Megan T. Stevenson & Sandra G. Mayson

How dangerous must a person be to justify the state in locking her up for the greater good? The bail reform movement, which aspires to limit pretrial detention to the truly dangerous—and which has looked to algorithmic risk assessments to quantify danger—has brought this question to the fore. Constitutional doctrine authorizes pretrial detention when the government’s interest in safety “outweighs” an individual’s interest in liberty, but it does not specify how to balance these goods. If detaining ten presumptively innocent people for three months is projected to prevent one robbery, is it worth it?

This Article confronts the question of what degree of risk justifies pretrial preventive detention if one takes the consequentialist approach of current law seriously. Surveying the law, we derive two principles: 1) detention must avert greater harm (by preventing crime) than it inflicts (by depriving a person of liberty); and 2) prohibitions against pretrial punishment mean that the harm experienced by the detainee cannot be discounted in the cost-benefit calculus. With this conceptual framework in place, we develop a novel empirical method for estimating the relative personal cost of incarceration and crime victimization that we call relative harm valuation: a survey method that asks respondents to choose between being the victim of certain crimes or being jailed for varying time periods. The results suggest that even short periods of incarceration impose grave harms, such that a person must pose an extremely high risk of serious crime in order for detention to be justified. No existing risk assessment tool is sufficient to identify individuals who warrant detention. The results demonstrate that the stated consequentialist rationale for pretrial detention cannot begin to justify our current detention rates. They suggest that the existing system is instead inflicting pretrial punishment, and they counsel a rethinking of pretrial law and policy.

Introduction

Suppose we can avert an armed robbery by incarcerating ten people for thirty days each. We do not know which of the ten would otherwise commit the crime, and the incarceration is not justified as punishment. Is it worth it? How many people should we be willing to lock up to prevent one future crime?

“None!” you may answer, on the ground that the state may never lock up any person solely to prevent future crime—at least not any person who is a responsible agent with her cognitive faculties intact. We live in a liberal democracy, not a dystopia.¹ 1.See generally Minority Report (20th Century Fox 2002) (depicting dystopian future in which future-criminals are incapacitated before they commit any crime).Show More You may be forgiven; this view has wide currency among thoughtful people.² 2.See, e.g., Laurence H. Tribe, An Ounce of Detention: Preventative Justice in the World of John Mitchell, 56 Va. L. Rev. 371, 376–77 (1970) [hereinafter Tribe, An Ounce of Detention]; Laurence H. Tribe, Trial by Mathematics: Precision and Ritual in the Legal Process, 84 Harv. L. Rev. 1329, 1370 (1971); see also infra notes 78–79 and accompanying text (addressing this perspective).Show More

But your indignation runs counter to the facts and the law. Contrary to common perception, preventive detention is not just the stuff of science fiction. Governments of contemporary liberal democracies routinely engage in preventive detention of many forms. Pretrial detention is one type. Other types include juvenile detention, immigration detention, and manifold variants of short- and long-term civil commitment.³ 3.See infra notes 58–64 and accompanying text.Show More In each of these fields, the government claims authority to deprive people of liberty solely on the basis that custody is necessary to prevent a person from committing future harm.⁴ 4.There are also forms of preventive detention that seek to avert unintentional rather than, or in addition to, willful acts of harm. Examples include quarantine to prevent the spread of communicable disease, as we know all too well, and jury sequestration. For discussions of the law of jury sequestration, see, e.g., Marcy Strauss, Sequestration, 24 Am. J. Crim. L. 63 (1996); James P. Levine, The Impact of Sequestration on Juries, 79 Judicature 266 (1996).Show More The state makes no claim that the person to be detained has forfeited her right to liberty or that the deprivation is deserved. The detention is not punishment. Instead, the detention is “regulatory.” The U.S. Supreme Court has long authorized such practices.⁵ 5.See infra notes 34–43 and accompanying text.Show More Indeed, it is hard to imagine functional governance without them.

Nonetheless, preventive detention is terrifying. It does not adhere to the central constraint on criminal punishment—that it may be imposed only for a past wrongful act.⁶ 6.See H.L.A. Hart, Punishment and Responsibility: Essays in the Philosophy of Law 1, 23 (2d ed. 2008) (explaining how this constraint on punishment “maximizes individual freedom within the coercive framework of law”); Douglas Husak, Overcriminalization: The Limits of the Criminal Law (2008) (describing conceptual constraints on punishment).Show More The justification for preventive detention is merely “risk,” and risk is amorphous. So the central question for any preventive detention regime is what kind and degree of risk is sufficient to justify the detention at issue. If we incarcerate people who have a 20% chance of otherwise committing an assault during the period of detention, for instance, we can expect to prevent one assault for every five detentions. Is such detention justified? How much liberty should we sacrifice to prevent one crime?

As is, there is nothing approaching a consensus answer to this question.⁷ 7.See generally Christopher Slobogin, A Jurisprudence of Dangerousness, 98 Nw. U. L. Rev. 1 (2003) (urging scholars and courts to develop more coherent standards for preventive deprivations of liberty).Show More Courts and legislatures routinely assert the regulatory power of the state to detain those who pose a great enough risk, but even in long-standing preventive detention regimes, the relevant legal standards are vague at best.⁸ 8.See infra Section I.A.Show More Generations of scholars have lamented the lack of legal guidance. Few have offered specific guidance themselves.⁹ 9.See Alan Dershowitz, The Origins of Preventive Confinement in Anglo-American Law — Part I: The English Experience, 43 U. Cin. L. Rev. 1, 60 (1974) (“We have not even begun to ask these kinds of questions, or to develop modes of analysis for answering them.”).Show More The problem is that the question requires an explicit tradeoff between liberty and security, values that are infrequently measured and difficult to compare.

Difficulties notwithstanding, the bail reform movement has now placed the question of what risk justifies preventive detention squarely at center stage.¹⁰ 10.See Sandra G. Mayson, Dangerous Defendants, 127 Yale L.J. 490, 494 (2018) [hereinafter Mayson, Dangerous Defendants] (“[Bail reform] holds great promise, but also raises an extremely difficult question: what probability that a person will commit unspecified future crime justifies detention . . . ?”).Show More Jurisdictions around the country are forsaking money bail in favor of more intentional decisions about pretrial custody. The new systems aspire to detain those arrested persons who pose a true threat and release everyone else on appropriate conditions.¹¹ 11.See Stephanie Wykstra, Bail Reform, Which Could Save Millions of Unconvicted People from Jail, Explained, Vox (Oct. 17, 2018, 7:30 AM), https://www.vox.com/future-perfect/2018/10/17/17955306/bail-reform-criminal-justice-inequality [https://perma.cc/V3Q​S-J69G].Show More Flight risk is also a concern in the pretrial context, but a distinctly secondary one in practice.¹² 12.Samuel R. Wiseman, Pretrial Detention and the Right to Be Monitored, 123 Yale L.J. 1344, 1351 (2014) (“Historically, the U.S. system of bail and associated pretrial detention was employed solely to prevent pretrial flight, but increasingly, the many individuals awaiting trial in jail are detained because a judge has deemed them potentially dangerous.”). For a thoughtful discussion of the various kinds of risk, often lumped together as “flight risk,” see Lauryn P. Gouldin, Defining Flight Risk, 85 U. Chi. L. Rev. 677 (2018).Show More The aspiration to limit detention to the dangerous requires each pretrial system to decide what kind of threat justifies detention. The advent of statistical risk assessment has crystallized the question further by forcing courts and stakeholders to deal in quantified probabilities and to confront the limits of prediction.¹³ 13.John Monahan & Eric Silver, Judicial Decision Thresholds for Violence Risk Management, 2 Int’l J. of Forensic Mental Health 1, 6 (2003) (“The necessity for choosing a decision threshold for risk management decisions, long implicit in clinical risk assessment, is made apparent in actuarial prediction.”).Show More Every jurisdiction that authorizes pretrial detention, and every court that imposes it, must decide what degree of risk warrants depriving a person of liberty.

This Article tackles the question of when pretrial detention is warranted to prevent future crime.¹⁴ 14.Cf. Mayson, Dangerous Defendants, supra note 10, at 557–60 (reserving judgment on the degree of risk that justifies preventive detention). This Article does not address the power of courts to detain an accused person who has violated a court-imposed condition of release.Show More Whereas the great bulk of prior scholarship on pretrial detention has focused on the shortcomings of current law,¹⁵ 15.E.g., Albert W. Alschuler, Preventive Pretrial Detention and the Failure of Interest-Balancing Approaches to Due Process, 85 Mich. L. Rev. 510 (1986); Chalmous G. Reemes, United States v. Salerno: The Validation of Preventive Detention and the Denial of a Presumed Constitutional Right to Bail, 41 Ark. L. Rev. 697, 719–22 (1988).Show More we take existing law as a given. This is not to endorse existing law as representing the best possible policy approach to detention. The project, rather, is to take existing legal doctrine seriously and to ask when detention meets the law’s criteria. We present a conceptual framework for answering the question and then a novel empirical method for implementing the framework.

The conceptual framework is a straightforward consequentialist one. Constitutional law authorizes pretrial detention when the government’s interest in safety “outweighs” the individual’s interest in liberty.¹⁶ 16.See infra notes 29–44 and accompanying text.Show More On our reading, this doctrine establishes a simple cost-benefit framework.¹⁷ 17.See infra Section I.B.Show More In order to be justified in those terms, pretrial detention must, at minimum, avert more harm than it inflicts. The most significant harms at stake are the cost of crime to the potential crime victim and the cost of pretrial detention to the detainee. Within this calculus, prohibitions against pretrial punishment mean that the well-being of the arrestee must be fully considered. The challenge is thus to develop a direct measure of the relative harms of incarceration and crime.

To meet the challenge, the Article deploys a novel form of contingent valuation that we call “relative harm valuation” (“RHV”).¹⁸ 18.We developed this concept and conducted our first study in 2017 but learned in the spring of 2020 that others have used the same method in other contexts. Most notably, the legal scholars Jane Bambauer and Andrea Roth have used a similar survey method to estimate when carceral punishment becomes “excessive” for constitutional purposes. See infra note 119. Conversations with Bambauer and Roth were valuable in refining our approach. We consider the existence of these other efforts to be a strength of the present study rather than a weakness. The other studies to have deployed RHV have also documented a surprising degree of aversion to incarceration or involuntary commitment among a sizable portion of respondents. See infra note 141.Show More It aims to estimate the relative harm of incarceration versus crime victimization while avoiding some of the distortions that plague traditional cost-benefit and contingent-valuation methods. Our method is intentionally simple, and it echoes John Rawls’ famous notion that the principles of justice are those that a rational person would choose behind a “veil of ignorance” as to her own traits and position in society.¹⁹ 19.John Rawls, A Theory of Justice 118 (2d ed. 1999).Show More Adapting his effort to detach normative analysis from self-interest, we conduct a survey that requires respondents to compare the costs of detention and crime directly, imagining themselves as both detainee and as crime victim. We ask questions like, “How much time in jail is as bad as being the victim of a burglary?” and, “If you had to choose between spending a month in jail or being the victim of a burglary, which would you choose?”²⁰ 20.This method is a variant of the survey technique formally known as “contingent valuation,” which has provided most of the commonly used estimates for the costs of crime. See infra notes 103–07 and accompanying text.Show More

The survey results suggest that people view incarceration as an incredibly harmful experience. Most would choose crime-victimization over even short jail stints. The median respondent says that a single day in jail is as costly as a burglary, that three days are as costly as a robbery, and that a month in jail is as costly as an aggravated assault. Notably, these results are very consistent across race, gender, and socioeconomic class of the respondent. They are also quite similar for those who have personal experience with incarceration or crime victimization. Aversion to incarceration is broad and widespread.

By focusing on costs to the crime victim and incarcerated person, our method puts aside indirect and fiscal costs. However, we expect these to accrue on both sides of the ledger. Crime can create fear in the community, but so can incarceration. Law enforcement and prosecution are expensive to the state, but so is incarceration. Including such costs might make the analysis more nuanced, but we do not think it would meaningfully change the main result: that even a short period of incarceration inflicts very grave harms.

The severity of the harm that incarceration inflicts (according to our median respondent) means that preventive detention can only be justified on consequentialist grounds if there is a very high risk that the person would otherwise commit serious crime. Jailing a person for thirty days is justifiable only if it is expected to prevent crimes at least as harmful as a serious assault. Jailing someone for just one day is justifiable only if it averts crime as serious as burglary. These risk thresholds are higher than we can meet with statistical evidence. In studies of one widely used risk assessment tool, for instance, even defendants in the highest risk group have only a 2.5% chance of rearrest for a violent offense within a month.²¹ 21.See Thomas Blomberg, William Bales, Karen Mann, Ryann Meldrum & Joe Nedelec, Ctr. for Criminology and Pub. Pol’y Rsch., Validation of the COMPAS Risk Assessment Classification Instrument 47 tbl.8 (2010).Show More We would have to detain forty such people for one month each, not just one person, to expect to avert one violent offense.

Given the high risk-threshold for preventive detention and the limits of our predictive abilities, pretrial detention on the basis of dangerousness should be rare. But it is not. On any given day, almost 500,000 people are held in jails awaiting trial.²² 22.Zhen Zeng, Bureau of Just. Stat., Jail Inmates in 2017, 1 fig.1, 5 tbl.3 (2019), www.bjs.gov/content/pub/pdf/ji17.pdf [https://perma.cc/NPM3-NP84].Show More Many more cycle through pretrial detention each year.²³ 23.Id. at 1.Show More A significant number of these detentions may be the unintentional result of a court setting money bail that the accused cannot afford.²⁴ 24.E.g., Sandra G. Mayson, Detention by Any Other Name, 69 Duke L.J. 1643, 1653 (2020) [hereinafter Mayson, Detention by Any Other Name] (citing statistics regarding detention on money bail).Show More A much smaller number may be justified on the basis of flight risk—a ground for detention that this paper does not address.²⁵ 25.But see Wiseman, supra note 12, at 1349 (arguing that detention is rarely necessary to manage flight risk given advancing surveillance technologies).Show More Yet the centrality of public-safety discourse in the growing backlash to bail reform efforts demonstrates that crime risk dwarfs flight risk, in the view of both courts and the public, as a concern in the pretrial phase.²⁶ 26.See, e.g., Jesse McKinley, The Bail Reform Backlash that Has Democrats at War, N.Y. Times (Feb. 14, 2020), https://www.nytimes.com/2020/02/14/nyregion/new-york-bail-reform.html [https://perma.cc/NU9F-B3RZ]; see also, e.g., H.R. 81, 2020 Leg., Reg. Sess. (Ala. 2020) (proposing amendment to state constitution to permit pretrial detention for dangerousness).Show More The focus on crime risk suggests that a substantial portion of the millions of people who cycle through jails each year are there because they were perceived to be dangerous.

There are many possible explanations for the dramatic gap between theory and practice. The most likely, we surmise, is that current practice reflects an implicit discounting of the value of detainees’ well-being relative to the well-being of potential crime victims. This might be because accused people are viewed as criminals who have forfeited the right to liberty; because accused people are disproportionately Black, brown, and poor while the paradigmatic crime victim in the public imagination is white and wealthy; because pretrial detention is assumed to be credited against legitimate punishment imposed after conviction; or all of the above.

Some of these grounds for discounting the welfare of arrestees are easier to dismiss than others. The most difficult ground to dismiss is the idea that arrestees are not entitled to the same concern as crime victims because they are not wholly innocent; they are in some manner culpable for having created the risk at issue. As one of us has written elsewhere, this notion runs headlong into the presumption of innocence and prohibition on pretrial punishment, foundational principles of the American legal order.²⁷ 27.Mayson, Dangerous Defendants, supra note 10, at 537–38.Show More It is extremely difficult to reconcile those principles with the idea that the state can discount the welfare of arrestees on the basis of their (probable) guilt. Yet the intuition that the state may treat accused persons as having impaired moral status is strong, and in some circumstances, it seems unjust not to discount an arrestee’s welfare relative to a person the arrestee is credibly alleged to have threatened.

This Article does not resolve the conflict between the prohibition on pretrial punishment and the human impulse to discount the welfare of arrestees in a cost-benefit calculus. Rather, it demonstrates that a rigorous consequentialist analysis raises deep questions about how the law ought to value individual liberty and welfare, questions that echo across many fields of law. It also demonstrates that, left unexamined, consequentialist rationales can mask decision-making processes that rely on judgments of worth or that are dictated by perverse incentives. Confronting these processes will be important to the long-term success of pretrial reform.

This Article makes four contributions. The first is to fully articulate the consequentialist conceptual framework for detention decisions that current law entails. The second is the method we devise to apply that framework: relative harm valuation, which allows for the comparison of intangible harms without resorting to the distorting intermediary of dollars. This Article’s third contribution is the information the survey reveals: Even short periods of jail detention impose harms as grave as serious crimes. The logical corollary is that if we value the liberty of accused people and crime victims by a common standard, pretrial detention for the purpose of preventing crime is almost never warranted on cost-benefit grounds. Finally, in illuminating the chasm between the cost-benefit rationale for pretrial detention and our actual practices, this Article highlights the need for policymakers, courts, and bail reformers to grapple with the retributive impulse and institutional incentives that shape detention practice on the ground.

This Article proceeds in three Parts. Part I describes the legal doctrine that authorizes pretrial preventive detention on cost-benefit grounds. It extrapolates the consequentialist conceptual framework that this doctrine implies, then explains why existing empirical methods are inadequate to weigh the harm of criminal victimization against the harm of incarceration. Part II presents our relative-harm-valuation surveys and explains the results. Part III explores the implications of the survey results for pretrial policy and beyond.

1. See generally Minority Report (20th Century Fox 2002) (depicting dystopian future in which future-criminals are incapacitated before they commit any crime). ↑
2. See, e.g., Laurence H. Tribe, An Ounce of Detention: Preventative Justice in the World of John Mitchell, 56 Va. L. Rev. 371, 376–77 (1970) [hereinafter Tribe, An Ounce of Detention]; Laurence H. Tribe, Trial by Mathematics: Precision and Ritual in the Legal Process, 84 Harv. L. Rev. 1329, 1370 (1971); see also infra notes 78–79 and accompanying text (addressing this perspective). ↑
3. See infra notes 58–64 and accompanying text. ↑
4. There are also forms of preventive detention that seek to avert unintentional rather than, or in addition to, willful acts of harm. Examples include quarantine to prevent the spread of communicable disease, as we know all too well, and jury sequestration. For discussions of the law of jury sequestration, see, e.g., Marcy Strauss, Sequestration, 24 Am. J. Crim. L. 63 (1996); James P. Levine, The Impact of Sequestration on Juries, 79 Judicature 266 (1996). ↑
5. See infra notes 34–43 and accompanying text. ↑
6. See H.L.A. Hart, Punishment and Responsibility: Essays in the Philosophy of Law 1, 23 (2d ed. 2008) (explaining how this constraint on punishment “maximizes individual freedom within the coercive framework of law”); Douglas Husak, Overcriminalization: The Limits of the Criminal Law (2008) (describing conceptual constraints on punishment). ↑
7. See generally Christopher Slobogin, A Jurisprudence of Dangerousness, 98 Nw. U. L. Rev. 1 (2003) (urging scholars and courts to develop more coherent standards for preventive deprivations of liberty). ↑
8. See infra Section I.A. ↑
9. See Alan Dershowitz, The Origins of Preventive Confinement in Anglo-American Law — Part I: The English Experience, 43 U. Cin. L. Rev. 1, 60 (1974) (“We have not even begun to ask these kinds of questions, or to develop modes of analysis for answering them.”). ↑
10. See Sandra G. Mayson, Dangerous Defendants, 127 Yale L.J. 490, 494 (2018) [hereinafter Mayson, Dangerous Defendants] (“[Bail reform] holds great promise, but also raises an extremely difficult question: what probability that a person will commit unspecified future crime justifies detention . . . ?”). ↑
11. See Stephanie Wykstra, Bail Reform, Which Could Save Millions of Unconvicted People from Jail, Explained, Vox (Oct. 17, 2018, 7:30 AM), https://www.vox.com/future-perfect/2018/10/17/17955306/bail-reform-criminal-justice-inequality [https://perma.cc/V3Q​S-J69G]. ↑
12. Samuel R. Wiseman, Pretrial Detention and the Right to Be Monitored, 123 Yale L.J. 1344, 1351 (2014) (“Historically, the U.S. system of bail and associated pretrial detention was employed solely to prevent pretrial flight, but increasingly, the many individuals awaiting trial in jail are detained because a judge has deemed them potentially dangerous.”). For a thoughtful discussion of the various kinds of risk, often lumped together as “flight risk,” see Lauryn P. Gouldin, Defining Flight Risk, 85 U. Chi. L. Rev. 677 (2018). ↑
13. John Monahan & Eric Silver, Judicial Decision Thresholds for Violence Risk Management, 2 Int’l J. of Forensic Mental Health 1, 6 (2003) (“The necessity for choosing a decision threshold for risk management decisions, long implicit in clinical risk assessment, is made apparent in actuarial prediction.”). ↑
14. Cf. Mayson, Dangerous Defendants, supra note 10, at 557–60 (reserving judgment on the degree of risk that justifies preventive detention). This Article does not address the power of courts to detain an accused person who has violated a court-imposed condition of release. ↑
15. E.g., Albert W. Alschuler, Preventive Pretrial Detention and the Failure of Interest-Balancing Approaches to Due Process, 85 Mich. L. Rev. 510 (1986); Chalmous G. Reemes, United States v. Salerno: The Validation of Preventive Detention and the Denial of a Presumed Constitutional Right to Bail, 41 Ark. L. Rev. 697, 719–22 (1988). ↑
16. See infra notes 29–44 and accompanying text. ↑
17. See infra Section I.B. ↑
18. We developed this concept and conducted our first study in 2017 but learned in the spring of 2020 that others have used the same method in other contexts. Most notably, the legal scholars Jane Bambauer and Andrea Roth have used a similar survey method to estimate when carceral punishment becomes “excessive” for constitutional purposes. See infra note 119. Conversations with Bambauer and Roth were valuable in refining our approach. We consider the existence of these other efforts to be a strength of the present study rather than a weakness. The other studies to have deployed RHV have also documented a surprising degree of aversion to incarceration or involuntary commitment among a sizable portion of respondents. See infra note 141. ↑
19. John Rawls, A Theory of Justice 118 (2d ed. 1999). ↑
20. This method is a variant of the survey technique formally known as “contingent valuation,” which has provided most of the commonly used estimates for the costs of crime. See infra notes 103–07 and accompanying text. ↑
21. See Thomas Blomberg, William Bales, Karen Mann, Ryann Meldrum & Joe Nedelec, Ctr. for Criminology and Pub. Pol’y Rsch., Validation of the COMPAS Risk Assessment Classification Instrument 47 tbl.8 (2010). ↑
22. Zhen Zeng, Bureau of Just. Stat., Jail Inmates in 2017, 1 fig.1, 5 tbl.3 (2019), www.bjs.gov/content/pub/pdf/ji17.pdf [https://perma.cc/NPM3-NP84]. ↑
23. Id. at 1. ↑
24. E.g., Sandra G. Mayson, Detention by Any Other Name, 69 Duke L.J. 1643, 1653 (2020) [hereinafter Mayson, Detention by Any Other Name] (citing statistics regarding detention on money bail). ↑
25. But see Wiseman, supra note 12, at 1349 (arguing that detention is rarely necessary to manage flight risk given advancing surveillance technologies). ↑
26. See, e.g., Jesse McKinley, The Bail Reform Backlash that Has Democrats at War, N.Y. Times (Feb. 14, 2020), https://www.nytimes.com/2020/02/14/nyregion/new-york-bail-reform.html [https://perma.cc/NU9F-B3RZ]; see also, e.g., H.R. 81, 2020 Leg., Reg. Sess. (Ala. 2020) (proposing amendment to state constitution to permit pretrial detention for dangerousness). ↑

27. Mayson, Dangerous Defendants, supra note 10, at 537–38. ↑