Young and Dangerous: The Role of Youth in Risk Assessment Instruments

States are increasingly adopting risk assessment instruments (RAIs) to help judges determine the appropriate type and length of punishment for an offender. Although this sentencing practice has been met with a wide variety of scholarly criticism, there has been virtually no discussion of how RAIs treat youth as a strong factor contributing to a high risk score. This silence is puzzling. Not only is youth undoubtedly the most powerful risk factor in most RAIs, but youth also holds a special place in the criminal justice system as a “mitigating factor of great weight.” This Comment presents the first in-depth critique of RAIs with respect to their treatment of youth. It argues that, as currently designed and implemented, RAIs both contradict longstanding and widespread views about the proper role of youth as a factor in punishment and undermine efforts to craft proportionate sentences consistent with principles of justice and modern social science.


Risk assessment instruments (RAIs) are becoming increasingly influential in courtrooms throughout the country.1Sonja B. Starr, The Risk Assessment Era: An Overdue Debate, 27 Fed. Sent’g Rep. 205, 205 (2015) (“[W]e are already in the risk assessment era.”). These tools are also commonly referred to as “risk assessment tools,” “actuarial risk assessments,” or simply “risk assessments.” Developed from the crude psychological assessments of the 1930s, RAIs use actuarial science to calculate the likelihood of an individual committing a crime in the future.2See Sandra G. Mayson, Bias In, Bias Out, 128 Yale L.J. 2218, 2221–22, 2222 n.5 (2019). These tools are now used across the criminal justice system—informing decisions about bail, the length and nature of the sentence, and parole.3Starr, supra note 1, at 205. Many jurisdictions have already adopted RAIs for one or more of these purposes, and the evidence suggests that more are likely to follow.4See infra notes 24–28 and accompanying text. The popularity of RAIs, however, belies their problematic consequences: RAIs have the perverse effect of imposing longer sentences on young offenders than on older offenders for the same conduct.5See Megan T. Stevenson & Jennifer L. Doleac, Algorithmic Risk Assessment in the Hands of Humans, SSRN 3–4 (Apr. 21, 2021), []. Such a result is inconsistent with centuries of criminal justice doctrine regarding the punishment of youth and may well frustrate contemporary advancements in criminal justice policy.6See infra Part III.

The appeal of these tools is obvious. RAIs, if accurate and used correctly (two big ifs), could make huge advances toward a more effective, accurate, and equal criminal justice system.7See generally Jordan M. Hyatt, Mark H. Bergstrom & Steven L. Chanenson, Follow the Evidence: Integrate Risk Assessment into Sentencing, 23 Fed. Sent’g Rep. 266 (2011). Proponents argue that RAIs can decrease crime by helping judges effectively identify—and thus separate—the most dangerous individuals from the general population.8Id. At the same time, RAIs reveal which individuals are less likely to commit crimes in the future, allowing judges to recommend more lenient sentences for these offenders.9Id. Proponents also argue that RAIs promote accuracy and equality by providing judges with objective, statistically supported information to guide their subjective risk assessments.10Id.

Despite their potential benefits, RAIs have not been immune to criticism. In fact, a sizable body of literature challenges RAIs’ promises of scientific objectivity and accuracy.11See, e.g., Julia Angwin, Jeff Larson, Surya Mattu & Lauren Kirchner, Machine Bias, ProPublica (May 23, 2016), []; Julia Dressel & Hany Farid, The Accuracy, Fairness, and Limits of Predicting Recidivism, 4 Sci. Advances art. eaao5580 (2018), Critics argue that, rather than avoiding the subjectivity of human discretion, RAIs codify and veil the biased judgments and data on which they base their statistical analyses.12See Angwin et al., supra note 11. Critics point to studies showing that RAIs falsely identify Black men as future criminals at twice the rate that white men are falsely identified.13Id. In addition to these concerns about objectivity, critics also raise doubts about the accuracy of RAIs more generally. For example, software engineer Julia Dressel and digital-forensics scholar Hany Farid concluded that a popular RAI was “no more accurate or fair than the predictions of people with little to no criminal justice expertise.”14Dressel & Farid, supra note 11, at 3.

Amid the crossfire between empiricists, another line of criticism has emerged. This critique, driven mostly by law professors and scholars, focuses not on RAIs’ predictions but on the means by which these tools make their determinations.15See, e.g., Sonja B. Starr, Evidence-Based Sentencing and the Scientific Rationalization of Discrimination, 66 Stan. L. Rev. 803 (2014). RAIs use statistical correlations between crime and characteristics such as age and sex to determine the likelihood of a specific individual with these characteristics committing certain crimes in the future.16Erin Collins, Punishing Risk, 107 Geo. L.J. 57, 64 (2018). Because these calculations are used to make decisions concerning bail, sentencing, and parole, the result is that an individual’s case can be significantly affected by personal traits in jurisdictions that use RAIs.17Starr, supra note 15, at 804–05. Such treatment sits uncomfortably with our idealized notions of impartial justice; capitalizing on these intuitions, scholars have urged states to reject RAIs as violative of the Equal Protection Clause and for being bad public policy.18Id. RAIs’ consideration of sex, employment status, income, education, dependence on government assistance, job skills, criminal history, family criminality, and other traits have been criticized on these grounds.19Id. at 870 (“The inclusion of demographic and socioeconomic variables in risk prediction instruments that are used to shape incarceration sentences is normatively troubling and, at least with respect to gender and socioeconomic variables, very likely unconstitutional.”); Melissa Hamilton, Back to the Future: The Influence of Criminal History on Risk Assessments, 20 Berkeley J. Crim. L. 75, 133 (2015) (“Actuarial risk assessment has unfortunately amplified the impact of criminal records, and, using questionable sources, can qualify as evidence of past offending. As illuminated by the Back to the Future theme, the potential reconstruction of an individual’s prior record often may have the unfortunate effect of altering an individual’s future.”).

But there has been virtually no discussion of RAIs’ use of youth in legal or empirical scholarship.20Only one other article has directly addressed the issue. See Megan T. Stevenson & Christopher Slobogin, Commentary, Algorithmic Risk Assessments and the Double-Edged Sword of Youth, 96 Wash. U. L. Rev. 681 (2018). While Professors Stevenson and Slobogin presents a compelling picture of the “double-edged” nature of youth, this Comment further explores these inconsistencies through a historical analysis of the criminal justice system’s punishment of young people. The literature’s silence on the matter is particularly surprising in light of the extensive body of law that has developed over the last several centuries surrounding the punishment of young people.21See infra Part II. And, considering that youth is the most predictive factor in a risk assessment score,22See Stevenson & Slobogin, supra note 20, at 691–93. a conversation about the advisability of RAIs is incomplete without a thorough analysis of their particular effects on young people and their (in)compatibility with our system’s longstanding practices regarding the punishment of youth.

This Comment fills this gap by arguing that the use of RAIs that treat youth as an aggravating factor in sentencing decisions contradicts our justice system’s well-established commitment to treating young people more leniently. Continued application of these tools, as currently designed and implemented, also replicates dangerous mistakes of the past. While RAIs are used throughout the criminal justice system, this Comment will focus primarily on their use in sentencing. An analysis of the sentencing stage provides a particularly compelling picture of the criminal justice system’s otherwise-tenderhearted treatment of youth and thus offers an exceptionally stark illustration of the hypocrisies relating to RAIs’ treatment of youth as an aggravating factor. These issues are not unique to sentencing, however, and the tools used at sentencing are often the very same tools used at bail or parole.23For example, the COMPAS algorithm is used for both pretrial and sentencing determinations. Farhan Rahman, COMPAS Case Study: Fairness of a Machine Learning Model, Towards Data Sci. (Sept. 7, 2020), []; see also discussion infra Section I.A. As such, this Comment’s arguments have broader implications for the other uses of RAIs as well.

Part I provides an overview of the structure, development, and use of RAIs. It also explains the role of youth in RAI calculations. Part II then traces society’s evolving understanding of the proper role of youth in punishment, beginning with the common law infancy defense and progressing to the science-based understanding we have today. Lastly, Part III analyzes the ways that RAIs’ treatment of youth as an aggravating factor conflicts with principles concerning the proper role of youth as a factor in punishment.

I.        Background on RAIs

In 1994, Virginia became the first state to adopt an RAI for use in sentencing.24Starr, supra note 15, at 809. By 2019, twenty-eight states had adopted a similar kind of RAI.25Risk Assessment Tool Led to Harsher Sentences for Young and Black Defendants, Equal Just. Initiative (Nov. 25, 2019), []. There is reason to think that there will be even more expansion in the future: politicians, sentencing commissioners, and other policymakers have argued in support of RAIs.26Collins, supra note 16, at 60. So have organizations like the National Center for State Courts and the American Law Institute.27Id. As sentencing scholars Mirko Bagaric and Gabrielle Wolf stated, “[i]t seems inevitable that transparent and validated risk assessment tools will progressively be used in sentencing.”28Mirko Bagaric & Gabrielle Wolf, Sentencing by Computer: Enhancing Sentencing Transparency and Predictability, and (Possibly) Bridging the Gap Between Sentencing Knowledge and Practice, 25 Geo. Mason L. Rev. 653, 680 (2018).

This Part provides an overview of RAIs and considers how youth affects an offender’s risk of recidivism. Section I.A describes the different kinds of RAIs, from simple checklists to complex machine-learning algorithms. Section I.B summarizes how judges use RAIs to make sentencing decisions. Section I.C reviews the biological differences between young people and older people and assesses how these differences contribute to higher risk scores for the former.

A.     Structure and Development of RAIs

While most RAIs consider youth, the structure of these tools differs greatly. The most basic are essentially checklists, most often consisting of seven to fifteen factors.29Sandra G. Mayson, Dangerous Defendants, 127 Yale L.J. 490, 509–11 (2018). Each factor is associated with a number of points depending on that factor’s association with future crime.30Id. at 509. The points are then totaled into a cumulative risk score.31Id. Some RAIs further categorize risk scores into low-, medium-, or high-risk groups, while others have a cutoff score under which an individual is recommended for alternative sentencing.32Id. at 513; see also, e.g., Va. Crim. Sent’g Comm’n, Re-validation of the Nonviolent Offender Risk Assessment Instrument: Study Update 3, []. When RAIs categorize offenders’ scores into risk groups, judges typically receive only the offender’s overall risk category, not the risk factors or calculations used to generate these labels.33See Collins, supra note 16, at 63–68.

One example of a checklist-style RAI is Virginia’s Nonviolent Offender Risk Assessment (NVRA). The NVRA was developed by the Virginia Criminal Sentencing Commission to “identify drug and property offenders who were at the lowest risk of committing a new crime.”34Brandon L. Garrett, Alexander Jakubow & John Monahan, Va. Crim. Just. Pol’y Reform Project, Nonviolent Risk Assessment in Virginia Sentencing: The Sentencing Commission Data 3 (2018),‌/news/spr18/UVA%20Law%20NVRA%20Sentencing%20Analysis%20and%20Judicial%20Survey,%20March%201,%202018.pdf []. Its goal was to divert these low-risk offenders from prison to noncarceral interventions, like outpatient drug or mental health care programs.35Id. Potential alternative punishments include “jail, release, probation, community service, outpatient substance-abuse treatment, or electronic monitoring.” Id. The NVRA uses a variety of factors—such as offense type, sex, age at time of offense, and prior felony convictions—to determine the offender’s total score.36Va. Crim. Sent’g Comm’n, supra note 32, at 3. Those who score thirty-eight points or fewer are recommended for an alternative punishment, while those who score thirty-nine points or more are not.37Id.

By contrast, the most advanced RAIs are developed through machine-learning techniques in which computers use data to repeatedly revise their own predictive algorithms.38See Richard Berk & Jordan Hyatt, Machine Learning Forecasts of Risk to Inform Sentencing Decisions, 27 Fed. Sent’g Rep. 222, 223–24 (2015). The way these tools work is less clear. Although the programmer selects the training data, the factors to use, and the outcomes to predict, the algorithm determines the rest.39See id. It decides how heavily to weigh each factor, as well as how the factors interact with one another.40See id.

One example of a machine-learning RAI is the Correctional Offender Management Profiling for Alternative Sanctions (COMPAS). The COMPAS was developed by a corporation called Northpointe and has been adopted in multiple states, including New York, Wisconsin, California, and Florida, for both pretrial and sentencing determinations.41Rahman, supra note 23. The COMPAS algorithm uses the results of a 137-item questionnaire to evaluate an individual’s risk of recidivism.42Id. Once calculated, an individual’s risk is assigned a risk score between one and ten—one being the least likely to commit another crime and ten being the most likely.43Northpointe, COMPAS Risk & Need Assessment System (2012), http://www‌ []. A score of one indicates that the offender’s score is in the bottom ten percent of all scores, a score of two indicates that the offender’s score is in the bottom twenty percent of all scores, and so on.44Id.

It is important to mention that because the COMPAS was developed by a corporation, its algorithm is proprietary and subject to trade secret protections.45Stevenson & Slobogin, supra note 20, at 690. While Northpointe (now known as Equivant) has voluntarily released the risk factors used in the COMPAS, it has not explained how these factors are used to calculate an individual’s risk of recidivism.46See Risk Scores: The Not-So-Secret Recipe, equivant (Aug. 14, 2020), []. In 2016, Eric Loomis challenged this lack of transparency, arguing that the use of the COMPAS in the determination of his sentence violated his due process rights, but the Wisconsin Supreme Court rejected his claim.47State v. Loomis, 881 N.W.2d 749 (Wis. 2016).

B.     Use of RAIs in Sentencing

Sentencing procedure is mostly structured by the legislature.486 Wayne R. LaFave, Jerold H. Israel, Nancy J. King & Orin S. Kerr, Criminal Procedure § 26.1(a) (4th ed. 2015). Due to the few constitutional restraints on sentencing, legislatures decide which punishments can be imposed, who can impose them, and what guidance the sentencing authority receives.49Id. Statutes authorizing incarceration often impose mandatory minimum and maximum sentences for specific crimes.50Id. § 26.1(c); see also Charles Doyle, Cong. Rsch. Serv., R45074, Mandatory Minimum Sentencing of Federal Drug Offenses 6–7 (2018) (listing mandatory maximum and minimum sentences for various federal drug offenses). Some statutes, however, authorize the sentencing authority to depart from the typical range when certain mitigating or aggravating factors are present in the case.51LaFave et al., supra note 48, § 26.5(b); see also, e.g., 21 U.S.C. §§ 841(b), 851 (increasing the statutory maximum when the defendant is found to have committed a prior “felony drug offense”).

Typically, the trial judge determines an offender’s punishment.52LaFave et al., supra note 48, § 26.2(a). Some states give judges broad discretion to impose any sentence within the provided statutory range, but others have adopted certain mechanisms to limit judicial discretion.53Id. § 26.1(a). To make their sentencing decisions, judges rely on information provided in a presentence report.54Id. § 26.5(b). Presentence reports are prepared by probation officers after an in-depth interview with the offender.55Id. They contain information about the offender’s background, prior criminal record, medical evaluations, employment history, and financial circumstances.56Id. In jurisdictions that use RAIs, an offender’s risk score is usually included in the presentence report, often without any guidance on how the judge should use this information.57Collins, supra note 16, at 66. The presentence report has been called “the single most important document at both the sentencing and correctional levels of the criminal process.”58Stephen A. Fennell & William N. Hall, Due Process at Sentencing: An Empirical and Legal Analysis of the Disclosure of Presentence Reports in Federal Courts, 93 Harv. L. Rev. 1613, 1623 (1980).

RAIs can impact two major sentencing decisions. First, RAIs can influence what type of punishment an offender receives.59Collins, supra note 16, at 69–72. For example, depending on the risk score of the individual, the NVRA recommends that judges either sentence the offender to prison or to an alternative punishment.60See Va. Crim. Sent’g Comm’n, supra note 32. While this recommendation is discretionary, studies show that low risk scores do, in fact, increase an offender’s likelihood of receiving an alternative punishment, and high risk scores increase an offender’s likelihood of being sent to prison.61Garrett et al., supra note 34, at 3 (finding that 42.2 percent of low-risk offenders received alternative sentences compared to only 23.4 percent of high-risk offenders).

Second, RAIs can affect a judge’s decision regarding the length of an offender’s punishment.62Collins, supra note 16, at 67–69. Some jurisdictions authorize judges to exceed sentencing-guideline recommendations for offenders who pose a high risk of recidivism.63Id. at 67–68. One such jurisdiction is Kansas, which has adopted RAIs for the sentencing of juvenile offenders.64Id. Under the Kansas Sentencing Commission guidelines, low- or medium-risk juveniles may be sentenced to up to fifteen months, while high-risk juveniles may be sentenced to up to eighteen months.65Id.

Even in jurisdictions that do not authorize departures from sentencing maximums for high risk scores, empirical studies suggest that a high risk score nonetheless induces judges to give offenders longer sentences.66See Starr, supra note 15, at 867–70. Some of this happens subconsciously through a psychological phenomenon known as the anchoring effect.67See J.J. Prescott & Sonja Starr, Improving Criminal Jury Decision Making After the Blakeley Revolution, 2006 U. Ill. L. Rev. 301, 325–33. Anchoring occurs when an initial data point, even if completely arbitrary, provides a reference point for all subsequent judgments.68Id. When making the sentencing decision with an RAI, judges may start with an initial anchor—namely, the sentence they would have given absent the risk score—and increase or decrease the length of punishment in the direction of the risk score.69Collins, supra note 16, at 68. A high risk score may thus result in a sentence length significantly longer than the anchor, while a low risk score may result in a sentence length significantly shorter than the anchor.

Furthermore, judges tend to give RAIs additional weight in sentencing because of their perceived objectivity, legitimacy, and determinacy. Sentencing is often a high-stakes, emotional process that involves the complicated balancing of many indeterminate factors.70E.g., Starr, supra note 15, at 866; Thomas M. Hardiman, Foreword, 49 Duq. L. Rev. 637, 637 (2011) (“Balancing the principles animating the criminal law—incapacitation, retribution, rehabilitation, and deterrence—is difficult in and of itself, and the inherent subjectivity of the enterprise makes the task even more challenging.”). Of these factors, risk predictions can be one of the most difficult to accurately assess. Judges might be hesitant to heavily weigh considerations that cannot be neatly measured when conducting their own risk calculations.71Starr, supra note 15, at 865–66. RAIs, however, provide judges with concrete and seemingly precise assessments of risk. Their scientific “legitimacy” can persuade judges to rely on them more heavily.72Id. (noting that judges overweigh other “expert” evidence as well). At the very least, judges might be more likely to rely on RAI risk scores because they are more determinate than other sentencing factors.73See id.

The trend towards automation is likely to result in harsher punishments being imposed on young offenders. Because judges use an offender’s risk to determine the type and length of punishment, the weight that RAIs give to youth can increase the likelihood and length of incarceration for young offenders.74See infra Section I.C. After Virginia adopted an RAI that heavily weighed age at time of offense as an aggravating factor, offenders younger than twenty-three were 4 percent more likely to be incarcerated than older offenders.75Stevenson & Doleac, supra note 5, at 3–4. Their sentences were also 13 percent longer than those of their older peers.76Id. Note that these figures reflect the fact that judges sentenced young offenders less harshly than recommended by their risk scores. Full compliance with the sentencing recommendations would have resulted in a 15 percent increase in the probability of incarceration and a 60 percent increase in the length of incarceration. Id. For example, an older offender sentenced to three years in prison would have received, on average, four extra months in prison if the offender had been younger than twenty-three. To put it differently, young offenders in Virginia often spend more time in prison because of the adoption of an RAI.

C.     Youth’s Role in RAIs

This Section discusses the relationship between youth and RAIs’ main focus, namely, assessing recidivism risk. Before proceeding, however, a few terminological clarifications are needed. This Comment uses the words “youth” and “young people” in accordance with their scientific, rather than legal, meanings. While the legal age of majority in most states is eighteen,77Alexandra O. Cohen, Richard J. Bonnie, Kim Taylor-Thompson & B.J. Casey, When Does a Juvenile Become an Adult? Implications for Law and Policy, 88 Temp. L. Rev. 769, 770 (2016). The age of majority is the legal age of adulthood. Id. this bright-line rule is merely a crude approximation of what science has made clear: there are characteristics associated with youth that justify treating younger people more leniently than older people.78See infra Part II. Recent studies have concluded that many of these characteristics remain present in individuals well into their twenties.79See infra Section II.C.1. Thus, scientific discourse uses youth to refer to the lower end of a sliding scale of developmental, neurological, and emotional maturity. In light of these two conceptions of youth—the scientific sliding-scale definition and the legal bright-line definition—this Comment will employ different terms to refer to each. When discussing the former, more indeterminate category, this Comment will refer to “young people” (in contrast to “older people”). When discussing bright-line legal definitions, this Comment will refer to “juveniles” (in contrast to “adults”).

1.       The Age-Crime Curve

Youth is a highly influential aggravating factor in many RAIs, including the COMPAS and the NVRA. According to a study reverse-engineering the COMPAS algorithm, age alone explains 57 percent of the variation in scores indicating risk of violent recidivism, “substantially more” than any other factor.80Stevenson & Slobogin, supra note 20, at 691–93. Equivant admits as much on their website, which states that “‘age at first arrest’ is weighted the heaviest.”81Risk Scores: The Not-So-Secret Recipe, supra note 46. Similarly, the NVRA assigns an individual thirteen points for being younger than thirty when the crime was committed.82Va. Crim. Sent’g Comm’n, supra note 32. For comparison, the tool only assigns nine extra points if the individual has had five or more adult incarcerations.83Id.

In addition to the fact that age heavily factors into a high risk score, age may also correspond to other risk factors, essentially “double-counting” the offender’s age. For example, the NVRA assigns nine points to an offender who is not regularly employed during the two years prior to arrest.84Id. But, as young people are expected to go to school rather than work, young people are less likely to be regularly employed and less likely to have been regularly employed for long periods of time. Thus, the offender’s age is counted twice: once for being younger than thirty at the time of the offense and a second time for being not regularly employed. In this case, age contributes twenty-two points to the young offender’s risk score, representing well over half of the points needed to be disqualified for alternative punishment under the NVRA.85Id.

Youth is so heavily weighted in RAIs due to its strong correlation with criminal behavior.86Jeffery T. Ulmer & Darrell Steffensmeier, The Age and Crime Relationship: Social Variation, Social Explanations, in The Nurture Versus Biosocial Debate in Criminology 377, 378 (Kevin M. Beaver, J.C. Barnes & Brian B. Boutwell eds., 2015) (“It is now a truism that age is one of the strongest factors associated with criminal behavior.”). This correlation, referred to by sociologists as the “age-crime curve,” demonstrates that all types of crime rise quickly during adolescence, peak around late adolescence or early adulthood, and rapidly decline thereafter, leveling off slowly well into adulthood.87Id. at 389. Although some variation exists regarding the parameters of the age-crime curve, sociologists have discovered similarly shaped relationships between crime and age across numerous societies and time periods.88Id. at 380–81.

The age-crime curve can be explained by both biological and social factors. Neuroscientific evidence, such as noninvasive brain imaging and postmortem studies, indicates that different regions of the brain develop at different paces during adolescence and early adulthood.89See Mariam Arain et al., Maturation of the Adolescent Brain, 9 Neuropsychiatric Disease & Treatment 449, 451–52 (2013). Specifically, the prefrontal cortex—responsible for judgment and self-control—does not fully develop until an individual’s mid-twenties, whereas the subcortical limbic regions—responsible for desires and fear—develop mostly between the ages of thirteen and seventeen.90Cohen et al., supra note 77, at 783. This asymmetric development results in a decreased ability to inhibit inappropriate desires, emotions, and feelings.91Id. When faced with high social or emotional pressure, “the limbic regions of the brain may hijack less mature prefrontal regions leading to an imbalance or overreliance on these emotional regions.”92Id. at 784. As a result, when young people experience negative emotional stress, they demonstrate diminished cognitive capacity when compared to adults.93Id. at 786. Thus, young people are inherently more susceptible to criminal activity.

Additionally, adolescence and early adulthood are both periods of significant identity exploration, increasing young people’s willingness to engage in risky behaviors like crime.94Jeffrey Jensen Arnett, Emerging Adulthood: A Theory of Development from the Late Teens Through the Twenties, 55 Am. Psych. 469, 473–75 (2000). Generally speaking, parent and teacher surveillance decreases during this time, and adolescents and young adults have not yet formed new attachments to significant others, children, or occupations that impose restraining influences.95Michael Rocque, Chad Posick & Justin Hoyle, Age and Crime, Encyclopedia of Crime and Punishment 5 (Oct. 2, 2015), This instability provides young people with the opportunity to experiment with their identities in a way that is impossible during other age periods.96Arnett, supra note 94, at 475. Part of this experimentation might involve engaging in criminal activity.97Elizabeth S. Scott & Laurence Steinberg, Rethinking Juvenile Justice 53 (2008). Psychologists believe that this feature of young adulthood explains why certain risky behaviors, like unprotected sex, substance abuse, and driving while intoxicated, peak between the ages of eighteen and twenty-five.98Arnett, supra note 94, at 474–75.

But as young people enter into their twenties, they settle down, commit to partners, have kids, and take on steady employment.99Id. at 473–75. Research indicates that these prosocial bonds strongly decrease an individual’s likelihood to engage in criminal activity.100Rocque et al., supra note 95, at 5. Additionally, young people tend to reach neurological maturity during this time, specifically in the prefrontal cortex region.101Arain et al., supra note 89, at 452. Thus, both biological and social factors contribute to a natural tendency of young people desisting from crime after early adulthood.102See, e.g., Michael Massoglia & Christopher Uggen, Settling Down and Aging Out: Toward an Interactionist Theory of Desistance and the Transition to Adulthood, 116 Am. J. Socio. 543 (2010). Only about five percent of young offenders continue to commit crimes into adulthood.103Scott & Steinberg, supra note 97, at 53.

2.       Implications for Criminal Justice Policies

Young people’s natural desistance has powerful implications for criminal justice policies regarding youth. First, sentences that keep young offenders incarcerated long after they’ve aged out of their prime crime years are unlikely to prevent future crime and thus may be unnecessary and undesirable.104Rocque et al., supra note 95, at 6. Second, incarceration of young offenders may actually increase criminality by interfering with young offenders’ natural tendency to “age out” of their crimes.105Id. Young offenders are less likely to make prosocial connections with adult role models or other law-abiding peers while incarcerated.106Ian Lambie & Isabel Randell, The Impact of Incarceration on Juvenile Offenders, 33 Clinical Psych. Rev. 448, 451 (2013); Tirza A. Mullin, Note, Eighteen Is Not a Magic Number: Why the Eighth Amendment Requires Protection for Youth Aged Eighteen to Twenty-Five, 53 U. Mich. J.L. Reform 807, 816 (2020). Additionally, they are less likely to develop an identity, sense of purpose, or self-awareness in a prison environment.107Mullin, supra note 106, at 816. Failure to make prosocial connections and develop a sense of identity or purpose increases a young offender’s risk of recidivism once released back into society.108Id. Thus, while it may be tempting to address young people’s increased risk of reoffending by locking them up, such measures will likely be counterproductive. Instead, alternative punishments (like those recommended for low-risk offenders by the NVRA) may be more effective in reducing recidivism among young offenders.

In light of young offenders’ unique qualities, the criminal justice system has created an extensive system of juvenile justice that deemphasizes incarceration and incapacitation in favor of rehabilitation and leniency. Part II discusses this system in greater detail.

II.      The Past and Present of Punishing Youth

The criminal justice system has a widespread and longstanding policy of treating young people more leniently than older people. Although society’s views on how best to deal with young offenders have changed over time, scholars, judges, and lawmakers have consistently emphasized that youth should mitigate punishment. Only during the superpredator era—a deplorable period in the late 1980s and 1990s characterized by mass hysteria over a predicted rise in juvenile violent crime—were these principles temporarily abandoned. Today, there is widespread acknowledgement that the superpredator theory and its resulting reforms were misguided and destructive. As a result, scholars, judges, and lawmakers have reverted to the previously established policy of lenient punishments for youth. Section II.A surveys society’s understanding of youth’s relationship to criminal behavior at common law and through the development of the juvenile justice system. Section II.B then examines the superpredator era. Section II.C observes modern-day views concerning the punishment of young offenders, finding support for leniency in recent legislative reforms and constitutional case law.

A.     The History of the Juvenile Justice System

American criminal law has always recognized a fundamental distinction between juveniles and adults. As early as the fourteenth century,109Emily Buss, What the Law Should (and Should Not) Learn from Child Development Research, 38 Hofstra L. Rev. 13, 19 (2009). the common law recognized an infancy defense that afforded significant protections to young people of all ages.110Craig S. Lerner, Originalism and the Common Law Infancy Defense, 67 Am. U. L. Rev. 1577, 1583, 1587–94 (2018). The scope of the defense depended on the kind of crime at issue in the case. In capital cases, the defense created the presumption that children lacked criminal capacity, meaning that children were presumed incapable of possessing the requisite criminal intent to warrant punishment.111Id. at 1590–94. This presumption was irrebuttable for those under seven and rebuttable for those between seven and fourteen.112Id. In misdemeanor cases, the infancy defense was even more expansive, offering a complete defense for individuals up to the age of twenty-one in some situations.113Id. at 1588–90.

The infancy defense was the sole protection for juvenile offenders during the early years of American history.114See Buss, supra note 109, at 12. However, some scholars did contemplate additional protections. For example, in debates over the imposition of the death penalty on juvenile offenders, Sollom Emlyn, a prolific eighteenth-century legal writer, argued that the death penalty was inappropriate because of young people’s potential for “reformation.”1151 Matthew Hale, The History of the Pleas of the Crown 25 n.t (Sollom Emlyn ed., London, E. & R. Nutt & R. Gosling 1736) (“[Y]et in the common instances of larceny and stealing some other punishment might be found, which might leave room for the reformation of young offenders.”). Thus, during the first few centuries of American history, criminal law and scholars acknowledged two distinctions between juveniles and adults that justified leniency towards the former: juveniles’ decreased culpability and their potential for reformation.

Starting in the late nineteenth century, conceptions of juvenile culpability changed as the Progressive Era brought widespread social reform to all aspects of society, including criminal law.116See Buss, supra note 109, at 20–21. The Progressives adopted a positivist view of criminal behavior.117Andrew M. Carter, Age Matters: The Case for a Constitutionalized Infancy Defense, 54 U. Kan. L. Rev. 687, 719–20 (2006). They believed that crime was a product not of intrinsic moral failings but of extrinsic factors that could be addressed through rehabilitation.118Id. at 720. In accordance with these views, the Progressives advocated for a separate justice system for juveniles, one where criminal law and procedure would have very limited influence.119See Lara A. Bazelon, Note, Exploding the Superpredator Myth: Why Infancy Is the Preadolescent’s Best Defense in Juvenile Court, 75 N.Y.U. L. Rev. 159, 171 (2000). Under this system, juvenile courts would treat offenders as misguided yet fundamentally innocent—in need of proper supervision, treatment, and control rather than punishment.120See Ellen Ryerson, The Best-Laid Plans: America’s Juvenile Court Experiment 36 (1978) (“Perhaps the most influential idea in shaping the juvenile court system was the thesis that the defect which produced juvenile crime lay not so much in the child as in the environment from which he had come and, therefore, that no child should be treated as a criminal.”); Elizabeth S. Scott & Laurence Steinberg, Essay, Blaming Youth, 81 Tex. L. Rev. 799, 804 (2003). The state would act as juveniles’ guardians, promoting their wellbeing while using individualized treatments to direct them on a law-abiding path.121Bazelon, supra note 119, at 172. One of the fundamental tenets of American juvenile justice is parens patriae, the principle that the state is authorized to intervene to protect people who are unable to care for themselves. See Buss, supra note 109, at 23. The first juvenile court was established in Illinois in 1899,122Lisette Blumhardt, Comment, In the Best Interests of the Child: Juvenile Justice or Adult Retribution?, 23 U. Haw. L. Rev. 341, 343 (2000). and a little over a decade later, all but two states had created their own juvenile alternatives to criminal courts.123Id.

But the Progressives’ vision for the juvenile justice system never really took hold. After the mid-twentieth century, the juvenile justice system underwent an identity crisis of sorts as scholars and lawmakers argued over its future.124Scott & Steinberg, supra note 120, at 805–06. States had grown disillusioned with the rehabilitation model, and lawmakers passed statutes to reformulate the goals of the juvenile justice system to include accountability and protection of the public.125Id. The public abandoned the rehabilitation model because “little evidence suggested that it worked to reduce crime.” Id. at 805 n.27. Additionally, because of the focus on rehabilitation, juveniles in the juvenile justice system were not given access to certain rights that adults received in the more punitive adult system, leading some to argue for the abolition of the juvenile justice system. See, e.g., Barry C. Feld, Abolish the Juvenile Court: Youthfulness, Criminal Responsibility, and Sentencing Policy, 88 J. Crim. L. & Criminology 68 (1997). And, recognizing that juvenile courts had become more punitive, the Supreme Court issued a series of decisions applying procedural protections from the adult system to juveniles, eroding some of the distinctions between the two court systems.126In Kent v. United States, the Supreme Court noted that juvenile offenders often receive the “neither the protections accorded to adults nor the solicitous care and regenerative treatment postulated for children.” 383 U.S. 541, 556 (1966). The Court went on to issue a series of decisions holding that certain constitutional protections were required in juvenile proceedings. See, e.g., In re Gault, 387 U.S. 1, 33–34, 41, 55, 57 (1967) (extending rights to notice, counsel, confrontation, and fair and impartial hearing, as well as protection against self-incrimination, to juvenile offenders).

Despite this new emphasis on public safety and procedural protections, the juvenile justice system still viewed offenders as having diminished capacity and focused on sentencing these offenders in proportion to their culpability.127Even after Gault, the Court still embraced the idea of “compassionate justice” in juvenile courtrooms. Bazelon, supra note 119, at 175. Although it extended some constitutional protections to juveniles, it refused to transform the juvenile system into “a fully adversary process [that would] put an effective end to what has been the idealistic prospect of an intimate, informal protective proceeding.” McKeiver v. Pennsylvania, 403 U.S. 528, 545 (1971) (plurality opinion). It also continued to recognize juvenile offenders’ potential for growth, focusing on providing juveniles with services designed to prepare them for adult life.128Scott & Steinberg, supra note 120, at 806. But the juvenile justice system was soon to face a strong challenge to its founding commitments to proportionate, rehabilitative, and individualized sentencing. Starting in the late 1980s, mass hysteria involving unsubstantiated fears of juvenile violence overtook the nation, prompting large-scale legislative reform and altering society’s long-held views of juvenile punishment.

B.     The Superpredator Era

In the late 1980s and 1990s, spurred by an increase in juvenile violent crimes,129See Barry C. Feld, Bad Kids: Race and the Transformation of the Juvenile Court 201 (1999) (stating that the juvenile violent-crime arrest rate increased 67.3 percent between 1986 and 1995). Although few disputed that juvenile violent crime rose during this time, there was substantial debate over whether these statistics actually suggested a new generation of juvenile offenders. Compare Franklin E. Zimring, The Youth Violence Epidemic: Myth or Reality?, 33 Wake Forest L. Rev. 727 (1998) (attributing the increase to a rise in crimes committed with handguns and changing police standards of what constitutes an assault or aggravated assault), with Judy Briscoe, Breaking the Cycle of Violence: A Rational Approach to At-Risk Youth, Fed. Prob., Sept. 1997, at 3, 3 (attributing the rise to an increase in children “born into and reared in abject moral poverty”). widespread panic over a predicted rise of a particularly cruel and violent breed of juvenile offenders overshadowed all other discussions of juvenile justice.130See Scott & Steinberg, supra note 120, at 807 (describing the superpredator era as a “moral panic, in which the media, politicians, and the public reinforce each other in an escalating pattern of alarmed reaction to a perceived social threat”). In a 1995 article, social scientist John DiIulio labeled these offenders “super-predators,” describing them as “morally impoverished” juveniles who grew up “surrounded by deviant, delinquent, and criminal adults in abusive, violence-ridden, fatherless, Godless, and jobless settings.”131John J. DiIulio, Jr., The Coming of the Super-Predators, Wkly. Standard, Nov. 27, 1995, at 23, 25 []. Other scholars reinforced this depiction, writing that superpredators were “radically impulsive, brutally remorseless youngsters . . . who murder, assault, rape, rob, burglarize, deal deadly drugs, join gun-toting gangs, and create serious communal disorders.”132William J. Bennett, John J. DiIulio, Jr. & John P. Walters, Body Count: Moral Poverty . . . and How to Win America’s War Against Crime and Drugs 27 (1996). They were incorrigible,133See, e.g., Grim Reality Check on Youth Crime, Chi. Trib. (Jan. 31, 1996), [] (“The violence of their crimes and the apparent hardness of their hearts mock our notions of ‘treatment’ and ‘rehabilitation.’”). undeterrable,134Briscoe, supra note 129, at 4 (“[M]any of today’s persistent young offenders cannot be deterred from committing crimes simply by toughening the criminal penalties. . . . In fact, for many youth, going to prison is a badge of honor or a rite of passage.”). and deadly.135Criminologist James Wilson warned of “the prospect of innocent people being gunned down at random, without warning and almost without motive, by youngsters who afterward show us the blank, unremorseful faces of seemingly feral, presocial beings.” James Q. Wilson, What to Do About Crime, Commentary, Sept. 1994, at 25, 26 []. Unless something was done, this “demographic time bomb” would result in a “bloodbath of teenage violence.”136Richard Zoglin, Now for the Bad News: A Teenage Time Bomb, Time (Jan. 15, 1996),,33009,983959,00.html []. Experts predicted that the population of juveniles would increase over the next fifteen years, “bod[ing] disastrously for society.” Briscoe, supra note 129, at 3–4. According to some, the increase in juveniles, particularly male juveniles, would result in a doubling of the number of juvenile arrests for murder, rape, robbery, and aggravated assault between 1996 and 2010. John J. DiIulio Jr., Opinion, Stop Crime Where It Starts, N.Y. Times (July 31, 1996), [].

The media soon adopted these problematic tropes, saturating American viewers with images of violent and morally depraved juvenile offenders.137Perry L. Moriearty, Framing Justice: Media, Bias, and Legal Decisionmaking, 69 Md. L. Rev. 849, 851–52 (2010). In many ways, the media had been primed for the dissemination of the superpredator theory. First, the media began adopting an “infotainment” approach to reporting in the early 1990s, which prioritized celebrity and scandal and deemphasized policy and data. Id. at 860. Second, the media started focusing much more on crime reporting. Id. at 861. The 1989 “Central Park Jogger” case provided the groundwork for the media’s later obsession with juvenile crime; by the 1990s, crime was the most covered topic on the evening news. Id. at 861–64. Five years after “superpredator” was first introduced into the American lexicon, the term was used nearly three hundred times by forty major news outlets,138Carroll Bogert & Lynnell Hancock, Superpredator: The Media Myth That Demonized a Generation of Black Youth, Marshall Project,‌/superpredator-the-media-myth-that-demonized-a-generation-of-black-youth []. including in articles such as “Moral Poverty” by the Chicago Tribune and “‘Superpredators’ Arrive: Should We Cage the New Breed of Vicious Kids?” by Newsweek.139Moral Poverty, Chi. Tribune (Dec. 15, 1995), []; Peter Annin, ‘Superpredators’ Arrive: Should We Cage the New Breed of Vicious Kids?, Newsweek, Jan. 22, 1996, at 57 [‌/GTW9-CFBF]. Print media was not the only source advancing this narrative; popular talk shows, prime-time news programs, and television dramas all used real-life juvenile violent crimes as inspiration for entertainment.140Bazelon, supra note 119, at 165–66 (compiling sources). Even after juvenile crime began declining, news coverage continued to explode.141Lori Dorfman & Vincent Schiraldi, Building Blocks for Youth, Off Balance: Youth, Race & Crime in the News 18 (2001), []. According to one study surveying news outlets between 1987 and 1996, coverage of juvenile delinquency increased thirty-fold during that period.142Id.

The media’s incessant portrayals of juvenile violence reinforced the superpredator narrative, magnifying the perceived threat juvenile offenders posed to the public. A 1997 poll by the Los Angeles Times showed that 80 percent of respondents indicated that the media increased their personal fear of being victimized by crime.143Greg Braxton, Ratings vs. Crime Rates, L.A. Times (June 4, 1997, 12:00 AM), https://‌ []. In another study conducted in 1999, 82 percent of respondents felt that youth “d[id] not have as strong a sense of right and wrong” as fifty years earlier.144Meg Bostrom, Frameworks Inst., The 21st Century Teen: Public Perception and Teen Reality 6 (2001) []. The public’s belief that juveniles were dangerous and morally immature likely influenced its preference for punitive responses to juvenile offending. Indeed, in 1995, the vast majority of Americans believed that juveniles should be tried as adults if charged with a serious violent or property crime or with selling illegal drugs.145Id. at 32 (listing the results at 87 percent, 70 percent, and 63 percent, respectively).

Lawmakers bought in as well, employing the term “superpredator” and reinforcing the same fears about increasing juvenile crime rates.146See, e.g., Dole Seeks to Get Tough on Young Criminals, L.A. Times (July 7, 1996, 12:00 AM), [‌/K34U-S65J] (reporting that then presidential candidate Bob Dole warned that “[u]nless something is done soon, some of today’s newborns will become tomorrow’s super-predators”); User Clip: Hillary Clinton on ‘Superpredators’ in 1996, C-SPAN (Feb. 25, 2016), (“They are not just gangs of kids anymore. They are often the kinds of kids that are called superpredators.”). Furthermore, legislatures passed a series of reforms designed to make it easier to funnel juveniles into the adult justice system.147Elizabeth S. Scott, Speech, Miller v. Alabama and the (Past and) Future of Juvenile Crime Regulation, 31 Law & Ineq. 535, 537 (2013). States expanded the discretionary judicial waiver process, lowering the age and increasing the range of offenses for which juveniles could be transferred.148Id. States also created two new forms of transfer: mandatory transfer and prosecutorial transfer.149Janet C. Hoeffel, The Jurisprudence of Death and Youth: Now the Twain Should Meet, 46 Tex. Tech L. Rev. 29, 38 (2013). The former made transfer to the adult system mandatory for certain crimes, while the latter empowered prosecutors to choose whether to try juveniles in the adult or juvenile justice systems.150Id. at 50, 60–61. As a result of these reforms, the number of juveniles confined in adult facilities increased from approximately 1,600 in 1988 to more than 9,000 in 1997.151Moriearty, supra note 137, at 853.

The superpredator theory is now widely viewed as a mistake—and for good reason.152See, e.g., Peter Nicholas, Bernie Sanders Says Bill Clinton Owes Americans an Apology, Wall St. J. (Apr. 9, 2016, 11:42 PM), [‌/XE6Q-7ABJ]; Kim Taylor-Thompson, Opinion, Why America Is Still Living with the Damage Done by the ‘Superpredator’ Lie, L.A. Times (Nov. 27, 2020, 4:00 AM),‌/opinion/story/2020-11-27/racism-criminal-justice-superpredators []; Clyde Haberman, When Youth Violence Spurred ‘Superpredator’ Fear, N.Y. Times (Apr. 6, 2014), []. Despite a significant increase in the overall size of the juvenile population, juvenile crime rates actually began to decline in the mid-1990s, and by 2009, juvenile crime rates were the lowest since 1980.153Charles Puzzanchera & Benjamin Adams, U.S. Dep’t of Just., Juvenile Arrests 2009, at 8 (2011), []. The same phenomenon occurred with juvenile homicide rates: during the decade from 2000 to 2009, there were fewer juvenile homicide arrests than in the four years between 1992 and 1995, the year the term “superpredator” was coined.154Id. at 9. Scholars now agree that the decrease in juvenile crime rates was not attributable to the reforms enacted during the superpredator era.155See, e.g., Elizabeth S. Scott & Laurence Steinberg, Social Welfare and Fairness in Juvenile Crime Regulation, 71 La. L. Rev. 35, 37 (2010) (noting that studies examining these measures concluded that there was “little support for the claim that the declining crime rates [were] largely due to the enactment of harsher laws”); Franklin E. Zimring & Stephen Rushin, Did Changes in Juvenile Sanctions Reduce Juvenile Crime Rates? A Natural Experiment, 11 Ohio St. J. Crim. L. 57, 59 (2013) (conducting an empirical study which “calls into question” the theory that superpredator era reforms decreased juvenile crime rates). Rather, as the Surgeon General concluded in 2001, the initial increase in juvenile violent crimes during the late 1980s and early 1990s was actually the result of “increased lethality result[ing] from gun use, which ha[d] since decreased dramatically.”156U.S. Pub. Health Serv., U.S. Dep’t of Health & Hum. Servs., Youth Violence: A Report of the Surgeon General 5 (2001) [].

Even the superpredator theory’s most vocal proponents have now retracted their support. John DiIulio, the social scientist that first labeled these offenders “superpredators,” told the New York Times in 2001 that he had tried “to put the brakes on the superpredator theory,” but he “couldn’t write fast enough to curb the reaction.”157Elizabeth Becker, As Ex-theorist on Young ‘Superpredators,’ Bush Aide Has Regrets, N.Y. Times (Feb. 9, 2001), []. DiIulio later joined others in filing an amicus brief in support of the petitioners in Miller v. Alabama,158567 U.S. 460 (2012); see also infra Section II.C. stating that “the predictions by the proponents of the juvenile superpredator myth . . . were wrong.”159Brief of Jeffrey Fagan et al. as Amici Curiae in Support of Petitioners at 37, Miller, 567 U.S. 460 (No. 10-9646), 2012 WL 174240. The brief concluded that, although the laws passed during this era “threw thousands of children into an ill-suited and excessive punishment regime,” they “had no material effect on the subsequent decrease in crime rates.”160Id. at 37.

But the superpredator theory had an even more pernicious effect. In addition to needlessly subjecting juvenile offenders to adult prison environments where they were more likely to be assaulted161Scott & Steinberg, supra note 155, at 68 (noting that juveniles in prison are ten times more likely to report being sexually assaulted than juveniles in juvenile facilities). and deprived of necessary developmental opportunities,162Id. (“Further, with little in the way of education, occupational training, or rehabilitation, many prisons provide minimal positive structure for inmates’ daily lives. In these facilities, much time is spent in cells or in the prison yard with other prisoners under the surveillance of guards on the perimeter.”). the theory created what has now been called the “superpredator virus.”163Moriearty, supra note 137, at 887–90. This “virus” is an unconscious and racialized belief about which juveniles are most likely to commit crimes.164Id. It “installed [itself] into viewers’ racial schemas, where, without their knowledge, the messages increased the viewers’ implicit biases about age, race, and causes of and solutions to violent crime.”165Id. at 889. During the superpredator era, proponents crafted the narrative of the dangerous Black juvenile, linking youth and Blackness with violence and moral depravity.166See id. at 865. DiIulio himself referred to the predicted rise of juvenile crime as a “[B]lack crime problem” and estimated that “as many as half of these juvenile super-predators could be young [B]lack males.”167John J. DiIulio, Jr., My Black Crime Problem, and Ours, City J., Spring 1996, https://‌ []. The media also played a prominent role in enforcing this stereotype: local news programs across the country overrepresented juveniles of color as criminals and underrepresented them as victims of crime.168Dorfman & Schiraldi, supra note 141, at 13. As a result, the superpredator era was especially harmful for juveniles of color, who were much more likely to be arrested, detained, and transferred to the adult justice system than white juveniles.169Eleanor Hinton Hoytt, Vincent Schiraldi, Brenda V. Smith & Jason Ziedenberg, Annie E. Casey Found., Pathways to Juvenile Detention Reform: Reducing Racial Disparities in Juvenile Detention 10 (2001), []. Specifically, four out of five juveniles detained between 1983 and 1997 were juveniles of color.170Id.

C.     The Modern Approach to Juvenile Justice

Today, a new era of juvenile justice has emerged—one that reflects the pre-1980s approach of treating youth as a mitigating factor. Recent surveys indicate that the public now favors a rehabilitative, rather than carceral, approach to juvenile crime.171GBA Strategies, Poll Results on Youth Justice Reform 1 (2017) [‌/H4HR-Y5HX]. Recognition of the mistakes made during the superpredator era has motivated many legislatures to pass statutes undoing some of these ill-conceived reforms.172Scott, supra note 147, at 548. More and more, states are changing their juvenile justice systems to reprioritize rehabilitation and keep young people out of the adult justice system.173Id. at 548–50. Some states are eliminating automatic transfers of juveniles from the juvenile justice system to the adult justice system, as well as increasing restrictions on discretionary transfers.174Jeree Thomas, Campaign for Youth Just., Raising the Bar: State Trends in Keeping Youth Out of Adult Courts (2015–2017), at 25–34 (2017), [] (identifying ten states that have restricted transfer mechanisms). Others are raising the age of their juvenile justice jurisdiction cutoffs in light of new studies suggesting that youth development continues into one’s twenties.175Tim Requarth, Neuroscience Is Changing the Debate over What Role Age Should Play in the Courts, Newsweek (Apr. 18, 2016, 10:01 AM),‌/young-brains-neuroscience-juvenile-inmates-criminal-justice-449000.html []. For example, Vermont passed a statute in 2017 making any offender under twenty-two charged with a nonviolent crime eligible for juvenile-offender status.176See Act of June 13, 2017, No. 72, §§ 5, 2017 Vt. Acts & Resolves 521, 523–28 (codified as amended at Vt. Stat. Ann. tit. 33, §§ 5280–5288 (Supp. 2020)). Illinois, Connecticut, and Massachusetts have proposed similar bills.177John Kelly, In Another Big Year for “Raise the Age” Laws, One State Now Considers All Teens as Juveniles, Imprint (June 25, 2018, 8:59 AM), [].

The Supreme Court has also embraced a more nuanced understanding of juvenile culpability that treats youth as a mitigating factor, holding that “children are constitutionally different from adults for purposes of sentencing.”178Miller v. Alabama, 567 U.S. 460, 471 (2012). Starting in the mid-2000s, the Supreme Court issued three decisions addressing juveniles’ Eighth Amendment protections. In the first of these cases, Roper v. Simmons, the Court held that the Eighth Amendment prohibited the imposition of the death penalty on a juvenile offender.179543 U.S. 551, 555–56, 578 (2005). The Court’s opinion, written by Justice Anthony Kennedy, rested in large part on the understanding that juveniles are fundamentally different from adults.180Roper, 543 U.S. at 569–70. Relying heavily on modern social science and neuroscience, Justice Kennedy identified three crucial distinctions between juveniles and adults: juveniles’ “lack of maturity” and “underdeveloped sense of responsibility,”181Id. (quoting Johnson v. Texas, 509 U.S. 350, 367 (1993)). their unique “susceptib[ility] to negative influences and outside pressures,” and their undeveloped characters that were more likely to change over time.182Id. In light of these differences, Justice Kennedy concluded that juveniles had diminished culpability when compared to adults, thus warranting additional protections.183Id. at 570–71.

Five years later, the Court reaffirmed this understanding of juvenile culpability. In Graham v. Florida, the Court held that the Eighth Amendment prohibited states from sentencing juveniles who commit nonhomicide offenses to life without parole.184560 U.S. 48, 82 (2010). Justice Kennedy, again writing for the Court, emphasized that while the unique characteristics of a juvenile do “not absolve[] [him] of responsibility for his actions, . . . his transgression ‘is not as morally reprehensible as that of an adult.’”185Graham, 560 U.S. at 68 (quoting Thompson v. Oklahoma, 487 U.S. 815, 835 (1988)). This decreased culpability rendered the severe penalty of life without parole disproportionate.186Id. at 74.

Most recently, in 2012, the Court addressed juvenile Eighth Amendment protections in Miller v. Alabama.187567 U.S. 460 (2012). Here, the Court addressed whether mandatory life-without-parole sentences for juvenile offenders who commit capital offenses violate the Eighth Amendment.188Miller, 567 U.S. at 465. In concluding that they do, Justice Elena Kagan, writing for the majority, again invoked the idea that biological differences between adults and juveniles justify treating the latter group less harshly.189Id. at 465, 471–72. She pointed to prior opinions which stated that “criminal procedure laws that fail to take defendants’ youthfulness into account at all would be flawed”190Id. at 473–74 (quoting Graham, 560 U.S. at 76). and that “the chronological age of a minor is itself a relevant mitigating factor of great weight.”191Id. at 476 (quoting Eddings v. Oklahoma, 455 U.S. 104, 116 (1982)).

This view of youth as a mitigating factor extends beyond the Court’s Eighth Amendment jurisprudence. In J.D.B. v. North Carolina, the Court held that an individual’s age is relevant in Fifth Amendment Miranda cases,192Miranda v. Arizona, 384 U.S. 436 (1966) (holding that the Fifth Amendment prohibits prosecutors from using a suspect’s statements against him at trial unless the suspect was informed of the right to remain silent and the right to have an attorney). relying on many of the same differences between juveniles and adults that the Court noted in Roper, Graham, and Miller.193J.D.B. v. North Carolina, 564 U.S. 261, 264, 271–277 (2011) (holding that age is relevant when determining whether a suspect is in police custody, thus triggering the suspect’s Miranda rights). The J.D.B. Court pointed to numerous other areas of the law—property, contract, family—that demonstrated a similar understanding.194Id. at 273 (“Like this Court’s own generalizations, the legal disqualifications placed on children as a class—e.g., limitations on their ability to alienate property, enter a binding contract enforceable against them, and marry without parental consent—exhibit the settled understanding that the differentiating characteristics of youth are universal.”) In interpreting their own state constitutions, some state supreme courts have issued similar opinions.195See, e.g., State v. Lyle, 854 N.W.2d 378, 404 (Iowa 2014); State v. Houston-Sconiers, 391 P.3d 409, 420 (Wash. 2017) (“In accordance with Miller, we hold that sentencing courts must have complete discretion to consider mitigating circumstances associated with the youth of any juvenile defendant, even in the adult criminal justice system . . . .”). In State v. Lyle, for example, the Iowa Supreme Court held that the Iowa Constitution prohibits the application of mandatory minimum sentences to all juvenile offenders.196Lyle, 854 N.W.2d at 404. Mandatory sentencing, the court held, is “simply too punitive” insofar as it precludes courts from taking into consideration the culpability-mitigating characteristics of youth.197Id. at 400–03.

A clear and consistent picture emerges from these cases. Courts recognize that juveniles’ youth make them less mature, less responsible, and more susceptible to outside influence than adults. As juveniles grow older, their characters will change and therewith their inclination to engage in criminal activity. The Supreme Court held that these unique qualities are reasons to treat juveniles more leniently—either by categorically protecting them from severe punishment, as in Roper and Graham, or by requiring judges to consider the mitigating effects of youth before imposing certain punishment, as in Miller and Lyle. Although the specific holdings differ, each case concludes that youth is a mitigating factor for punishment and that this factor has significant weight when determining a juvenile’s sentence.

III.     Problems with RAI’s Treatment of Youth

Courts’ treatment of youth as a mitigating factor in determining the culpability of juvenile offenders derives from traditional principles concerning the punishment of youth. This Part argues that RAIs that recommend harsher sentences based on an offender’s youth are inconsistent with these principles. The argument here is not so much that the use of RAIs violates a specific state or federal law but that given the sheer quantity and quality of criminal justice practices that view youth differently, RAIs’ treatment of youth as an aggravating factor is an anomalous and wrongheaded discontinuity.198Thank you to Professor Rebecca Eisenberg for this wonderful description. The only other time in American history when the criminal justice system treated young people this way was during the superpredator era. Since we have moved away from these destructive views, states should reject RAIs that threaten to replicate these mistakes. Section III.A argues that RAIs are incompatible with our historical and present-day views of juvenile punishment, while Section III.B illustrates the ways in which RAIs mirror the policies and practices of the superpredator era. Section III.C concludes by explaining this Comment’s broader implications for the punishment of young offenders.

A.     Inconsistency with Principles of Juvenile Punishment

RAIs are incompatible with fundamental principles of juvenile punishment. The U.S. criminal justice system has long treated youth as a mitigating factor.199The superpredator era was the only exception to this general rule. See supra Section II.B. Two primary justifications have been given for this leniency: first, juveniles have been considered less culpable than adults; second, it has been believed that the proper focus of juvenile punishment was rehabilitation.

At almost all times in American history—and even before200The infancy defense predates the formation of the United States, having emerged in English common law in the 1300s. See supra note 109.—it was understood that juveniles were less culpable than adults. At common law, this view was enshrined in the infancy defense, which was justified on the basis that young people lacked the moral understanding necessary for criminal culpability.201Lerner, supra note 110, at 1586–87. The Progressives took this idea to its limit, arguing that all juveniles, regardless of age, were fundamentally innocent.202See Ryerson, supra note 120, at 36. The Supreme Court’s decisions in Roper, Graham, and Miller, which concluded that the differences between youth and adults justify heightened Eighth Amendment protections for juveniles, further support the general view that juveniles are less culpable than adults.203See, e.g., Roper v. Simmons, 543 U.S. 551, 571 (2005) (“Whether viewed as an attempt to express the community’s moral outrage or as an attempt to right the balance for the wrong to the victim, the case for retribution is not as strong with a minor as with an adult.”).

Moreover, punishment of juveniles has historically emphasized rehabilitation over protection of the public. As far back as the eighteenth century, scholars used juveniles’ unique capacity for reformation as an argument against the juvenile death penalty.204Hale, supra note 115, at 25. The rehabilitation ideal was then reified into the juvenile justice system, the original purpose of which was to transform juvenile offenders into law-abiding adults.205Carter, supra note 117, at 719–20. Today, after almost two decades of public hysteria over the danger of superpredators, the juvenile justice system is refocusing on its original goal of rehabilitation, as scholars and lawmakers advocate for and adopt new reforms to serve this purpose.206See supra Section II.C. The Supreme Court has similarly recognized that juveniles are “most in need of and receptive to rehabilitation” and that “the absence of rehabilitative opportunities or treatment makes the disproportionality of the [life-without-parole] sentence all the more evident.”207Graham v. Florida, 560 U.S. 48, 74 (2010).

RAIs’ treatment of youth is completely at odds with these long-established principles. Most obviously, RAIs substantially increase the risk of juvenile offenders being incarcerated and of their incarceration lasting for longer periods of time.208Stevenson & Doleac, supra note 5, at 3–4. This directly contravenes the Supreme Court’s unwavering position that a juvenile’s age should be treated as “a relevant mitigating factor of great weight.”209See, e.g., Miller v. Alabama, 567 U.S. 460, 476 (2012) (quoting Eddings v. Oklahoma, 455 U.S. 104, 116 (1982)). Youth’s strong aggravating effect in RAI calculations is due to the fact that young people are more likely to engage in criminal activity because they are more immature, irresponsible, and susceptible to negative peer influence than adults.210See supra Section I.C.1. But these are the exact same traits upon which the Court has “[t]ime and again” relied when affording juveniles extra protections under the Constitution.211J.D.B. v. North Carolina, 564 U.S. 261, 272 (2011). The Roper Court explicitly dismissed youth’s criminogenic effects as irrelevant to sentencing, acknowledging that “adolescents are overrepresented statistically in virtually every category of reckless behavior”212Roper v. Simmons, 543 U.S. 551, 569 (2005) (quoting Jeffrey Arnett, Reckless Behavior in Adolescence: A Developmental Perspective, 12 Developmental Rev. 339, 339 (1992)). while still suggesting that using the juvenile defendant’s age against him would be “overreaching.”213Id. at 573. The prosecutor in Roper argued that the defendant’s age was an aggravating factor. In response to the defense attorney’s argument that the defendant’s age should be considered a mitigating factor, the prosecutor responded: “Seventeen years old. Isn’t that scary? Doesn’t that scare you? Mitigating? Quite the contrary I submit. Quite the contrary.” Id. at 558.

The difference between the Court’s and RAIs’ treatment of youth stems from a disagreement over the proper goal of juvenile punishment. The Court evaluates punishment through rehabilitative and retributivist lenses, emphasizing juveniles’ greater potential for reform and their decreased culpability. On the other hand, RAIs focus exclusively on incapacitation, which seeks to incarcerate dangerous individuals to prevent them from doing harm to society.214Jessica M. Eaglin, Constructing Recidivism Risk, 67 Emory L.J. 59, 99–100 (2017) (“Of the myriad rationales for punishment that might underpin sentencing in general, the incapacitation rationale most squarely justifies the use of predictive evidence.”). RAIs promote incapacitation by providing judges with assessments of offenders’ recidivism risk, allowing judges to tailor offenders’ sentences to keep them in prison for as long as they are still dangerous.215Id. Because youth increases an offender’s recidivism risk,216See supra Section I.C. incapacitation justifies its use as an aggravating factor for RAIs.217Even under the incapacitation theory, RAIs do a poor job of helping judges determine a young offender’s appropriate punishment. Most RAIs only predict an offender’s risk one to three years into the future, failing to take into account the possibility that an offender’s risk drops significantly after that point. Eaglin, supra note 214, at 100. This problem is especially relevant for age, which, due to the age-crime curve, has time-sensitive predictive power. Because RAIs do not account for the decreased predictive power of youth over time, they encourage sentences that incarcerate youth long after they have aged out of their main crime-committing years. Another problem that scholars have identified is that RAIs don’t consider the impact that the sentence itself has on an offender’s recidivism risk. Starr, supra note 15, at 855–61. This concern is especially grave for young offenders, who naturally age out of crime and for whom incarceration is more likely to have counterproductive effects. See supra Section I.C.1.
It may be tempting to suggest that states develop more nuanced tools that can account for these problems; however, such arguments miss the mark. It ultimately does not matter whether RAIs can accurately or effectively advance an incapacitation goal of punishment because our criminal justice system has already determined that incapacitation is not the point of youth punishment. Even if RAIs successfully promote incapacitation, this Comment contends that their use in the sentencing of young offenders is still inappropriate given these principles of youth punishment.
In focusing only on incapacitation, however, RAIs are incongruent with our juvenile sentencing principles, which have long considered culpability and the potential for rehabilitation as crucial components of the sentencing analysis for juvenile offenders.

To be clear, this Comment does not argue that incapacitation is an improper consideration in the sentencing of young offenders. Young offenders who pose a significant risk to the public can—and perhaps should—be incapacitated for the safety of others. Even so, incapacitation is not the goal of youth punishment. As past and present doctrine demonstrates, the criminal justice system deems culpability and rehabilitation more important when it comes to the punishment of young offenders. For this reason, the criminal justice system has time and again provided juveniles more lenient sentences and offered young offenders the opportunity to change, despite the recognition that young people are more likely to recidivate.218See supra Sections I.C, II.A, II.C. RAIs violate this commitment by focusing exclusively on incapacitation rather than rehabilitation or culpability.

Even worse, RAIs encourage judges to overvalue incapacitation at the expense of other penological goals. As previously discussed, judges believe that RAIs have statistical credibility, which encourages judges to rely heavily on risk scores.219See supra notes 70–73 and accompanying text. Compared to other sentencing factors which can be indeterminate, risk scores provide normative labels or explicit recommendations that foster overreliance.220See Starr, supra note 15, at 866. One study conducted by criminal law professor Sonja Starr supports this idea.221Id. at 867–69. In the study, subjects were more likely to give a harsher sentence to a defendant with a higher risk of recidivism for the same crime, even though the facts suggested that the riskier defendant was less morally culpable.222Id. Thus, RAIs encourage judges to weigh incapacitation over the traditional goals of juvenile punishment.

RAIs also undermine the legal premise that youth is a mitigating factor for culpability,223See supra Section II.C. encouraging judges instead to treat youth as an aggravating factor. Although RAIs only provide information about recidivism risk and not an offender’s character, judges may mistakenly believe that a high risk score is also evidence of an offender’s increased culpability.224See Stevenson & Slobogin, supra note 20, at 700–02. Judges often confuse high risk for bad character in judicial decisions;225See, e.g., Deck v. Missouri, 544 U.S. 622, 633 (2005) (observing that evidence of the defendant’s “danger to the community . . . almost inevitably affects adversely the jury’s perception of [his] character”). such conflation is especially dangerous when sentencing juvenile offenders, whose most powerful culpability-mitigating factor also acts as a strong indication of future risk.226See Stevenson & Slobogin, supra note 20, at 700–02. Judges may interpret a juvenile offender’s high risk score as evidence of a depraved character, when in reality, the risk score is heavily determined by age.227Remember that for the COMPAS, youth alone accounts for 57 percent of the offender’s risk score. See supra note 80 and accompanying text. As a result, the judge will have indirectly used youth as a culpability-aggravating factor, the opposite of its traditional role in the criminal justice system. Criminal law professors Megan Stevenson and Christopher Slobogin describes this inconsistency as the “wors[t] sort of double-edged swordism” and notes that this problem is especially severe with respect to opaque RAIs like the COMPAS.228Id. at 702. Because these tools fail to disclose the factors contributing to an offender’s risk score, a judge using these tools may treat youth as both a culpability-mitigating and culpability-aggravating factor without even intending to do so.229Id. (arguing that such a “result is both illogical and unacceptable”).

RAIs also undermine the criminal justice system’s goal of rehabilitating juvenile offenders. As previously discussed, RAIs increase juvenile offenders’ likelihood of being incarcerated and the risk that they will be incarcerated for longer periods of time. This kind of punishment is counterproductive for rehabilitation purposes.230See supra Section I.B. The longer a juvenile remains incarcerated, the greater the disruption to the juvenile’s natural tendency to age out of crime. While incarcerated, juveniles have fewer developmental opportunities and fewer chances to develop prosocial connections with law-abiding peers and adults.231See supra Section I.C.2. Thus, RAIs that treat youth as an aggravating factor frustrate juvenile-rehabilitation sentencing goals. For these same reasons, RAIs may have the self-defeating effect of increasing recidivism even if incapacitation is deemed a legitimate basis for punishment.

In addition to undermining the criminal justice system’s fundamental goal of sentencing juvenile offenders with an eye toward culpability and rehabilitation, RAIs also violate another juvenile sentencing principle: the individualized-sentencing mandate. During the formation of the juvenile justice system, the Progressives argued that individualized sentencing was necessary for effectively rehabilitating juvenile delinquents.232See Bazelon, supra note 119, at 172. This call for individualized sentencing was later echoed by the Supreme Court in Miller, in which the Court held that mandatory penalty schemes violate the Eighth Amendment because they prevent judges from considering the youthfulness of the offender.233Miller v. Alabama, 567 U.S. 460, 480 (2012) (“Although we do not foreclose a sentencer’s ability to make that judgment in homicide cases, we require it to take into account how children are different, and how those differences counsel against irrevocably sentencing them to a lifetime in prison.”). Although Miller focused on mandatory life-without-parole sentences, its rationale can be used to justify the broader principle that the sentencing of juvenile offenders should take into account “an offender’s age and the wealth of characteristics and circumstances attendant to it.”234Id. at 476. The Court deemed it unacceptable that all juveniles receive the same sentence and “still worse” were juveniles to receive the same sentence as most adults who committed similar offenses.235Id. at 477.

Like mandatory-sentencing schemes, RAIs undercut judges’ ability to engage in individualized analysis. This is true in two ways. First, RAIs inherently rely on generalizations about groups of people (e.g., young people, men, people who live in certain areas) to make inferences about specific members of those groups.236See Collins, supra note 16, at 65. This problem is especially prominent with checklist-style RAIs, which tend to make cruder assessments about even larger groups of people. For example, Virginia’s Nonviolent Offender Risk Assessment (NVRA) gives all offenders thirteen points for committing a crime under the age of thirty.237Va. Crim. Sent’g Comm’n, supra note 32. Thus, judges who rely on RAIs make their sentencing decisions based on individuals who share similarities with the offender rather than analyzing the specific offender himself, thereby violating the individualized-sentencing mandate. Second, RAIs only consider certain characteristics of an offender—others not preprogrammed into the algorithm are totally ignored for risk assessment purposes. As a result, judges who rely on RAIs may not fully consider “the wealth of characteristics and circumstances attendant” to the juvenile offender’s age.238Miller, 567 U.S. at 476. Of course, RAIs differ from mandatory-sentencing schemes in a very crucial way: RAI use is discretionary. See Eaglin, supra note 214, at 61–62. Even still, the fact that RAIs are so at odds with the mandate in Miller is problematic due to the influence that RAIs have on sentencing. See supra Section I.B. The rationale behind Miller suggests that a juvenile’s sentence should be made with a complete analysis of the offender’s “wealth of characteristics and circumstances,” but a sentence that is based largely on group-based classifications ignores this teaching. Some RAIs even recommend certain sentences based on an offender’s risk score. In these jurisdictions, judges are invited to impose recommended sentences on juvenile offenders without first considering the unique circumstances of their particular cases. See supra note 37 and accompanying text. In all these ways, RAIs are fundamentally inconsistent with our legal tradition of leniency for juvenile offenders.

B.     The Superpredator Theory Reincarnated

The push to adopt RAIs is a part of a broader movement that “aims to be smart, rather than tough, on crime.”239Collins, supra note 16, at 59. The “smart on crime” movement attempts to use RAIs to address mass incarceration and to reprioritize rehabilitation by identifying low-risk offenders who are capable of serving their sentences in community-based programs.240See id. Despite this marketing, however, RAIs are just another iteration of the overly punitive superpredator-era reforms. Like the superpredator theory, RAIs result in the overincarceration of juvenile offenders in two important ways: RAIs both perpetuate the misconception that juvenile offenders are morally depraved and encourage judges to prioritize incapacitation over rehabilitation when sentencing juvenile offenders.

First, RAIs perpetuate the superpredator-era misconception that juvenile offenders are morally depraved. During the superpredator era, juvenile offenders were not viewed as normal children whose lack of maturity, greater susceptibility to peer pressure, and underdeveloped sense of self resulted in bad decisionmaking.241DiIulio, supra note 131. Rather, juvenile offenders were considered “kids who [had] absolutely no respect for human life and no sense of the future.”242Id. RAIs similarly encourage judges to view juveniles as morally depraved, as judges are likely to misunderstand juvenile offenders’ high risk scores as evidence of increased culpability rather than as products of their age.243See Stevenson & Slobogin, supra note 20.

Second, both the superpredator theory and RAIs encourage judges to emphasize incapacitation over rehabilitation. During the superpredator era, lawmakers believed that juvenile offenders were being “coddled by a justice system that clings to a discredited belief in rehabilitation.”244Elliot Currie, Crime and Punishment in America 2 (rev. & updated ed. 2013). It was thought these juveniles were so incorrigible that only by incarcerating them could the public be protected.245See Grim Reality Check on Youth Crime, supra note 133 (“Increasingly, we wonder not whether they can be saved, but whether we can save ourselves from them.”). No rehabilitative intervention could possibly fix these individuals: if put in the juvenile justice system, they would simply do their time until they turned eighteen, after which they would be released from their cell back onto the streets as full-grown adults even more predisposed to violence and destruction.246See Krista Larson & Hernan Carvente, Juvenile Justice Systems Still Grappling with Legacy of the “Superpredator” Myth, Vera Inst. of Just.: Think Just. Blog (Jan. 24, 2017), [].

Because RAIs reinforce these misguided beliefs, their use increases young offenders’ likelihood of being incarcerated and decreases their likelihood of being rehabilitated—a result completely at odds with the goals of the “smart on crime” movement. This counterintuitive effect is not as surprising as it may initially appear. The first RAI for sentencing was adopted in Virginia in 1994 near the peak of the superpredator myth’s prominence.247Starr, supra note 15, at 809. Although the NVRA was designed to facilitate diversion from prison, the tool was not meant to uproot the previously existing public safety-focused approach.248See Garrett et al., supra note 34, at 3. Its adoption simply meant that RAIs would be labeling offenders as dangerous or not dangerous rather than lawmakers or judges—mechanizing the system, not overthrowing it. Put simply, RAIs cannot promote a “smart on crime” approach to juvenile sentencing because, at their core, they are merely a reincarnation of one of our country’s most misguided “tough on crime” policies.

In addition to undermining the “smart on crime” movement, use of RAIs also subverts recent efforts to promote racial equality in the criminal justice system. Like the superpredator theory, use of RAIs results in racially discriminatory sentencing. A study that analyzed the risk scores of seven thousand individuals in Broward County, Florida, found significant racial disparities in the COMPAS’s false-negative and false-positive rates.249Jeff Larson, Surya Mattu, Lauren Kirchner & Julia Angwin, How We Analyzed the COMPAS Recidivism Algorithm, ProPublica (May 23, 2016), [] (“The algorithm is more likely to misclassify a black defendant as higher risk than a white defendant. . . . The test tended to make the opposite mistake with whites, meaning that it was more likely to wrongly predict that white people would not commit additional crimes if released compared to black defendants.”). Though white defendants were mislabeled as low risk more often than Black defendants, the RAI was twice as likely to mislabel a Black defendant as high risk than a white defendant.250Id. Because high risk scores are more likely to result in longer and harsher sentences, the fact that Black defendants are more likely to be mislabeled as high risk is likely to result in oversentencing of Black individuals.251See supra Section I.B. Just as the superpredator theory resulted in more Black juveniles being arrested, detained, and transferred to the adult justice system, so too will the increased use of RAIs that treat youth as an aggravating factor.

Although RAIs do not directly take race into account, Black offenders receive higher risk scores because the factors used to calculate these risk scores often act as proxies for race.252See Starr, supra note 15, at 838. For example, many RAIs factor the crime rate of an offender’s neighborhood into their risk score; crime rates are often higher in urban neighborhoods that are densely populated with residents of color.253Id. Similarly, RAIs that use education as a risk factor will likely pick up on the fact that Black and Hispanic individuals are less likely to graduate from high school and college than are white people.254See Camille L. Ryan & Kurt Bauman, U.S. Census Bureau, Educational Attainment in the United States: 2015, at 2 (2016),‌/dam/Census/library/publications/2016/demo/p20-578.pdf []. Furthermore, almost all RAIs rely on criminal history, another factor that is strongly correlated with race.255See Hamilton, supra note 19, at 77, 130–32. In 2015, for instance, Black juveniles were over four times as likely as white juveniles to be detained or committed to youth facilities.256The Sent’g Project, Black Disparities in Youth Incarceration (2021), []. Thus, despite RAI’s apparent race neutrality, the predictive factors they use are more likely to result in higher risk scores for Black offenders.

This result is not only discriminatory but also inevitable. Despite proponents’ hopes that RAIs offer a new path forward, these tools mirror the past. Risk predictions “identify patterns in past data and offer them as projections about future events.”257Mayson, supra note 2, at 2251. Unfortunately, due in large part to policies enacted during the superpredator era, our past criminal data is racially skewed.258See supra Section II.B. Black people—especially young Black men—have been overpoliced and overincarcerated for decades, and RAIs will only reproduce these unequal results.259Id. As criminal law professor Sandra Mayson succinctly puts it, “bias in, bias out.”260Mayson, supra note 2, at 2224.

The premise of prediction “is that we can learn from the past because, absent intervention, the future will repeat it.”261Id. But the death of George Floyd in the summer of 2020 has galvanized a national reckoning with the criminal justice system’s unequal and unjust treatment of Black people262Eliott C. McLaughlin, How George Floyd’s Death Ignited a Racial Reckoning That Shows No Signs of Slowing Down, CNN (Aug. 9, 2020, 11:31 AM),‌/08/09/us/george-floyd-protests-different-why/index.html []. and a determination not to let past harms go unremedied.263Orion Rummler, The Major Police Reforms Enacted Since George Floyd’s Death, Axios (Oct. 1, 2020), []. States that adopt and use RAIs only project the past into the future.

Lawmakers must not succumb to the false hope held out by RAIs. Despite attempts to undo the regimes of the superpredator era, many (if not most) of these draconian laws remain in effect, and many of those sentenced under these laws remain in prison.264See Scott, supra note 147, at 548. Expansive legislative changes, once on the books, can be very difficult to undo.265Id. at 548 n.81. For this reason, lawmakers should avoid compounding the destructive mistakes of the superpredator era by adopting RAIs at sentencing. America has already incarcerated thousands of individuals based on unfounded fears of juvenile violence,266Moriearty, supra note 137, at 853. and the system created to imprison these young people has proven to be especially challenging to dismantle. Until more is done to reconcile RAIs’ treatment of youth with the traditional and modern goals of the criminal justice system, states should not adopt RAIs that treat youth as an aggravating factor.

C.     Beyond the Juvenile Justice System

Though the inconsistencies between RAIs and long-standing criminal justice principles are most striking when applied to juveniles,267See supra Section I.C (explaining that this Comment uses the term juvenile to refer to individuals under the age of eighteen). the use of RAIs is problematic for the sentencing of all young people. This is true even for young people in the adult justice systems or who are above the legal age of majority. Many (though not all)268Consider, for example, the infancy defense for misdemeanors, which applied to offenders as old as twenty-one in some cases. See Lerner, supra note 110, at 1588–90. of the historical and constitutional sources described in this Comment refer only to juveniles. Nevertheless, these perspectives are just as informative for an understanding of the criminal justice system’s treatment of young people generally.

Age-based cutoffs tend to be arbitrary and inconsistent, changing across many different time periods and localities.269This section only focuses on the age-based cutoffs within the criminal justice system, but there are many other examples of conflicting and inconsistent age limits outside the criminal realm. For example, individuals can travel unaccompanied at twelve, serve in the military and vote at eighteen, drink alcohol at twenty-one, and rent a car at twenty-five. David Pimentel, The Widening Maturity Gap: Trying and Punishing Juveniles as Adults in an Era of Extended Adolescence, 46 Tex. Tech L. Rev. 71, 84–85 (2013). Take the juvenile justice system as an example. As of 2021, the highest age of juvenile-court jurisdiction in forty-six states and the District of Columbia is seventeen years old, but three states (Georgia, Texas, and Wisconsin) have a maximum age of sixteen years old and another (Vermont) has a maximum age of eighteen.270Anne Teigen, Juvenile Age of Jurisdiction and Transfer to Adult Court Laws, Nat’l Conf. of State Legislatures (Apr. 8, 2021), []. Missouri and Michigan passed laws raising their maximum ages to seventeen; both laws went into effect in 2021. Id. Vermont’s upper age of juvenile-court jurisdiction will increase to twenty in 2022. See Vt. Stat. Ann. tit. 33, § 5103(c)(2)(A) (Supp. 2020) (effective July 1, 2022). This is quite different from the breakdown in 2009, where only thirty-eight states had a maximum age of seventeen, ten states had a maximum age of sixteen, and three states had a maximum age of fifteen.271Jurisdictional Boundaries, Juv. Just. Geography Pol’y Prac. & Stat., http://www‌ []. The modern approach also contrasts starkly with the common law approach, where the infancy defense provided the only special treatment that young offenders received prior to the creation of the juvenile justice system.272See Buss, supra note 109, at 19; see also supra Section II.A.

Similarly, the Supreme Court has created additional constitutional protections for juveniles under the age of eighteen, but this was not always the case. Before the Court’s decision in Roper, prohibiting the death penalty for individuals under eighteen,273Roper v. Simmons, 543 U.S. 551 (2005). a plurality of the Court decided in Thompson v. Oklahoma that the Eighth Amendment prohibited the death penalty of any offender under the age of sixteen.274487 U.S. 815, 838 (1988). In Roper, Justice Kennedy extended the decision in Thompson to include sixteen- and seventeen-year-olds because “[t]he logic of Thompson extends to those who are under 18.”275Roper, 543 U.S. at 574.

Judges sentencing young offenders should not engage in this kind of arbitrary line drawing. Even if the Supreme Court and state legislatures must make such bright-line rules for workability purposes, sentencing judges are not subject to the same limitations. Sentencing is an extremely fact-intensive process; judges are given presentencing reports and hold sentencing hearings in an effort to discern all relevant information about the offender.276See supra notes 70–73 and accompanying text. This process, especially for young offenders, is supposed to consider the individual characteristics of the offender instead of the categories to which they are arbitrarily assigned.277See supra notes 232–238 (discussing RAIs’ incompatibility with the individualized-sentencing mandate). It is not the offender’s label as a “juvenile” or “adult” that justifies offering them additional protections in the criminal justice system. Rather, young offenders should be given additional protections because of certain unique qualities that have consistently been deemed important in sentencing.278See supra Sections II.A, II.C. The labels of “juvenile” or “adult” establish a binary system by which individuals are classified as having or not having these qualities. But this system is both over- and underinclusive and is thus unsuitable for the individualized, fact-specific inquiry required at sentencing.

Rather than looking to these age-based cutoffs, this Comment instead looks at the underlying justification for these cutoffs, namely, the distinct qualities of juveniles that warrant their special treatment in the criminal justice system. This Comment contends that the logic of Thompson—and of Roper, Graham, Miller, the infancy defense, and the juvenile justice system, for that matter—apply to all young offenders, not just juveniles.279At least one federal district court has recognized that the logic of the Supreme Court’s juvenile Eighth Amendment cases applies to young adults as well. Cruz v. United States, No. 11-CV-787, 2018 WL 1541898, at *25 (D. Conn. Mar. 29, 2018) (“‘[T]he Eighth Amendment forbids a sentencing scheme that mandates life in prison without possibility of parole’ for offenders who were 18 years old at the time of their crimes.” (quoting Miller v. Alabama, 567 U.S. 460, 479 (2012))), vacated, 826 F. App’x 49 (2d Cir. 2020). The Second Circuit vacated the district court’s opinion in a summary order, holding that Miller applies only to those who were younger than eighteen at the time of their crimes. Cruz v. United States, 826 F. App’x 49, 51–52 (2d Cir. 2020), cert. denied, 141 S. Ct. 2692 (2021). As a summary order, the decision has no precedential effect. Id. at 49. The Thompson plurality stressed the cognitive differences between juveniles and adults, arguing that because of these differences, juveniles’ “irresponsible conduct is not as morally reprehensible as that of an adult.”280Thompson v. Oklahoma, 487 U.S. 815, 838 (1988). But the unique qualities associated with juveniles are present in young people generally. Recent scientific studies demonstrate that the brain continues to develop into the mid-twenties and that young adults possess many of the same traits as juveniles.281See supra Section I.C.1. Justice Kennedy acknowledged as much in Roper, stating that “[t]he qualities that distinguish juveniles from adults do not disappear when an individual turns 18.”282Roper v. Simmons, 543 U.S. 551, 574 (2005).

The push to give young adults additional legal protections is gaining traction. Many scholars have argued in favor of an older cutoff age between juveniles and adults, citing neuroscientific research, mass incarceration, and recent reforms as support.283See, e.g., Josh Gupta-Kagan, The Intersection Between Young Adult Sentencing and Mass Incarceration, 2018 Wis. L. Rev. 669; Mullin, supra note 106. And several states have in turn implemented new programs designed to treat young adults differently than older adults. Examples include youthful-offender statutes that mitigate the sentences of offenders under the age of twenty-five, young adult courts modeled on juvenile courts, and expanded jurisdiction of the juvenile justice system.284See Gupta-Kagan, supra note 283, at 683–84; Requarth, supra note 175.

These reforms bolster the argument that offenders older than eighteen ought to receive some of the protections offered to juvenile offenders, but it is worth stressing that this Comment’s arguments do not focus on changing the legal cutoff. The argument is instead that because young adults possess many of the same characteristics that justify lenient sentencing for juveniles, use of RAIs in the sentencing of young offenders is problematic regardless of the legal age of majority. As explained, the criminal justice system has long recognized that it is these unique qualities, not mere juvenile status, that justifies treating youth as a “mitigating factor of great weight.”285Miller v. Alabama, 567 U.S. 460, 476 (2012) (citing Eddings v. Oklahoma, 455 U.S. 104, 116 (1982)). Insofar as RAIs encourage judges to treat youth as an aggravating factor, they distort the sentencing process and create the unacceptable risk of disproportionate sentences for all young offenders.


The use of RAIs that treat youth as an aggravating factor in sentencing not only contradicts fundamental assumptions of the criminal justice system but also undermines a new era of juvenile justice jurisprudence that attempts to synthesize modern social science with principles of proportionality and justice. This inconsistency must be addressed by states that currently use such RAIs in sentencing as well as by those contemplating adoption of similar RAIs. Although this Comment does not propose a specific reform, its analysis strongly suggests that lawmakers, judges, and scholars should—at a minimum—seriously consider eliminating the use of RAIs against young offenders or developing new tools that avoid consideration of youth. These arguments also provide support for solutions that have been proposed by scholars to address other concerns related to RAIs, including increased transparency surrounding how RAIs generate risk scores and increased education of judges about RAIs.286See, e.g., Stevenson & Slobogin, supra note 20, at 703–05. These proposals may allow judges to use their discretion to neutralize some of the problematic features of RAIs and to reprioritize rehabilitation, diminished culpability, and individualized consideration when sentencing young offenders. More research is needed to determine which reforms are best suited to addressing the problems identified by this Comment. But this much is clear: as currently designed and implemented, RAIs mirror the destructive and discriminatory mistakes of the past and subvert the goal of building a smarter and fairer criminal justice system for the future.

*      J.D., May 2021, University of Michigan Law School. Thank you to Professors Sonja Starr, Gabriel Mendlow, Rebecca Eisenberg, and Nina Mendelson for their feedback and guidance. I am grateful for the helpful comments of participants in the Michigan Law Student Scholarship Workshop and for the wonderful editing by the Volume 120 Notes Office. Special shoutout to my fellow Volume 119 Notes Editors and to Carrie Darden of Volume 118. Lastly, I want to give a heartfelt thank you to Marie, whose never-ending patience and encouragement helped make this dream a reality.