TEST PAGE IGNORE
The very idea of using actuarial risk assessment in sentencing decisions poses several further problems at the level of policy or even philosophy. By assuming that the future will be like the past, that the individual is like the group, and that the individual as an actor is like the outward signs by which we know the individual, the methodology underlying prediction makes the offender into an object of scientific study. This approach is not altogether wrong, and it may be useful. However, it also risks reducing the offender to an automaton bound by a chain of cause and effect that leaves no room for free will. Because these problems reach fundamental questions about the nature of human agency, judges are not likely to solve them in a completely satisfying way. Instead, the solutions will need to balance the defendant’s individual rights against the demands of society.
ARAS assume that the future will be more or less like the past: there exists a continuity between defendants’ past behavior and what they will do in the future. As explained above, it is this principle of continuity that makes prediction possible. In this context, continuity operates on at least two levels. First, the system assumes that the relationships between risk factors and future recidivism that past offenders demonstrated will continue to hold true, such that if the defendant being evaluated displays the same risk factors, he or she will also likely recidivate. Second, systems often use criminal history as a risk factor, assuming that one who broke the law in the past will continue his or her misbehavior into the future. These assumptions are not entirely without merit, and they may enable reasonably accurate predictions, but they nonetheless threaten the defendant’s right to self-determination.
The second assumption brings into focus the danger of assuming that the past must repeat itself. A central goal of punishment, and of the criminal law in general, is the reduction of crime, i.e., intervening in people’s lives to break the links of cause and effect that would lead to future crime. This includes leading those who have strayed back to the path of lawful behavior, but it also includes offering the offender the opportunity to change his or her ways. By assuming that an offender with a criminal history is something of a career criminal and must be locked up, lest he or she commit further crimes, a criminal justice system that relies upon ARAS delays or denies the offender’s chance to return to a lawful life. Given the lack of opportunities within prisons for inmates to develop their skills and the stigma that follows incarceration and frequently curtails lawful employment, assuming that an offender with a history of crime will extend his or her misdoings into the future could become a self-fulfilling prophecy.
Even the assumption that the future will mirror the past at the group level risks creating the undesirable reality that it predicts. As explained above, by using data from the past, an ARAS drags the criminal justice system’s past failings into the future, even as reformers are desperate for law enforcement and the courts to break free of their history of racial discrimination. Much as the courts seek to intervene in offenders’ lives to redirect them to a better future, progressives within and without the legal world hope for a break from the past. In contrast, the fundamentally conservative assumptions underlying prediction imagine that past and future are continuous, and thereby entrench the failings of yesteryear.
Two related problems in the prediction methodology stem from the assumption that the individual is like the group of which he or she is a member. First, such an assumption may simply be wrong as an empirical matter because the individuals within the group vary so much that they defy any attempts to reduce their attributes to some simple measure. Sonja B. Starr illustrates this problem with the example of human height. Sampling the height of a few hundred people lets one calculate a reasonably accurate mean height, but it does little to improve one’s chance of predicting the height of the next person one will meet. Even within a single gender, height varies too much from one person to the next for one to form a precise and accurate prediction. Much like IQ and Body Mass Index, the group measure may be a useful tool for understanding the group at a population level, but its utility breaks down at the individual level. Similarly, though meticulous criminological studies may be able to gather data about criminal behavior at a group level, it may be impossible to apply any such insights to the individuals within the group with reasonable precision and accuracy. The predictions would likely be not only wrong, but also unfair.
Second, even if one assumes that group data can accurately predict individual recidivism risk in most cases, applying such a principle mechanically once again denies the offender the chance to change course. Implicit in the criminal law is the notion that defendants act with the freedom to choose their path in life; it is that choice that generates responsibility and permits punishment. So long as the law respects the offender’s free will and hopes that he or she will exceed whatever expectations group-based predictions may create, it would be wrong to impose upon him or her the computer’s prediction.
The group-to-individual problem is a particular form of the more general assumption that from manifest signs—including group membership—one can predict the individual’s future behavior. Insofar as we attribute to the defendant the ability to choose freely his or her actions, this assumption is suspect. The defendant’s freedom carves out an interior space into which one cannot see and from within which the defendant may defy any prediction that is made about his or her future choices.
Finally, the use of ARAS in sentencing decisions suffers from yet another philosophical problem—confusion about the proper goals of punishment. Recidivism risk assessment primarily addresses the goal of incapacitation while neglecting the goals of retribution, deterrence, rehabilitation. Because ARAS scores have the potential to anchor and tempt judges into giving them undue weight, courts including the scores in presentence investigation reports risk privileging one penal goal over the others on no rational grounds. Answering the question of whether this potential focus on incapacitation harms defendants awaits a coherent theory of punishment that defines its proper ends. This is not likely to be forthcoming in the near future.
The constellation of quasi-philosophical problems examined above center around a disagreement about the nature of the person as such. On the one hand, advocates of using ARAS approach the problem of sentencing from a scientific standpoint. They speak in the discourse of science and view the defendant through a scientific lens. In order to make the defendant into an object of study amenable to their methods, they impose upon him or her a certain conceptual apparatus, including the notion that things in nature—including people—obey uniform natural laws. Much as in other domains, they apply the techniques of the mathematical sciences to uncover these laws so that they may predict and control nature. On the other hand, followers of the common law and the various moral theories from which it grew see the defendant from another perspective. They approach the situation equipped with their own collection of concepts and speak with a very different vocabulary: freedom, choice, responsibility, moral desert. At times, these two groups seem to be speaking different languages altogether, or speaking the same language but not talking about the same thing.
Because reconciling these two perspectives is unlikely, courts ought to navigate a course between them by following the beacon of an overarching value such as fairness. The disagreement outlined above approaches fundamental conflicts like materialism-idealism that cannot be resolved. One cannot eliminate the tension by reducing one position to the other, e.g., one cannot explain away idealism in entirely materialist terms. Nor can one ignore either side of the debate without losing a valuable perspective. The trouble is that though these views are incompatible, neither is strictly wrong. Each view contributes something useful to our understanding of the problem of proper sentencing, and the judge who tries to avoid the conflict between these positions by ignoring one side of the debate will suffer from a willful ignorance. Instead, the judge ought to consider both perspectives, take from each what is useful, and apply those lessons in his or her pursuit of a sentence that satisfies the goals of punishment while remaining fair and just.
 Starr, supra note 8 at 838.
 Cf. id. at 855 (“instruments tell us, at best, who has the highest risk of recidivism. They do not tell us whose risk of recidivism will be reduced the most by incarceration”)
 Id. at 842–47.
 Id. at 844.
 See id. at 842 (“They do not provide anything even approaching a precise prediction of an individual’s recidivism risk. The models are designed to predict the average recidivism rate for all offenders who share with the defendant whichever characteristics are included as variables in the model. If the model is well specified and based on an appropriate and large enough sample, then it might perform this task well. But because individuals vary much more than groups do, even a relatively precisely estimated model will often not do well at predicting individual outcomes in particular cases.”).
 Cf. id. at 850 (“it is not harmless to base an individual’s incarceration on a statistical inference that, based on his poverty or gender, treats him as the human equivalent of a loaded gun”).
 Insofar as prison reduces opportunities for lawful employment and socially-positive association after prisoners’ release, incapacitation may militate against the last two of these goals.
 This problem seems to lurk below the surface of criminal and tort law and surfaces in, among other places, disputes over the insanity defense, diminished capacity, and the culpability of addicts.
 In the absence of an overarching theory of punishment, the same approach will help the judge balance the sometimes contradictory goals of punishment against each other.