By Benjamin Ewing*
*Assistant Professor, Queen’s University Faculty of Law. Earlier versions of this paper were presented at USC Gould School of Law, Princeton University, the University of Alabama School of Law, Queen’s University Faculty of Law, Yale Law School, Duke Law School, the conference on “Overcriminalization and Indigent Legal Care” at Georgia State University, and the Mid-Atlantic Junior Faculty Forum at the University of Richmond School of Law. For useful comments and fruitful exchanges in those and other contexts, I am particularly grateful to Matt Adler, Michael Adler, Scott Altman, Emad Atiq, Ian Ayres, Jack Balkin, Kate Bartlett, Sara Sun Beale, Charles Beitz, Stuart Benjamin, Jamie Boyle, Curt Bradley, Sam Buell, Guy-Uriel Charles, Walter Dellinger, Deborah DeMott, Mihailis Diamantis, Charles Ewing, Kim Ferzan, John de Figueiredo, Chad Flanders, Lisa Griffin, Ben Grunwald, Paul Haagen, Sharon Harris-Ewing, Eisha Jain, Greg Keating, Alex Kirshner, Doug Kysar, Robert Leider, Marin Levy, Stephen Macedo, Jesse McCarthy, Erin Miller, Philip Pettit, Jed Purdy, Jonathan Quong, Barak Richman, Stephen Sachs, Richard Schmalbeck, Steven Schwarcz, Neil Siegel, Tim Stoll, Tom Tyler, Gary Watson, Ernest Young, my assistant at Duke, Sue Hinson, and my advanced criminal law students at Duke in the spring of 2017. Finally, many thanks to Tori Easton Clark, Rebecca Yung, and the editorial staff of the American Journal of Criminal Law for their work preparing this article for publication.
This paper presents a novel theory of why recidivists appear to have a weaker moral complaint about punishment than first-timers: we implicitly assume that crime and punishment give people valuable “moral” opportunities to reflect upon the sources of their fallibility as agents and take steps to guard against them. This helps to secure them against succumbing to crime, thereby diminishing the reasonableness of their objection to future punishment. Though compelling in theory, this line of thought is problematic in practice because ex-offenders’ opportunities to avoid reoffending are arguably worsened by criminogenic prison conditions and collateral consequences of conviction to a greater extent than they are improved by the moral opportunities inherent in crime and punishment for it. The moral link between ex-offenders’ opportunities and recidivist “premiums” implies we should bolster the former or scale back the latter.
Our response to a crime depends on its perpetrator’s history of offending. We see this in both the administration and the content of criminal law: in exercises of discretion by police, prosecutors, and judges; in laws that single out first-time offenders for special solicitude; in sentencing guidelines that accord significant weight to criminal history in the sentencing recommendations they generate; and in three-strikes laws and similar “habitual offender” statutes dictating exceptionally harsh punishments for serious crimes by offenders with multiple serious past convictions. What could justify such systematic lenience toward first-time offenders and harshness toward recidivists?
Responding to crimes of recidivism with increasing harshness may help us to attain a lower crime rate for any given level of punishment and to avoid punishment that is useless or nearly so (a particularly attractive goal for those who see punishment as a necessary evil that must be justified by instrumental goods it produces). Yet a practice of punishing equally culpable offenders differently based on their race, sex, age, associations, education, employment, or family status—among other variables—might also help us to achieve those valuable ends. And lay intuitions and legal practices accord greater and more systematic significance to criminal history, implying that past convictions are a sturdier basis for differential punishment than other offender characteristics that might be used to estimate an offender’s future dangerousness and otherwise predict the usefulness of punishing him. That suggests the moral justification for treating prior convictions as aggravating does not lie exclusively in the significance of minimizing crime and avoiding useless punishment.
Perhaps recidivists have often had greater legal or moral knowledge, acted with more defiance or premeditation, or shown worse character than first-timers. But treating crimes as necessarily more culpable simply in virtue of being acts of recidivism would require us to adopt a conspicuously broad conception of criminal culpability. And that in turn would cut against the grain of criminal law’s vaunted narrow focus on bad acts rather than bad people—which is reflected in the minimal and occasional roles it accords idiosyncratic differences in (a) offenders’ legal or moral knowledge, (b) exceptionally repugnant, unexpressed motives with which they happen to have committed their crimes, and (c) their free floating bad character over and above the inadequate regard for other people that each of their crimes has itself manifested.
The limits of crime control and the awkwardness of moral culpability as explanations of our treatment of recidivists may not seem troublesome for those inclined toward functionalist explanation and instrumentalist justification. Yet to ground our relative lenience toward first-timers and harshness toward recidivists in a mishmash of prudential generalizations and inchoate sentiments is unsatisfying. Our practice of scaling sanctions to wrongdoers’ prior transgressions is pervasive, intuitive, and consequential and, as such, appears to reflect a coherent and foundational moral ideal.
Tailoring our response to wrongdoing to its perpetrator’s history of offending is pervasive in that it is entrenched across not only the criminal justice systems of the United States and other countries, but also the sanction schemes enforcing the rules of a broad range of public and private institutions across history and around the world. Its pervasiveness reflects and reinforces its intuitive moral appeal, which has been confirmed by surveys of public opinion and of offenders themselves.
Our tendency to take prior wrongs to license enhanced responses to subsequent ones is also enormously consequential. Contrary to popular belief, there is evidence that most felons—perhaps two-thirds—are not reimprisoned after their initial release. But this evidence is consistent with the idea—for which there is also empirical support—that at any given time most felony defendants have prior felony arrests (and very often convictions). A Bureau of Justice Statistics report on felony defendants in the 75 most populated counties in the United States found that in 2009 roughly 62 percent of felony defendants had a prior felony arrest (more than one, in more than 8 in 10 instances) and 43 percent had a prior felonyconviction (more than one, in about 7 of 10 instances). Moreover, the impact of scaling punishment to criminal history is a function not only of the frequency of recidivism but also the magnitude of recidivist sentencing enhancements—which is often substantial. Prior convictions are accorded a place and weight in sentencing that appears unrivaled by any other aggravating or mitigating factors and is sometimes on the same order of magnitude as the seriousness of one’s present offense itself.
Given how pervasive, intuitive, and consequential it is—and how intimately connected our treatment of recidivists is to mass incarceration—the practice of scaling sanctions to an offender’s history of transgressions is woefully undertheorized. Moreover, leading critics of mass incarceration have recently stressed that we cannot hope to end it if we push only for fewer and more lenient punishments for non-serious offenders—particularly non-violent offenders. A similar point can and should be made about recidivists. We succumb to wishful thinking if we suppose that we could undo mass incarceration merely by keeping more first-time offenders out of prison.
The problem, however, is that it is easy to muster sympathy for the plight of “good people” who have made mistakes and difficult to rethink our harshness toward people who commit crimes of violence, people who are seemingly incorrigible recidivists, and especially people who commit multiple violent crimes. Is it possible to make effective use of the special sympathy most people have for first-time offenders without perversely hardening our deep-seated commitment to increasing harshness toward successive acts of recidivism? The most extreme ways of scaling punishment to criminal history—such as California’s three-strikes law prior to a recent reform—have engendered substantial public criticism. But legal theorists who reject the idea of scaling punishment to criminal history altogether are at risk of missing the full force of first-timers’ claims to special sympathy and failing to critique the so-called recidivist “premium” in a way that can speak to the vast majority of people whose support for it is deep and intuitive.
This paper takes seriously the intuitive appeal of scaling punishment to criminal history. And it uses that appeal to mount an internal critique of the practices of scaling punishment to criminal history that we find in the criminal justice systems of America. Unfortunately, there are good reasons to believe those practices fail to do justice to the importance people attribute to second chances, on the one hand, and learning one’s lesson, on the other.
In Part II of the paper, I offer a brief account of why scaling punishment to criminal history might be useful for controlling crime but also why this cannot, alone, fully explain or justify our attitudes and practices toward first-timers and recidivists. The problem is that there are many different distinguishing characteristics of an offender besides his criminal history that likely increase the marginal usefulness of punishing him for controlling crime. Nor can we convincingly fill the gap by arguing that a recidivist’s past crimes somehow necessarily make his present ones more morally culpable in a way that licenses greater state condemnation and punishment. Rather than stretch our conception of criminal culpability and conclude that the state is warranted in condemning and punishing recidivists for non-criminal personal failings over and above their crimes themselves, we do better to start anew: in search of a third determinant of justified punishment beyond crime control and moral culpability.
Part III fills the void by presenting a novel theory of the deep-seated intuition that a wrongdoer has a weaker complaint about a harsh response if she has already transgressed before. The idea behind this theory of “prior convictions as moral opportunities” is that we implicitly view a wrongful act and a process of accountability for it as valuable “moral” opportunities that help us to guard against the external and internal sources of our fallibility as moral agents. Hence, we suppose that repeat offenders have had better opportunities to avoid their second crimes than their first and thus less complaint about punishment for them. This explains well why harsher punishments for subsequent offenses can seem both intuitive in theory and suspect in practice. For whereas some practices of accountability do more than others to buttress the moral opportunities inherent in them, others—including American criminal punishment—impose countervailing constraints on ex-offenders’ opportunities that arguably make it more difficult and costly for them to reform.
In Part IV, I venture from theory to practice and back. First, I consider implications of the theory for such concrete questions of law and policy as whether, why, and how recidivist premiums should depend on (1) the presence of an intervening conviction, (2) the kind of crimes a defendant has committed, (3) the elapsed time between crimes, and (4) the number of previous convictions a defendant has accrued. Second, I discuss reasons for thinking that in America ex-offenders may have poorer opportunities to avoid reoffending than they had to avoid their initial crimes. And I contend that if we are to persist in scaling punishment to criminal history, fairness requires that we take steps to ensure that the moral opportunities arising from prior convictions are not outweighed in significance by criminogenic prison conditions and collateral consequences of a criminal record. Finally, I suggest that the theory of prior convictions as moral opportunities points to a novel and attractively minimalist conception of the rehabilitative ideal. Rather than see rehabilitation as an affirmative aim that justifies punishing people, or some people more than others, we might instead conclude that securing rehabilitative opportunities for ex-offenders is a moral precondition we must satisfy to be warranted in punishing them more harshly the next time around.
 Police officers may and often do, for instance, respond to low-level crimes, such as misdemeanors, with warnings or other forms of “non-enforcement.” See, e.g., Michael K. Brown, Working the Street: Police Discretion and the Dilemmas of Reform 182–220 (1981). Warnings are by nature a less attractive option for repeat offenders.
 See, e.g., Richard S. Frase, The Decision to File Federal Criminal Charges: A Quantitative Study of Prosecutorial Discretion, 47 U. Chi. L. Rev. 246, 261–66 (1980).
 For instance, although the advisory sentencing ranges of the federal sentencing guidelines are themselves determined by the intersection of a defendant’s offense level and criminal history category, the guidelines manual also authorizes judges to depart upward from a guideline range “[i]f reliable information indicates that the defendant’s criminal history category substantially under-represents the seriousness of the defendant’s criminal history . . . .” United States Sentencing Commission, Guidelines Manual § 4A1.3(a)(1) p.s. (Nov. 2016).
 See, e.g., Wash. Rev. Code § 9.94A.650(2) (2018) (“In sentencing a first-time offender [defined in the previous subsection to exclude those who have committed a “violent offense,” “sex offense,” or certain drug or alcohol related offenses] the court may waive the imposition of a sentence within the standard sentence range and impose a sentence which may include up to ninety days of confinement . . . .”); Nora V. Demleitner, Douglas A. Berman, Marc L. Miller & Ronald F. Wright, Sentencing Law and Policy: Cases, Statutes, and Guidelines 361 (2d ed. 2007) (“As the statute from Washington highlights, this tendency to cut first-time offenders a break has been codified in many structured sentencing systems.”).
 See, e.g., USSG Ch.4; Minn. Sentencing Guidelines and Commentary § 2.B (2018).
 “[B]etween 1993 and 1995, 24 States and the Federal Government enacted new laws using the ‘three strikes’ moniker . . . .” John Clark, James Austin & D. Alan Henry, National Institute of Justice, “Three Strikes and You’re Out”: A Review of State Legislation 1 (1997). Clark et al. found that “violent felonies such as murder, rape, robbery, arson, aggravated assault, and carjacking [we]re typically included as strike offenses,” though several states’ provisions have swept more broadly, including certain drug offenses, for instance. Id. at 7.
 In the most obvious and central case, a person sentenced as a “recidivist” or someone with a “criminal history” is someone who committed a crime and was convicted of it before he committed his present offense. However, as Richard Frase has emphasized:
[S]ome recidivist enhancements require an intervening conviction between the ‘prior’ and current offences . . . . [O]ther recidivist enhancements apply with or without an intervening conviction. For example, under the federal, Kansas, Minnesota and Washington guidelines, criminal history is calculated as of the date of sentencing, not the date on which the offence being sentenced . . . was committed . . . .
Richard S. Frase, Prior-conviction Sentencing Enhancements: Rationales and Limits Based on Retributive and Utilitarian Proportionality Principles and Social Equality Goals, in Previous Convictions at Sentencing: Theoretical and Applied Perspectives 117, 119 (Julian V. Roberts & Andrew von Hirsch eds., 2010). When speaking of criminal history and recidivism, unless I indicate otherwise, I have in mind principally—but not exclusively—cases in which a defendant committed a prior crime, and was convicted of it, before the present offense for which his sentencing is at issue.
 See infra Part II.A.
 For an empirical overview of the usefulness of a long list of variables as predictors of recidivism, see J.C. Oleson, Risk in Sentencing: Constitutionally Suspect Variables and Evidence-Based Sentencing, 64 SMU L. Rev. 1329, 1347–68 (2011).
 Recognizing that many factors beyond prior convictions might be used to estimate more precisely an offender’s risk of reoffending, one prominent commentator has argued that criminal history-based sentencing enhancements are a covert way of “punishing dangerousness” that obscures the tradeoff between punishment of the deserving and incapacitation of the dangerous. See Paul H. Robinson, Punishing Dangerousness: Cloaking Preventive Detention as Criminal Justice, 114 Harv. L. Rev. 1429 (2001).
 See generally Julian V. Roberts, Punishing Persistent Offenders: Exploring Community and Offender Perspectives (2008).
 One might argue that a defendant’s criminal history is simply a predictor of dangerousness that is especially salient, familiar, and strong. However, reviewing evidence on potential predictors of recidivism, Oleson suggests that criminal history is high on the list but not clearly an outlier with greater predictive value than such factors as a defendant’s “criminal companions,” “criminogenic needs,” “antisocial personality,” or race. See Oleson, supra note 9, at 1347–68. Criminal history differs, of course, from many other actuarial factors in the degree to which it was once under a defendant’s control through choice. But to appeal to that difference is to give up the game of trying to show that the reasons for favoring criminal history over other predictive offender characteristics in sentencing are merely administrative and epistemic—not moral.
 See infra Part II.B.
 In unusual cases in which the moral or legal wrongfulness of a crime would otherwise be far from obvious, the law sometimes requires that to commit it, a defendant needs to have acted in willful violation of the law, or with “consciousness of wrongdoing.” See Samuel W. Buell & Lisa Kern Griffin, On the Mental State of Consciousness of Wrongdoing, 75 Law & Contemp. Probs. 133 (2012).
 Although criminal guilt does not generally presuppose that an accused acted with any particular motive, there are offences for which some ulterior aim is required, and motive is arguably relevant in a range of direct and indirect ways to various forms of partial exculpation, as well as aggravation and mitigation. See generally Carissa Byrne Hessick, Motive’s Role in Criminal Punishment, 80 S. Cal. L. Rev. 89 (2006) (arguing that motive plays more of a role in the criminal law than is often supposed and should play an even greater one).
 Despite the criminal law’s famed resistance to punishing mere bad character, evidence of character may have some impact at sentencing, particularly in the capital context, in which mitigation inquiries are most protected and expansive. See generally Lockett v. Ohio, 438 U.S. 586, 604 (1978) (plurality opinion) (holding that a capital sentencer generally must be allowed to consider “any aspect of a defendant’s character or record and any of the circumstances of the offense that the defendant proffers as a basis for a sentence less than death”) (emphasis added) and 18 U.S.C. § 3592(a)(8) (2012) (specifying that facts about a defendant’s character are among the potential grounds for mitigation in a capital case).
 Scaling punishment to criminal history may achieve a range of potential penal ends—and perhaps even constitute the subject of an overlapping consensus (in Rawls’s sense), or an incompletely theorized agreement (in Cass Sunstein’s sense), among reasonable citizens whose shared commitment is supported by a range of diverse, non-foundational arguments. See John Rawls, Political Liberalism 133–72 (expanded ed. 2005); Cass R. Sunstein, Incompletely Theorized Agreement, 108 Harv. L. Rev. 1733 (1995).
 See Roberts, supra note 11, at 2–7.
 See id. at 137–84.
 One must be careful about the “recidivism rate” for several reasons. First, when recidivism is defined to include rearrests without convictions, or to focus on reincarceration but even for parole violations, it is much more prevalent than the kind of recidivism about which I am concerned—namely, recidivism as rearrest, conviction, and sentencing for a new offense. See Marie Gottschalk, Caught: The Prison State and the Lockdown of American Politics 101–04 (2015). Second, as John Pfaff has observed, frequently cited data on recidivism compiled by the Bureau of Justice Statistics capture recidivism rates of prisoners released in particular years, which are likely to be much higher than the recidivism rate of all felons ever released from prison. Pfaff notes that a more recent study tracking specific individuals reimprisoned over 13 years “found that only about one-third of all people admitted to prison at any point ended up returning at any later time—not one-half as the BJS reports suggested.” See John F. Pfaff, Locked In: The True Causes of Mass Incarceration—and How to Achieve Real Reform 66–68 (2017) (quoting 67). Cf. Patrick A. Langan & David J. Levin, Bureau of Justice Statistics, Recidivism of Prisoners Released in 1994 (2002), and Matthew R. Durose, Alexia D. Cooper & Howard N. Snyder, Bureau of Justice Statistics, Recidivism of Prisoners Released in 30 States in 2005: Patterns from 2005 to 2010 (2014), with William Rhodes et al., Following Incarceration, Most Released Offenders Never Return to Prison, 62 Crim. & Delinquency 1003 (2016). As Marie Gottschalk notes, “Pew’s findings [from data on prisoners released in 2004 and 1999] were comparable to those of the 2002 Bureau of Justice Statistics recidivism report and of studies of recidivism rates in other developed countries.” Gottschalk, supra note 20, at 102 (citing Pew Center on the States, States of Recidivism: The Revolving Door of America’s Prisons (2011) and Amanda Petteruti & Jason Fenster, Justice Policy Institute, Finding Direction: Expanding Criminal Justice Options by Considering Policies of Other Nations 41–42 (2011)).
 Brian A. Reaves, Bureau of Justice Statistics, Felony Defendants in Large Urban Counties 2009—Statistical Tables 11, 13 (2013).
 For example, the federal sentencing range for a defendant with an offense level of 14 (out of a possible 43) would be as low as 15–21 months of imprisonment if he were in the lowest criminal history category and as high as 37–46 months if he were in the highest criminal history category. A defendant with an offense level of 28 could face a sentencing range of anywhere from 78–97 months to 140–175 months, depending on his criminal history category. See USSG § 5A.
 Two collections of essays on the moral foundations of recidivist punishment have been published in recent years, and their tone is tentative and exploratory. See Recidivist Punishments: The Philosopher’s View (Claudio Tamburrini & Jesper Ryberg eds., 2012); Previous Convictions at Sentencing, supra note 7.
 See especially Pfaff, supra note 20;Gottschalk, supra note 20, at 165–95; James Forman, Jr., Racial Critiques of Mass Incarceration: Beyond the New Jim Crow, 87 N.Y.U. L. Rev. 21, 45–52 (2012).
 Reforms focused on non-serious offenders may redound to the detriment of their “serious” counterparts when scaled-back punishment for the “non-serious” is paired and sold with ramped-up punishment for the “serious.” See, e.g.,Gottschalk, supra note 20, at 167–68.
 A recent ballot measure brought California in line with the standard practice of restricting third strikes to a narrow set of “serious” offenses. See generally J. Richard Couzens & Tricia A. Bigelow, The Amendment of the Three Strikes Sentencing Law 5 (May 2017), http://www.courts.ca.gov/documents/Three-Strikes-Amendment-Couzens-Bigelow.pdf.
 See, e.g., Matt Taibbi, Cruel and Unusual Punishment: The Shame of Three-Strikes Laws, Rolling Stone,March 27, 2013, available at http://www.rollingstone.com/politics/news/cruel-and-unusual-punishment-the-shame-of-three-strikes-laws-20130327.
 Use of the term “recidivist premium” is ubiquitous in discussions of sentencing enhancements based on criminal history and dates back at least as far as to George P. Fletcher, The Recidivist Premium, 1 Crim. Just. Ethics 54 (1982).
 Cf. Frase, Prior-conviction Sentencing Enhancements, supra note 7, at 120 (“Strict theories, rejecting repeat-offender enhancements or positing reduced desert only for offenders with little or no criminal history, would invalidate so much of current sentencing practice that they are likely to be ignored by policymakers and courts.”).