-LMJH (“[U]npaid criminal justice debt puts individuals at risk of imprisonment and can impact everything from their employment and housing opportunities, to their financial stability, to their right to vote.”).
• 94Tabarrok, 11 Crimin & Pub Pol at 462–63 (cited in note 91).
• 95Goldkamp, 11 Crimin & Pub Pol at 429–30 (cited in note 45) (describing this as a form of “deterrence ‘backfire’”).
• 96Tabarrok, 11 Crimin & Pub Pol at 464 (cited in note 91).
• 97See Part II.A.
• 98See Part II.B.
• 99If the judge is unable to manage the risk of nonappearance using conditions of release, detention may be warranted. For example, as Professor Samuel R. Wiseman has observed, “No matter how ingenious the [electronic monitoring] technology, it is likely that highly motivated defendants will find a way to defeat it, perhaps by damaging or removing the tracking device or by blocking its signal.” Wiseman, 123 Yale L J at 1371 (cited in note 29). These “highly motivated defendants” would qualify as the sort of “extreme cases” cited in the text. Wiseman speculates, however, that under current practices in many jurisdictions, these defendants may already be at risk of being detained for dangerousness. Id.
• 100See Timothy R. Schnacke, Fundamentals of Bail: A Resource Guide for Pretrial Practitioners and a Framework for American Pretrial Reform *2 (National Institute of Corrections, Aug 2014), archived at http://perma.cc/P4EL-XEP8 (“‘[B]ail’ [refers to] a process of releasing a defendant from jail . . . with [financial or nonfinancial] conditions set to provide reasonable assurance of court appearance or public safety.”).
• 101Gouldin, 2016 BYU L Rev at 856 n 80 (cited in note 16).
• 102The origins of bail can be traced back to medieval England, where bail in its earliest forms was set to approximate the debt that a defendant might owe to a victim at the resolution of a case. See Schnacke, Fundamentals of Bail at *26–28 (cited in note 100). See also Clara Kalhous and John Meringolo, Bail Pending Trial: Changing Interpretations of the Bail Reform Act and the Importance of Bail from Defense Attorneys’ Perspectives, 32 Pace L Rev 800, 803 (2012) (“Because the amount of the pledge was equal to the potential penalty upon conviction, the ‘system necessarily linked the amount of the pretrial pledge to the seriousness of the crime.’”), quoting June Carbone, Seeing through the Emperor’s New Clothes: Rediscovery of Basic Principles in the Administration of Bail, 34 Syracuse L Rev 517, 520 (1983). In the United States, however, bail has always been described as an incentive (and not as any sort of prepayment or approximation of liability). Schnacke, Fundamentals of Bail at *40 (cited in note 100).
• 103Habeas Corpus Act of 1679, 31 Car II ch 2, reprinted in 5 Statutes of the Realm 935 (1819).
• 104Habeas Corpus Act of 1679, 31 Car II ch 2, reprinted in 5 Statutes of the Realm at 935 (cited in note 103).
• 105Samuel Wiseman, Discrimination, Coercion, and the Bail Reform Act of 1984: The Loss of the Core Constitutional Protections of the Excessive Bail Clause, 36 Fordham Urban L J 121, 127 (2009). See also Hermine Herta Meyer, Constitutionality of Pretrial Detention, 60 Georgetown L J 1139, 1189 (1972) (“By imposing excessive bail, the judges had made the Habeas Corpus Act inoperative with respect to those prisoners whom the King did not want to release.”).
• 106English Bill of Rights of 1689, 1 Wm & Mary sess 2, ch 2, reprinted in 6 Statutes of the Realm 142 (1819).
• 107English Bill of Rights of 1689, 1 Wm & Mary sess 2, ch 2, reprinted in 6 Statutes of the Realm at 143 (cited in note 106) (“[E]xcessive bail ought not to be required . . . .”).
• 108US Const Amend VIII.
• 109Wiseman, 36 Fordham Urban L J at 128 (cited in note 105) (tracing the history of the Excessive Bail Clause). See also id at 130 (“[The clause’s] complex and obscure history . . . has made consensus over the precise function of the constitutional prohibition against excessive bail elusive.”).
• 110Id at 123 (“There has been relatively little innovation in the law and scholarship on bail . . . since Salerno.”). See also United States v Salerno, 481 US 739, 752–54 (1987) (first citing Stack v Boyle, 342 US 1, 5 (1951), and then citing Carlson v Landon, 342 US 524, 545–46 (1952), as the only two previous Supreme Court cases addressing the Excessive Bail Clause).
• 111342 US 1 (1951).
• 112Id at 5.
• 113481 US 739 (1987).
• 114Pub L No 98-473, 98 Stat 1976, codified in various sections of Title 18.
• 115See Salerno, 481 US at 746–49 (explaining that the challenged preventive detention provisions were “regulatory” in nature and deeming them reasonable because they furthered the government’s legitimate interest in the safety of the community).
• 116Id at 754.
• 117Id.
• 118Stack, 342 US at 5. The Court granted petitioners’ motion to reduce bail after the trial court set bail based solely on the nature of the charge and without considering each defendant’s individual circumstances. Id. Modern bail litigation decisions ground this right to an “individualized hearing” in the Due Process Clause. See, for example, Jones v City of Clanton, 2015 WL 5387219, *2 (MD Ala).
• 119Salerno, 481 US at 754. See also Wiseman, 123 Yale L J at 1393 (cited in note 29):

  At least an intermediate level of scrutiny is consistent with precedent and warranted by the text and purpose of the Eighth Amendment. Applying this standard, the question is whether the use of money bail to ensure the defendant’s presence at trial is excessive under Salerno—that is, substantially broader than necessary to achieve the governmental interests at stake.

  See also Richard S. Frase, Excessive Prison Sentences, Punishment Goals, and the Eighth Amendment: “Proportionality” Relative to What?, 89 Minn L Rev 571, 603 (2005) (explaining that the Stack language “implies a form of means proportionality—if a lower bail amount would suffice, any higher bail is excessive”).

• 120The focus in Salerno was on whether detention was “excessive” in light of the public-safety concerns articulated in that case. The focus in this Article is on different forms of nonappearance risk, but similar specificity is constitutionally required for predicting different types of public-safety risk. See Salerno, 481 US at 751 (“When the Government proves by clear and convincing evidence that an arrestee presents an identified and articulable threat to an individual or the community, we believe that, consistent with the Due Process Clause, a court may disable the arrestee from executing that threat.”).
• 121See, for example, Campbell v Johnson, 586 F3d 835, 842 (11th Cir 2009) (“To determine whether bail is excessive, we must compare the terms of the bail against the interest the government seeks to protect.”); Galen v County of Los Angeles, 477 F3d 652, 662 (9th Cir 2007) (“Excessiveness cannot be determined by a general mathematical formula, but rather turns on the correlation between the state interests a judicial officer seeks to protect and the nature and magnitude of the bail conditions imposed in a particular case.”); United States v Scott, 450 F3d 863, 866 n 5 (9th Cir 2006) (“In some instances—when flight would be irrational, such as when the crime involves a minor traffic infraction—any amount of bail may be excessive because the bail amount would not serve the purpose of ensuring appearance in court to answer the charges.”).
• 122See Criminal Justice Policy Program, Moving beyond Money: A Primer on Bail Reform *8 (Harvard Law School, Oct 2016), archived at http://perma.cc/T7S8-QV38 (arguing that Stack’s “functional analysis of bail suggests that the Eighth Amendment imposes a sliding scale, linking constitutionally permissible bond amounts (or other conditions of release) to the amount needed to incentivize particular defendants to appear at court proceedings,” but noting that “[i]n practice, [ ] the courts have not applied this Eighth Amendment principle in a way that has meaningfully constrained the use of bail”) (citation omitted).
• 123Mayson, 127 Yale L J at 506 (cited in note 14).
• 12418 USC § 3142(b) (emphasis added).
• 125See, for example, 18 USC § 3142(c) (requiring a judge to order the pretrial release of a defendant on personal recognizance or unsecured appearance bond if “[the judge] determines that the release . . . will not reasonably assure the appearance of the [defendant] as required . . .”) (emphasis added).
• 126See, for example, DC Code § 23-1322(b)(1) (“The judicial officer shall hold a hearing to determine whether any condition or combination of conditions . . . will reasonably assure the appearance of the person . . . .”).
• 12718 USC § 3142(f)(2)(A).
• 12818 USC § 3142(d)(2), (e)(1) (contemplating both shorter- and longer-term pretrial detention).
• 12918 USC § 3142(e)(1). For a detention hearing to be held, either the government or the judicial officer must move to hold the hearing. 18 USC § 3142(f)(2). The motion must be made by the government if the case involves a narrow list of violent crimes or crimes with severe maximum sentences, 18 USC § 3142(f)(1), but can be made either by the government or the judicial officer for cases that involve a “serious risk that [the defendant] will flee,” 18 USC § 3142(f)(2)(A), or “a serious risk that [the defendant] will obstruct or attempt to obstruct justice, or threaten . . . or attempt to threaten . . . a prospective witness or juror,” 18 USC § 3142(f)(2)(B).
• 130See, for example, 18 USC § 3142(d)(2), (e)(1). The statute’s lack of clarity about the definitions of flight and nonappearance and the inconsistent use of these terms in other contexts make it difficult to state this with more certainty.
• 13118 USC § 3142(e)(1).
• 13218 USC § 3142(g)(3)(A).
• 13318 USC § 3142(c)(1)(B).
• 13418 USC § 3142(c)(1)(B).
• 13518 USC § 3142(c)(1)(B).
• 136See Wiseman, 123 Yale L J at 1395 n 229 (cited in note 29) (collecting state statutes “that require the use of the least restrictive means”).
• 137See, for example, 18 USC § 3142(g); NY Crim Proc Law § 510.30(2)(a); Ohio Rev Code Ann § 2937.222(C); Tenn Code Ann § 40-11-115(b); Fla Stat Ann § 903.046(2). See also Baradaran and McIntyre, 90 Tex L Rev at 503–04 (cited in note 24) (describing the historical development of statutory risk factors).
• 138See Part I.A. See also Gouldin, 2016 BYU L Rev at 865–66 (cited in note 16) (surveying state and federal laws that prescribe factors to be considered in pretrial decisionmaking and concluding that “the statutes do not indicate which factors are relevant to flight risk and which are believed to predict dangerousness”).
• 139See, for example, United States v Bustamante-Conchas, 557 Fed Appx 803, 806 (10th Cir 2014) (describing conditions that will “protect against [defendant’s] risk of flight” and “reasonably assure [his] appearance”); Ex parte Castillo-Lorente, 420 SW3d 884, 888–90 (Tex App 2014) (evaluating whether high bail set by trial court would “secure the defendant’s presence at trial” and “deter [him] from fleeing the jurisdiction”); Hernandez v State, 669 SE2d 434, 435 (Ga App 2008) (“The trial court’s foremost consideration when fixing the amount of bail should be the probability that the defendant, if freed, will appear at trial.”).
• 140See, for example, Fry v State, 990 NE2d 429, 446 (Ind 2013) (ruling that a court can release a defendant on “his own personal recognizance, unless the state shows evidence of a flight risk”), citing Ind Code § 35-33-8-3.2(a) (“[A] court may admit a defendant to bail and impose any of the following conditions to assure the defendant’s appearance at any stage of the legal proceedings, or . . . to assure the public’s physical safety.”) (emphasis added).
• 141Wiseman, 84 Geo Wash L Rev at 438 (cited in note 15) (“Judges make pretrial release decisions with only weak legislative guidance, and this grant of discretion gives rise to agency costs.”) (citation omitted). In light of these concerns, Wiseman proposes taking discretion away from judges entirely. See id (insisting that “[t]he next wave of bail reform should aim not at guiding judges’ discretion, but at radically limiting it—a classic means of reducing certain agency costs”).
• 14218 USC § 3142(g)(1)–(2).
• 143See, for example, United States v English, 629 F3d 311, 320 (2d Cir 2011) (upholding a detention order by a district-court judge who would have granted a bail motion had the defendant not been charged with both firearm and drug offenses); United States v Timley, 236 Fed Appx 441, 442 (10th Cir 2007) (endorsing the district court’s detention order for a defendant facing a mandatory life sentence because of the high incentive to flee); United States v Craven, 1998 WL 196622, *2 (1st Cir) (finding that the potential penalties give the defendant “a greater incentive to flee”).
• 144See, for example, Sneed v State, 946 NE2d 1255, 1259 (Ind App 2011) (noting that the “severity of the charges against [the defendant] sufficiently counterbalances her ties in the community and history of appearing in court”). See also Garcia v Wasylyshyn, 2007 WL 2216971, *2 (Ohio App) (“The nature and number of counts, as well as the possible sentences if convicted, support the implication that petitioner may indeed be a flight risk.”).
• 145See 18 USC § 3142(e)(2)–(3), (f)(1). Many states have similar rebuttable presumptions against release. See, for example, Alaska Stat § 12.30.011(d)(2); NC Gen Stat § 15A-533(d), (e).
• 146See Statement of Interest of the United States, Varden v City of Clanton, Civil Action No 15-34, *9 (MD Ala filed Feb 13, 2015) (“Varden Statement of Interest”) (“Fixed-sum bail systems . . . [set money bail] based solely on the criminal charge.”); Subramanian, et al, Incarceration’s Front Door at *32 (cited in note 4) (“[Bail schedules] link bail amounts to the severity of the initial charge, with criminal charge serving as a proxy for risk of re-arrest and flight, and the bail amount meant to mitigate that risk.”) (citation omitted).
• 147See, for example, Jennifer Fratello, Annie Salsich, and Sara Mogulescu, Juvenile Detention Reform in New York City: Measuring Risk through Research *9 (Vera Institute of Justice, Apr 2011), archived at http://perma.cc/SV2T-CKKP (“Neither charge type nor charge severity . . . were found to be associated with failure to appear or rearrest [of juvenile defendants], even though these factors are often included in normatively based detention risk assessments.”); Kelly Dedel and Garth Davies, Validating Multnomah County’s Juvenile Detention Risk Assessment Instrument *8 (Multnomah County Department of Community Justice, June 11, 2007), archived at http://perma.cc/8JXR-G2R4 (analyzing a risk-assessment instrument and identifying that offense seriousness was negatively correlated with pretrial FTA or rearrest); John S. Goldkamp and Michael R. Gottfredson, Policy Guidelines for Bail: An Experiment in Court Reform 70 (Temple 1985) (“When failure rates are examined by the seriousness of the defendant’s charges, the relation assumed by the conventional wisdom is not found: failure rates do not increase directly with seriousness levels.”).
• 148See notes 58–59 and accompanying text.
• 149See note 147.
• 150See, for example, United States v Friedman, 837 F2d 48, 49 (2d Cir 1988) (“[T]he Bail Reform Act does not permit detention on the basis of dangerousness in the absence of risk of flight, obstruction of justice or an indictment for the [enumerated] offenses.”); id at 50 (requiring “more than evidence of the commission of a serious crime and the fact of a potentially long sentence to support a finding of risk of flight”).
• 151See Goldkamp and Vîlcică, Judicial Discretion and the Unfinished Agenda of American Bail Reform at 125 (cited in note 15) (discussing evidence that “pretrial detention could even be assigned as outright punishment based on the charge standard”).
• 152Id at 131–32 (explaining that some judges have been reluctant to give up on the “charge standard”—that is, setting bail according to “the seriousness of the lead charge in a defendant’s case”—because they view the seriousness of the charge as “a crude indication of the potential costs” of losing defendants to flight).
• 153Id. Judges taking this view clearly presume the guilt of the arrestee.
• 154See note 151.
• 155See Kalhous and Meringolo, 32 Pace L Rev at 804 (cited in note 102) (tracing consideration of “the strength of the evidence” and “the likelihood of conviction” to at least the Statute of Westminster in 1275), citing Carbone, 34 Syracuse L Rev at 526–27 (cited in note 102). These practices continued in colonial America. Throughout the colonial period, both the seriousness of the charge and the weight of the evidence were viewed as “effective proxies for the risk of flight—where conviction appeared more likely, the presumption that the accused would flee was stronger.” Id at 806–07.
• 156See, for example, 18 USC § 3142(g)(2); Ohio Rev Code Ann § 2937.222(C)(2).
• 157Compare, for example, Bustamante-Conchas, 557 Fed Appx at 804 (affirming the decision to release the defendant when there was “substantial circumstantial evidence, but no direct evidence” that the defendant was guilty), with, for example, United States v Berkun, 392 Fed Appx 901, 903 (2d Cir 2010) (“Because the evidence of guilt is strong, it provides [the defendant] with an incentive to flee.”).
• 158See Baradaran, 72 Ohio St L J at 771 (cited in note 24) (noting that a bail hearing can become “a mini-trial before the actual trial (and significantly, without many of the protections that accompany a defendant at trial)”).
• 159Id at 770–72. See also Stack, 342 US at 4:

  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.

• 160See, for example, Bacon v United States, 449 F2d 933, 944 (9th Cir 1971) (finding that, although the government alleged that the defendant was able to flee, the government failed to establish a likelihood of flight).
• 161See, for example, United States v El-Hage, 213 F3d 74, 80 (2d Cir 2000) (making a flight determination because of the defendant’s “apparent access to false documents, his extensive history of travel and residence in other countries, and his alleged ties to an extensive and well-organized terrorist group whose leader and seven other of whose indicted members are still at large”).
• 162See, for example, Bacon, 449 F2d at 944 (finding no flight risk when “[t]here was no showing of past attempts . . . to evade judicial process, nor of past clandestine travels”).
• 163See, for example, El-Hage, 213 F3d at 80 (highlighting the defendant’s “apparent access to false documents” as a reason for determining that he was a flight risk).
• 164See Truong Dinh Hung v United States, 439 US 1326, 1328–29 (1978) (“Truong II”).
• 165See id at 1326–27.
• 166United States v Truong Dinh Hung, 577 F2d 738 (4th Cir 1978) (table) (“Truong I”).
• 167Truong II, 439 US at 1329. Truong appeared as required and his case became known for establishing the “primary purpose” standard that was eventually incorporated into the Foreign Intelligence Surveillance Act. See United States v Truong Dinh Hung, 629 F2d 908, 915 (4th Cir 1980) (“Truong III”) (holding that warrantless surveillance that was “conducted primarily for foreign intelligence reasons” did not violate the Fourth Amendment) (quotation marks omitted).
• 168These issues arise most often for defendants with significant overseas ties. See Kalhous and Meringolo, 32 Pace L Rev at 829 (cited in note 101) (“Another very common reason [for a defendant to be denied bail] is the country of origin of a defendant—whether he is a citizen or not. A naturalized citizen with contacts abroad is viewed with skepticism.”). Relatedly, as outlined in the next Section, defendants with domestic connections and anchors are viewed as a lower risk of flight.
• 169See, for example, United States v Villapudua-Quintero, 308 Fed Appx 272, 273 (10th Cir 2009) (affirming magistrate judge’s conclusion that family ties to Mexico rendered the defendant a flight risk); United States v Kattar, 1992 WL 80317, *4–5 (1st Cir) (considering factors like the defendant’s Lebanese passport, real property, and bank accounts when deciding whether to grant petition for release).
• 170See, for example, Bustamante-Conchas, 557 Fed Appx at 805 (considering defendant’s frequent and recent visits to Mexico); United States v Bonilla, 388 Fed Appx 78, 80 (2d Cir 2010) (considering defendant’s frequent trips to the Dominican Republic in recent years); United States v Khanu, 370 Fed Appx 121, 121–22 (DC Cir 2010) (considering defendant’s citizenship and pre-indictment trip to Sierra Leone).
• 171See, for example, Kalhous and Meringolo, 32 Pace L Rev at 829–33 (cited in note 102) (“[T]he government has advanced the proposition that individual defendants with ties to Israel present an additional risk of flight given Israel’s Law of Return.”).
• 172See, for example, United States v Valdivia, 104 Fed Appx 753, 754–55 (1st Cir 2004) (relying on the defendant’s “resources and foreign contacts” and “established ties outside the United States” to conclude that he posed a flight risk); United States v Aitken, 898 F2d 104, 107 (9th Cir 1990) (concluding that defendant was a flight risk in part because he had “access to large sums of cash”). See also Craven, 1998 WL 196622 at *2 (discussing legislative history pertaining to flight risk and drug trafficking).
• 173See, for example, United States v Madoff, 586 F Supp 2d 240, 243–44 (SDNY 2009); United States v Sabhnani, 493 F3d 63, 80 (2d Cir 2007) (“The defendants shall pay all costs associated with electronic monitoring.”).
• 174Benjamin Weiser, Rich Defendants’ Request to Judges: Lock Me Up in a Gilded Cage (NY Times, June 1, 2016), online at http://www.nytimes.com/2016/06/02/nyregion/rich-defendants-request-to-judges-lock-me-up-in-a-gilded-cage.html (visited Nov 9, 2017) (Perma archive unavailable), quoting United States v Dreier, 596 F Supp 2d 831, 833 (SDNY 2009).
• 175See Part IV.B.1.
• 176Amended Memorandum: Reasons for Pretrial Release, United States v Smith, Criminal Action No 15-20053, *2 (ED Mich filed May 4, 2015) (available on Westlaw at 2015 WL 2084720).
• 177See, for example, 18 USC § 3142(g)(3) (listing among factors to be considered in determining conditions of release: “the history and characteristics of the person, including . . . family ties, employment, financial resources, length of residence in the community, community ties . . .”); Ohio Rev Code Ann § 2937.222(C)(3)(a) (“[Factors include] family ties, employment, . . . length of residence in the community, community ties [and others].”).
• 178Compare, for example, Valdivia, 104 Fed Appx at 754 (finding that the defendant’s strong family and community ties did not overcome the presumption in favor of detention), and Hernandez, 669 SE2d at 435 (upholding high bail amount in light of the fact that the defendant did not own a home in the state, among other factors), with, for example, United States v Xulam, 84 F3d 441, 442 (DC Cir 1996) (overturning denial of bail for a defendant who had strong community ties and posed no threat to the community), and State v Brown, 338 P3d 1276, 1291–92 (NM 2014) (overturning high bail amount when the defendant had employment as well as family and community ties).
• 179See Part IV.B.1 (discussing low-cost nonappearances).
• 18018 USC § 3142(g)(3)(A). See also, for example, DC Code § 23-1322(e)(3)(A) (using the same language as the federal statute).
• 181Stack, 342 US at 9 (Jackson concurring). Some statutes include factors that permit a more general inquiry into trustworthiness. See, for example, NY Crim Proc Law § 510.30(2)(a)(i) (“The principal’s character, reputation, habits, and mental condition . . . [are factors in] the issuance of . . . bail.”).
• 182Stack, 342 US at 9 (Jackson concurring).
• 183See, for example, United States v Clum, 492 Fed Appx 81, 85 (11th Cir 2012) (noting the defendant’s prior refusal to cooperate). See also Lauryn P. Gouldin, When Deference Is Dangerous: The Judicial Role in Material-Witness Detentions, 49 Am Crim L Rev 1333, 1349–50 (2012) (analyzing relevance of prior cooperation to flight-risk analysis in material-witness cases).
• 184See, for example, Berkun, 392 Fed Appx at 903 (upholding detention because the defendant committed the alleged crime while on bail for a different crime, revealing a “record of deceiving the court”); United States v Kisling, 334 F3d 734, 735 (8th Cir 2003) (upholding the denial of bail when defendant’s history of avoiding his legal troubles, including evading service and an FTA, made him a flight risk). See also, for example, Querubin v Commonwealth, 795 NE2d 534, 544 (Mass 2003) (holding that defendant was a flight risk because he was known to use an alias, originally eluded several police officers when they attempted to arrest him on several default warrants, and as a result failed to appear before the court on a cocaine trafficking charge); People v Gurule, 174 P3d 846, 846 (Colo App 2007) (upholding the denial of an appeal bond because the court determined that the defendant posed a flight risk due to his FTA for a sentencing proceeding and return only several months later).
• 185See, for example, Clum, 492 Fed Appx at 85 (upholding the district court’s denial of bail when the defendant resisted arrest, refused to cooperate, and had a history of defying the authority of the courts, including one FTA).
• 186See notes 214–28 and accompanying text (discussing the Arnold Foundation’s risk-assessment tool). See also, for example, United States v Rico, 551 Fed Appx 446, 447 (10th Cir 2014) (finding the defendant’s two recent FTAs relevant to nonappearance risk analysis); State v Dunn, 2014 WL 3714647, *5 (Vt) (upholding the defendant’s substantial monetary bail due to two previous FTAs, two probation violations, and an out-of-state address).
• 18718 USC § 3142(g)(3)(A). See also, for example, Fla Rules Crim Proc § 3.131(b)(3); Rico, 551 Fed Appx at 447 (finding the defendant to be a flight risk due in part to his history of drug and alcohol abuse).
• 188See, for example, Bustamante-Conchas, 557 Fed Appx at 804 (favoring release when, among other factors, the defendant lacked a history of drug or alcohol abuse); Brown, 338 P3d at 1291 (same).
• 189See Wiseman, 84 Geo Wash L Rev at 426–28 (cited in note 15) (discussing how judicial discretion creates agency costs).
• 190Developing a National Model for Pretrial Risk Assessment at *5 (cited in note 17). To be clear, risk assessment is not new. The statutory risk factors outlined in Part II.C were designed to help judges assess risk. Those factors derived from risk-assessment tools developed by the Vera Institute of Justice in the 1960s. Risk-assessment tools, however, have evolved considerably over time. For a detailed summary of the history of these tools and of their use in a range of criminal justice contexts, see Melissa Hamilton, Risk-Needs Assessment: Constitutional and Ethical Challenges, 52 Am Crim L Rev 231, 236–39 (2015); id at 233 (explaining that these tools are seeing their “heyday in criminal justice”). See also Melissa Hamilton, Back to the Future: The Influence of Criminal History on Risk Assessments, 20 Berkeley J Crim L 75, 91–95 (providing a general summary of how these tools work).
• 191As noted in Part III.C, these tools have prompted a range of other thoughtful and well-developed critiques, many of which are beyond the scope of this Article.
• 192Gouldin, 2016 BYU L Rev at 892 (cited in note 16).
• 193See Subramanian, et al, Incarceration’s Front Door at *31 (cited in note 4) (“[T]he best [risk-assessment] tools evaluate the person’s dynamic or changeable risk factors and needs, [so] they should be re-administered routinely to determine whether current supervision or custody levels and programming are still appropriate.”).
• 194See Developing a National Model for Pretrial Risk Assessment at *5 (cited in note 17) (“These systemic failures put the public in danger and place unnecessary strain on budgets, jails, law enforcement, families, and communities.”).
• 195FACT SHEET: Launching the Data-Driven Justice Initiative (cited in note 1).
• 196Shaila Dewan, Judges Replacing Conjecture with Formula for Bail (NY Times, June 26, 2015), online at http://www.nytimes.com/2015/06/27/us/turning-the-granting-of-bail-into-a-science.html (visited Nov 9, 2017) (Perma archive unavailable). In 2015, fewer than 10 percent of jurisdictions were employing these types of tools, but that number has been increasing. Id.
• 197Equal Justice Initiative, Delaware Access to Justice Commission’s Committee on Fairness in the Criminal Justice System: A Report on Bail and Pretrial Detention *9 (Oct 2015), archived at http://perma.cc/2VC7-R7EU (“In 2014, New Jersey passed two pieces of bail/pretrial detainment reform legislation concurrently . . . [to] shift[ ] New Jersey’s pretrial release system from a money-based bail system to a primarily risk-based system.”). See Chapter 31, 2014 NJ Sess Law 467, codified at NJ Stat Ann § 2A:162-15 to -26; NJ Const Art I, ¶ 11, amended by 2014 NJ Sess Law 865. See also Lisa W. Foderaro, New Jersey Alters Its Bail System and Upends Legal Landscape (NY Times, Feb 6, 2017), online at http://www.nytimes.com/2017/02/06/nyregion/new-jersey-bail-system.html (visited Nov 9, 2017) (Perma archive unavailable) (explaining that these changes, which went into effect in January 2017, “placed the state in the forefront of a national movement aimed at changing [the] bail system”). New Mexico voters recently amended their constitution. See NM Const Art II, § 13 (declaring that no defendant eligible for pretrial release shall “be detained solely because of financial inability to post a money or property bond”); Susan Montoya Bryan, Voters: Judges Can Deny Bail to Dangerous Defendants (Santa Fe New Mexican, Nov 8, 2016), archived at http://perma.cc/6SF6-LU2Z.
• 198See Press Release, Gov. Malloy Signs Legislation Reforming the State’s Pretrial Justice System to Help Break the Cycle of Crime and Poverty (Office of Governor Dannel P. Malloy, June 28, 2017), archived at http://perma.cc/2SWC-LY26; An Act concerning Pretrial Justice Reform, Public Act 17-145, codified at Conn Gen Stat § 54-64a(a) et seq.
• 199See Steve Hughes, Cuomo’s Bail Reform Effort Spurs Debate (Times Union, Feb 11, 2018), archived at http://perma.cc/7RZ9-UVBK (“New York appears increasingly likely to join several other states that have enacted bail reforms.”); Andrew M. Cuomo, New York State: Excelsior Ever Upward, 2018 State of the State *59–60 (2018), archived at http://perma.cc/HLG3-B4WG.
• 200Md Rule 4-216.1(b)(2) (calling for “individualized consideration” of a defendant’s “ability . . . to meet a special condition of release with financial terms”). Additional legislative reforms may follow. See Michael Dresser, Maryland Court of Appeals: Defendants Can’t Be Held in Jail Because They Can’t Afford Bail (Baltimore Sun, Feb 8, 2017), archived at http://perma.cc/PBX6-A4Y3.
• 201See, for example, Varden Statement of Interest at *1 (cited in note 146) (associated with a case in which Ms. Varden, an indigent defendant, was jailed after the Clanton Municipal Court used a generic bail schedule to impose bail in the amount of $2,000, which she could not afford to pay). The Department of Justice filed a statement of interest in the case asserting: “Incarcerating individuals solely because of their inability to pay for their release, whether through the payment of fines, fees, or a cash bond, violates the Equal Protection Clause of the Fourteenth Amendment.” Id. This statement of interest has since been cited in bail challenges across the country. See, for example, Thompson v Moss Point, 2015 WL 10322003, *1 (SD Miss).
• 202Commercial bail bondsmen or commercial sureties provide bail to defendants for a price and have, over time, become a fixture in American criminal justice. Their lobbyists frequently challenge bail reform efforts. Many groups advocate for abolition of these for-profit bail enterprises. See Justice Policy Institute, Finding Direction: Expanding Criminal Justice Options by Considering Policies of Other Nations *1–3 (Apr 2011), archived at http://perma.cc/YMF5-7PSU (observing that other countries like Australia, Canada, Finland, Germany, and England do not permit commercial bail).
• 203In 2015, a New York City council member recommended that the city allocate $1.4 million of community funds to pay “bail set at $2,000 or lower for defendants charged with low-level misdemeanors and unable to afford it.” Emily Ngo, NYC Council Speaker Melissa Mark-Viverito Proposes Bail Fund for Indigent Defendants (Newsday, June 17, 2015), archived at http://perma.cc/3BC3-7DPL. See also Jocelyn Simonson, Bail Nullification, 115 Mich L Rev 585, 599–606 (2017) (describing the growing phenomenon of community bail funds in detail).
• 204See note 14.
• 205Marie VanNostrand and Kenneth J. Rose, Pretrial Risk Assessment in Virginia *3 (Luminosity, May 1, 2009), archived at http://perma.cc/63S7-WNTQ. As I have explained in other work, in the pretrial context, there are constitutional and statutory requirements that call for separate measurements of flight (or nonappearance) risk and danger. Gouldin, 2016 BYU L Rev at 871–81 (cited in note 16). There are also compelling policy reasons for disentangling those risks. Id at 885–93.
• 206It is worth clarifying that “flight risk” as contemplated in this Article is not the same as a very high risk of nonappearance. For the reasons outlined in Part IV.A, flight risk should be treated as a different type of risk than other forms of nonappearance (not merely as a different degree of the same risk).
• 207See Mayson, 127 Yale L J at 510–15 (cited in note 14) (comparing current pretrial risk-assessment tools).
• 208Marie VanNostrand and Gena Keebler, Pretrial Risk Assessment in the Federal Court *5 (Luminosity, Apr 14, 2009), archived at http://perma.cc/8NHF-R63S.
• 209See Mayson, 127 Yale L J at 512 (cited in note 14).
• 210Jessica Eaglin, Constructing Recidivism Risk, 67 Emory L J 59, 79 (2017) (explaining that for the recidivism-predicting tools used in the sentencing context, tool developers identified predictive factors with a “statistically significant correlation” with recidivism).
• 211Nathan James, Risk and Needs Assessment in the Criminal Justice System *3 (Congressional Research Service, June 22, 2015).
• 212Vîlcică and Goldkamp, 42 Crim Just & Behav at 1162 (cited in note 81) (reviewing studies of current risk-assessment tools).
• 213See notes 132–36 and accompanying text.
• 214See Dewan, Judges Replacing Conjecture with Formula for Bail (cited in note 196).
• 215Id.
• 216Public Safety Assessment (cited in note 22) (noting that the PSA is “far less expensive . . . than previous techniques”). See also Dewan, Judges Replacing Conjecture with Formula for Bail (cited in note 196) (explaining that part of the appeal is that the PSA is “designed to be more economical than existing risk assessments”).
• 217Public Safety Assessment (cited in note 22).
• 218Mayson, 127 Yale L J at 511–12 (cited in note 14) (outlining the risk factors used in all pretrial tools currently in use). Risk factors that are “static” include, for instance, having a prior conviction and a history of FTAs, while “dynamic” factors include employment status and substance abuse. Id at 512.
• 219Laura and John Arnold Foundation, Public Safety Assessment: Risk Factors and Formula *2, archived at http://perma.cc/3GHR-GT8P. The problems with combining future dangerousness and flight risk into a single risk measure are discussed in an earlier article. See Gouldin, 2016 BYU L Rev at 885–89 (cited in note 16).
• 220Public Safety Assessment: Risk Factors and Formula at *2 (cited in note 219).
• 221See Eaglin, 67 Emory L J at 118–19 (cited in note 210) (arguing for greater transparency regarding risk-assessment tool inputs and outcomes to “facilitate public accountability”).
• 222Public Safety Assessment: Risk Factors and Formula at *4 (cited in note 219).
• 223Public Safety Assessment (cited in note 22). See also text accompanying note 215.
• 224Statutory factors are detailed in Part II.C. Risk factors used in other risk-assessment tools are outlined in Part III.B.1. See also Public Safety Assessment: Risk Factors and Formula at *2 (cited in note 219) (listing the risk factors used by PSA).
• 225Public Safety Assessment: Risk Factors and Formula at *2 (cited in note 219).
• 226Id at *3.
• 227Developing a National Model for Pretrial Risk Assessment at *4 (cited in note 17).
• 228See Public Safety Assessment (cited in note 22).
• 229See, for example, Public Safety Assessment: Risk Factors and Formula at *2–3 (cited in note 219) (noting that a prior FTA in the past two years is worth twice as much weight as other factors, such as prior convictions, in calculating a defendant’s risk of FTA for future court dates). Reliance on prior FTAs to predict a future FTA is based on studies indicating the strength of nonappearance history in predicting future nonappearance. See Cohen and Reaves, Pretrial Release of Felony Defendants in State Courts at *9–10 (cited in note 45) (showing that defendants with a prior FTA were more likely to fail to appear than defendants with no arrest record and than those with an arrest record but no previous FTA).
• 230See, for example, VanNostrand and Keebler, Pretrial Risk Assessment in the Federal Court at *3 (cited in note 208) (noting that the Bail Reform Act of 1984 identified the defendant’s “record of appearances at court proceedings” as one factor that courts should consider).
• 231Id at *21.
• 232Id at *20 (defining pretrial failure in part as “failing to appear for court”).
• 233There is a growing literature analyzing the methodology of developing risk-assessment tools and algorithms for the criminal justice system. See, for example, Andrew Guthrie Ferguson, The Rise of Big Data Policing: Surveillance, Race, and the Future of Law Enforcement 136–40 (NYU 2017) (emphasizing the need for greater accountability for tool developers and algorithm builders); Eaglin, 67 Emory L J at 105 (cited in note 210) (“[G]overnment entities and tool developers should adhere to various accountability measures in the construction of actuarial risk tools.”). See also generally Michael L. Rich, Machine Learning, Automated Suspicion Algorithms, and the Fourth Amendment, 164 U Pa L Rev 871 (2016) (focusing on use of machine-learning algorithms in the Fourth Amendment context).
• 234See notes 26–30 and accompanying text (highlighting inconsistent usage of the terms “flight risk” and “nonappearance risk”).
• 235See Wiseman, 84 Geo Wash L Rev at 420–21 (cited in note 15) (describing the focus in modern bail scholarship on predicting dangerousness and the neglect of flight risk).
• 236See, for example, Edward Latessa, et al, Creation and Validation of the Ohio Risk Assessment System: Final Report *16 (July 2009), archived at http://perma.cc/7ZZ2-RHGS (explaining that one key distinction between Ohio’s Pretrial Assessment Tool and similar tools used in other contexts (such as community supervision, prison intake, and reentry) is that the former requires prediction of the likelihood of a future FTA). See also Cecelia Klingele, The Promises and Perils of Evidence-Based Corrections, 91 Notre Dame L Rev 537, 564–67 (2015) (documenting the embrace of risk-assessment instruments in the sentencing and corrections contexts).
• 237See Solon Barocas and Andrew D. Selbst, Big Data’s Disparate Impact, 104 Cal L Rev 671, 691 (2016) (explaining that the disparate impact caused by certain tool features is the product of the creators’ “reasonable priorities as profit seekers”); Eaglin, 67 Emory L J at 66 (cited in note 210) (explaining that developers’ “desire for cheap, varied, and easily accessible data” influences the choices they make in constructing risk tools, sometimes in “conflict with a state’s existing sentencing policies and practices”); id at 80 (describing recidivism risk tool developers’ decision to use readily available arrest data as the measure of recidivism).
• 238See Barocas and Selbst, 104 Cal L Rev at 688 (cited in note 237) (explaining that sometimes the data selected for study “fail[s] to capture enough detail to allow for the discovery of crucial points of contrast”).
• 239See id at 674 (“Approached without care, data mining can reproduce existing patterns of discrimination, inherit the prejudice of prior decision makers, or simply reflect the widespread biases that persist in society.”); Eaglin, 67 Emory L J at 101–04 (cited in note 210).
• 240Joshua A. Kroll, et al, Accountable Algorithms, 165 U Pa L Rev 633, 640 (2017). See also id at 678 (highlighting the importance of assuring “fidelity to substantive policy choices”); Ferguson, The Rise of Big Data Policing at 139 (cited in note 233) (“Accountability rather than transparency provides a way out of the black data opacity problem. Accountability by design will require big data policing to confront [problems of] bias, fear, and fair application.”); Eaglin, 67 Emory L J at 105 (cited in note 210) (explaining that tool developers make decisions in constructing risk-assessment instruments that “implicate larger normative questions best left for criminal justice experts and the political process to resolve”).
• 241The risk typology proposed in Part IV is intended to improve pretrial decisionmaking, including but not limited to improving the accuracy and utility of risk-assessment tool predictions. It does not directly address or cure the critiques briefly described here.
• 242Frederick Schauer, Profiles, Probabilities, and Stereotypes 19 (Belknap 2003) (noting that “many people believe it wrong to make individual decisions on the basis of nonuniversal group characteristics even if the group attributions have a solid statistical grounding”). See also Christopher Slobogin, Proving the Unprovable: The Role of Law, Science, and Speculation in Adjudicating Culpability and Dangerousness 111–13 (Oxford 2007) (defending actuarial predictions of violence risk); Bernard E. Harcourt, Against Prediction: Profiling, Policing, and Punishing in an Actuarial Age 237 (Chicago 2007) (arguing that actuarial methods not only “aggravate social disparities,” but might “backfire” and “increase rather than decrease the overall amount of crime in society”).
• 243Schauer, Profiles, Probabilities, and Stereotypes at 19 (cited in note 242).
• 244See, for example, Sandra G. Mayson, Bias in, Bias out: Criminal Justice Risk Assessment and the Fantasy of Race Neutrality *32–33 (unpublished manuscript, 2017) (on file with author); Sonja B. Starr, Evidence-Based Sentencing and the Scientific Rationalization of Discrimination, 66 Stan L Rev 803, 851 (2014) (critiquing existing evidence-based regimes and urging approaches that “base actuarial prediction only on crime characteristics and criminal history” and that strip “socioeconomic and demographic variables”); Hamilton, 52 Am Crim L Rev at 242 (cited in note 190) (analyzing evidence-based sentencing schemes and determining that the use of socioeconomic and demographic variables to contribute to a defendant’s sentence may violate the Equal Protection and Due Process Clauses).
• 245See, for example, Developing a National Model for Pretrial Risk Assessment at *5 (cited in note 17).
• 246See notes 270–73 and accompanying text. See also Alexandra Natapoff, Misdemeanor Decriminalization, 68 Vand L Rev 1055, 1063–65 (2015) (describing challenges for poor defendants navigating the criminal justice system).
• 247See Eaglin, 67 Emory L J at 95 (cited in note 210) (“[M]ore frequent contact with the justice system does not necessarily mean higher risk to the public. Much of this contact comes from heightened scrutiny, not necessarily more criminal wrongdoing.”). See also note 239 (explaining that, when used improperly, data may simply reproduce existing discrimination).
• 248Goldkamp, 11 Crimin & Pub Pol at 430–31 (cited in note 45) (describing the “elusiveness of complete and accurate data relating to fugitives” and explaining that “just the task of counting fugitives to define the numerators and denominators of potential effectiveness measures presents difficult challenges”). See also Harmon, 115 Mich L Rev at 337 (cited in note 69) (“[T]he studies are too few, too limited, and too dated to draw strong conclusions.”).
• 249Although data about who flees the jurisdiction are not presently collected and analyzed in the way envisioned by this paper, it may not be particularly difficult to begin to gather relevant statistics. At present, “fugitives are selected into the [US Marshals Service] in part based on the presumption of an offender having crossed state or national boundaries.” David M. Bierie and Paul J. Detar, Geographic and Social Movement of Sex Offender Fugitives, 62 Crime & Delinq 983, 997 (2016).
• 250See Part II.B (statutes); Part II.C (cases).
• 251Brad Heath, The Ones That Get Away (USA Today, Mar 11, 2014), archived at http://perma.cc/STF7-A77D (explaining that notations in the NCIC database indicate whether the jurisdiction of initial arrest is willing to pay to extradite the defendant).
• 252See id.
• 253See notes 174–76 and accompanying text.
• 254The Bail Reform Act, for example, imposes obligations on judges to order release of defendants “subject to the least restrictive further condition, or combination of conditions,” that would ensure their appearances. 18 USC § 3142(c)(1)(B). The appropriateness of pretrial detention to manage public-safety risk (as opposed to nonappearance risk) is beyond the scope of this Article. See Baradaran and McIntyre, 90 Tex L Rev at 526–29 (cited in note 24) (analyzing predictions of pretrial dangerousness). See also Shima Baradaran Baughman, Costs of Pretrial Detention, 97 BU L Rev 1, 19 (2017) (comparing the costs associated with pretrial public-safety risk with the costs of pretrial detention).
• 255Gouldin, 2016 BYU L Rev at 894–95 (cited in note 16) (describing conditions of release that manage nonappearance risk).
• 256Legislatures and courts have long recognized this connection; a defendant’s financial resources are a listed federal statutory factor. 18 USC § 3142(g)(3)(A). See also notes 172–76 and accompanying text. See also Why It Matters (Uptrust), archived at http://perma.cc/9XY2-ZNUJ (“When poor people miss court, they do not flee.”).
• 257There is at least one decades-old study that suggests that money bail deters flight for released defendants. See Samuel L. Myers Jr, The Economics of Bail Jumping, 10 J Legal Stud 381, 395 (1981) (finding that setting higher bail discouraged flight among released defendants).
• 258Wiseman, 123 Yale L J at 1352–53 (cited in note 29). See also Harmon, 115 Mich L Rev at 339–40 (cited in note 69) (explaining that technology makes it increasingly easy to locate a suspect if he fails to appear); id (describing the “extensive digital trail” created by the use of “credit cards, bank cards, electronic benefits transfer (EBT) cards, monthly transit cards, electronic tolling devices (like FasTrak and EZPass), and many other location-based services and devices,” in addition to cell phones).
• 259Unfortunately, use of the database is not universal. See Law Enforcement Information Sharing (Information Sharing Environment), archived at http://perma.cc/THB7-MVHN.
• 260Miles, 48 J L & Econ at 281 (cited in note 80) (“[E]stimates show that broadcasting a fugitive’s profile on America’s Most Wanted substantially raises the apprehension hazard by a factor of seven and shortens the expected fugitive spell by roughly a fourth.”).
• 261Wiseman, 123 Yale L J at 1352–53 (cited in note 29).
• 262Id at 1362.
• 263Cohen and Reaves, Pretrial Release of Felony Defendants in State Courts at *7 (cited in note 45); Cohen Email (cited in note 47) (explaining that the term “fugitives” included defendants who “remained in the local jurisdiction”). See also Helland and Tabarrok, 47 J L & Econ at 109 (cited in note 57) (“Those [defendants] who remain at large more than 1 year are called fugitives.”).
• 264Bierie and Detar, 62 Crime & Delinq at 996 (cited in note 249).
• 265Id.
• 266Id at 983. Bierie and Detar found that 37 percent of the offenders in their study

  fled to a familiar area, 65% lived with friends or family at capture, and 50% traveled more than 370 miles (with 35% residing in an adjacent state to the last known address). Analyses also showed that these three outcomes varied as a function of offender demographics, geographic history, social networks, and criminal history.

  Id.

• 267The intent of a defendant is not being gauged as a measure of culpability. Instead, it functions as a sort of proxy for the likelihood of assuring future appearances and the costs of recovering or locating defendants who do not appear.
• 268It is worth noting that the seriousness of the offense of arrest is discussed throughout this Part as it intersects with each of the proposed categories. It is not, however, used to define the categories.
• 269See Harmon, 115 Mich L Rev at 338 (cited in note 69) (discussing the “familiar and manageable” reasons defendants might fail to appear). See also Why It Matters (cited in note 256) (contending that poor defendants who “miss court” do so “because they lack transportation, could not take time off work, had to care for their children, or simply don’t keep a good calendar”).
• 270Goldkamp, 11 Crimin & Pub Pol at 429–30 (cited in note 45) (asserting that these ideas are implied in Flannery and Kretschmar’s study of the Fugitive Safe Surrender program). See also Bierie, 79 Fed Probation at 27–28 (cited in note 48) (drawing similar distinction between “unintentional” and “intentional” fugitives).
• 271Goldkamp, 11 Crimin & Pub Pol at 429 (cited in note 45).
• 272Flannery and Kretschmar, 11 Crimin & Pub Pol at 451–52 (cited in note 54).
• 273Harmon, 115 Mich L Rev at 337–38 (cited in note 69) (describing the “malleab[ility]” of appearance rates for defendants who are issued citations instead of being arrested).
• 274Bornstein, et al, 19 Psychology, Pub Pol & L at 76 (cited in note 54) (discussing test results which “show[ed] that a reminder effectively reduces the FTA rate”). See also Marie VanNostrand, Kenneth J. Rose, and Kimberly Weibrecht, State of the Science of Pretrial Release Recommendations and Supervision *20 (Pretrial Justice Institute, June 2011), archived at http://perma.cc/QFK5-8UGG (summarizing studies of notifications in different jurisdictions and noting that “[a]ll of the studies concluded that court date notifications in some form are effective at reducing failures to appear in court”).
• 275VanNostrand, Rose, and Weibrecht, State of the Science at *15–20 (cited in note 274).
• 276Subramanian, et al, Incarceration’s Front Door, at *33–34 (cited in note 4) (describing data from Coconino County, Arizona). See also VanNostrand, Rose, and Weibrecht, State of the Science at *17–18 (cited in note 274).
• 277Barry Mahoney, et al, Pretrial Services Programs: Responsibilities and Potential *39 (National Institute of Justice, Mar 2001), archived at http://perma.cc/7L3C-F5XM.
• 278Id at *40 (describing a program in the District of Columbia).
• 279Id at *39:

  [M]any failures to appear can be averted by reminding the defendants of their upcoming court appointments. . . . We have had a positive effect on both defendants and the criminal justice system by explaining to these individuals how the system works, answering their questions, and explaining the importance of coming to court.

• 280See Simonson, 115 Mich L Rev at 634 (cited in note 203) (describing appearance rates of defendants whose release is funded by community bail funds).
• 281Uptrust (AngelList), online at http://angel.co/uptrust (visited Nov 10, 2017) (Perma archive unavailable). See also What We Do (Uptrust), archived at http://perma.cc/726W-TNXB (“Through a mix of automation and on-demand human interaction we humanize the confusing criminal justice process for low-income defendants.”).
• 282Russell M. Gold, Jail as Injunction *32–33 (unpublished manuscript, 2017) (on file with author).
• 283Id. See also notes 79–87 and accompanying text.
• 284Professor Rachel Harmon cites a study of defendants who failed to appear after being issued a citation that suggests that this would be a promising approach. Harmon, 115 Mich L Rev at 338 (cited in note 69) (“In one study, over half of the failures to appear were solved by continuing the case for a week and informing the suspect of the new day, with no additional penalty for the initial failure to appear.”). See also Schnacke, Fundamentals of Bail at *104 (cited in note 100) (citing a different study that examined the same question).
• 285Based on survey data collected from 112 of the 150 most populous counties in the United States, the Pretrial Justice Institute (PJI) found that 69 percent of respondents “report that staff of a pre-trial services program or similar entity make an effort to contact defendants and urge them to return to court voluntarily.” Survey of County Pretrial Release Policies at *10 (cited in note 50). Of those jurisdictions that attempt to contact defendants, 68 percent make telephone calls, 36 percent send letters, and 16 percent make home visits. Id. See also Mahoney, et al, Pretrial Services Programs: Responsibilities and Potential at *40 (cited in note 277) (“Pretrial services programs that have established specialized failure-to-appear units universally report that most wayward defendants are not ‘on the lam’ but, rather, can be quickly reached at home or work.”).
• 286The average cost of detaining a defendant pretrial is approximately $19,253. VanNostrand and Keebler, Pretrial Risk Assessment in the Federal Court at *36 (cited in note 208). In New York City, the daily cost of detaining a defendant pretrial is approximately $123 per day. See Cost of Pre-trial Detention in City Jails Takes Bite out of Big Apple’s Budget (NYC Independent Budget Office), archived at http://perma.cc/8DVX-VYY6. See also Wiseman, 123 Yale L J at 1372–74 (cited in note 29) (gathering daily cost statistics and comparing different types of supervision). In the federal prison system, the daily cost of pretrial detention is $73.03. Supervision Costs Significantly Less than Incarceration in Federal System (US Courts, July 18, 2013), archived at http://perma.cc/WD45-B4M2.
• 287See, for example, Vera Institute of Justice, Bail Bond Supervision in Three Counties: Report on Intensive Pretrial Supervision in Nassau, Bronx, and Essex Counties *15–16 (Aug 1995), archived at http://perma.cc/JJ7T-4ZJT. The Vera report cites a 1992 study in which researchers found that, when under intensive supervision, defendants released pretrial had an FTA rate of 0.7 percent. Id at *16. In the same year, a similar study of defendants released from jail “with no consistent supervision pending the disposition of their cases” found that those defendants who were released unsupervised had an FTA rate of 42 percent. Id.
• 288Helland and Tabarrok, 47 J L & Econ at 94 (cited in note 57) (“Defendants who fail to appear impose significant costs on others. Direct costs include the costs of rearranging and rescheduling court dates, the wasted time of judges, lawyers, and other court personnel, and the costs necessary to find and apprehend or rearrest fugitives.”).
• 289Id.
• 290Id at 98.
• 291Harmon, 115 Mich L Rev at 340 (cited in note 69). See also notes 67–71 and accompanying text.
• 292Goldkamp, 11 Crimin & Pub Pol at 431 (cited in note 45) (asserting that these ideas are implied in Flannery and Kretschmar’s study of the Fugitive Safe Surrender program), citing Flannery and Kretschmar, 11 Crimin & Pub Pol at 456 (cited in note 54).
• 293Tabarrok, 11 Crimin & Pub Pol at 466–67 (cited in note 91).
• 294Bierie and Detar, 62 Crime & Delinq at 985 (cited in note 249). See also Flannery and Kretschmar, 11 Crimin & Pub Pol at 439 (cited in note 54).
• 295Bierie and Detar have explained that defendants who are actively hiding from law enforcement to “avoid capture” struggle to meet other “core needs,” including “access to shelter, income, safety, and social or emotional support.” Bierie and Detar, 62 Crime & Delinq at 986 (cited in note 249). Because “traditional avenues to meeting these needs are often inhibited or blocked,” these defendants are forced to seek out alternative, often criminal arrangements, and they “experience increased risk of physical danger and risk of exploitation because . . . [of their] inability to seek protection from police or courts.” Id. See also notes 86–95 and accompanying text.
• 296See, for example, Ark Code Ann § 16-84-203(a) (excusing defendants for failing to appear when prevented by illness or by detention in a jail or correctional facility).
• 297This assumes that justice delayed for a long time imposes greater costs on the community than swift or merely briefly delayed justice. The assumption does not seem particularly controversial because the swiftness of punishment is relevant to both the retributive and the utilitarian goals of punishment.
• 298See Part I.C.
• 299Cohen Email (cited in note 47). See also Cohen and Reaves, Pretrial Release of Felony Defendants in State Courts at *8–10 (cited in note 45) (discussing the “fugitive rate” during their one-year study).
• 300See generally Cohen and Reaves, Pretrial Release of Felony Defendants in State Courts (cited in note 45).
• 301Senate Hearings on Fugitives, 106th Cong, 2d Sess at 1 (cited in note 74). In a 2000 hearing, Senator Strom Thurmond of the Senate Committee on Criminal Justice Oversight explained that “[f]ugitives represent not only an outrage to the rule of law, they are also a serious threat to public safety.” Id.
• 302Harmon, 115 Mich L Rev at 340–41 (cited in note 69).
• 303See Goldkamp, 11 Crimin & Pub Pol at 431 (cited in note 45) (calling for “better descriptive data” about fugitives and observing that “[t]ypes of fugitives can be identified that call possibly for a range of different responses, both preventive and reactive, that target specifically the different problems associated with each type”).
• 304See notes 55, 56, 287 and accompanying text.
• 305Tabarrok, 11 Crimin & Pub Pol at 468 (cited in note 91):

  Despite the difficulty of navigation, the criminal justice system can be unforgiving to those who fail to meet its dictates. Simplifying the process and offering one-stop shopping is not only more just, but it also means that punishment is more swift and certain, a benefit both for the defendants and for society.

• 306Id at 469. See also notes 311–12 and accompanying text.
• 307See Mary T. Phillips, The Past, Present, and Possible Future of Desk Appearance Tickets in New York City *42, 72 (NYC Criminal Justice Agency, Mar 2014), archived at http://perma.cc/M88D-MW5B (describing results from a study of appearance rates for desk appearance tickets issued in New York City, finding a “strong association between FTA and arrest-to-arraignment time,” and concluding that “FTA rates, although already far lower than in previous decades, could be reduced further by scheduling arraignments more quickly following the arrest”).
• 308For example, a 2016 federal class-action lawsuit filed by the Bronx Defenders indicated that the average pending age of misdemeanor cases at the end of 2015 was 827 days. Amended Complaint, Trowbridge v DiFiore, Civil Action No 16-3455, *24 (SDNY filed Jan 23, 2017). This litigation is currently stayed pending settlement negotiations. Trowbridge v. Cuomo (Bronx Defenders, June 27, 2016), archived at http://perma.cc/ML9J-H4AH.
• 309A Nightmare Worthy of Dickens (NY Times, May 12, 2016), online at http://www.nytimes.com/2016/05/12/opinion/a-nightmare-court-worthy-of-dickens.html (visited Nov 11, 2017) (Perma archive unavailable) (characterizing the Bronx criminal courts as a “horribly managed court system that has neither the resources nor the incentive to move any faster”).
• 310Although concern about court backlogs and delays has prompted some speedy-trial reforms, it is not clear how effective these reforms will be. The bail reforms that took effect in New Jersey at the beginning of 2017, for example, included New Jersey’s first speedy-trial rule. In New York, Kalief’s Law, legislation proposed to amend New York’s speedy-trial rules, is currently pending in the state senate. Squadron Passes 1st Step in Speedy Trial Reform through Codes Committee (NY State Senate, June 6, 2017), archived at http://perma.cc/5MPL-S4Q4.
• 311See Tabarrok, 11 Crimin & Pub Pol at 469 (cited in note 91) (concluding that the “popularity” of the Saturday surrender option in the Fugitive Safe Surrender program “indicates that many fugitives have jobs that they do not want to lose” and demonstrates that “a more flexible criminal justice system could better help individuals to reintegrate with civil society”). See also id (describing a 1990s night-court “experiment” in Cook County, Illinois, in which “disposition time fell from 245 days to 86 days, and the number of court dates per case fell from 11 to just over 6”).
• 312See id (“Whether through night courts, weekend courts, or otherwise, simplifying and speeding up the criminal justice system could improve both justice and efficiency.”). See also Summons Reform (cited in note 73) (describing summons reform efforts in New York City, including both greater clarity about how and when to respond to a summons and broadening the “window within which to satisfy the summons”).
• 313CourtCall, a vendor that provides remote-access services for courthouses, explains that its “remote video technology provides motorists the opportunity to save time and costs by being heard remotely and creates greater efficiencies for the courts and law enforcement. More importantly, those contesting traffic citations need not be required to miss work, school or family obligations.” What Is CourtCall (CourtCall), archived at http://perma.cc/H7AL-PPAD (emphasis omitted).
• 314This summer, the New York City Mayor’s Office of Criminal Justice announced a “warrants campaign” to research and implement “the best way to encourage individuals to come to court and clear their warrants.” Warrants Campaign (City Record Online, June 2017), archived at http://perma.cc/C5HK-BTPV.
• 315Flannery and Kretschmar, 11 Crimin & Pub Pol at 439 (cited in note 54) (describing the role of broad amnesty efforts in clearing backlogs and returning some defendants to the process).
• 316Cahill, 11 Crimin & Pub Pol at 476 (cited in note 58) (“Court fees and other fines required to clear a warrant can also represent a financial hardship on some individuals and families, and it could be the root cause of leaving a warrant outstanding.”).
• 317Investigation of the Ferguson Police Department at *3 (cited in note 71).
• 318Id.
• 319Barack Obama, The President’s Role in Advancing Criminal Justice Reform, 130 Harv L Rev 811, 844 (2017) (“We should all be able to agree that the justice system should never be used as a source of revenue. . . . I agree with Attorney General Lynch that this is ‘an unconscionable state of affairs in a nation that outlawed debtors’ prisons in 1833.’”), quoting Attorney General Loretta E. Lynch Delivers Remarks at the Eighth Annual Judge Thomas A. Flannery Lecture (Department of Justice, Nov 15, 2016), archived at http://perma.cc/6CDV-Y29T.