New York Law Journal Reviews Harcourt's New Book

Against Prediction: Profiling, Policing, and Punishing in an Actuarial Age
Aziz Huq
New York Law Journal
May 25, 2007

The history of American law we learn at school is a Whig one. It is a story of inexorable moral and practical progress that often go hand in hand. Constitutional rights expand to rectify one form of injustice after another. The regulatory state flourishes to address structural crises of different kinds, from the Depression to post-war economic inequality and environmental catastrophe.

Recently, an important part of this upward trajectory has been the promise of law and economics. Perfection is feasible through an optimal aligning of incentives, pronounce the acolytes of the Chicago school of law and economics, whose footnotes are larded with citations to Judge Posner and Professor Becker.

Bernard Harcourt, who teaches at the University of Chicago, is here to tell us that this tale of progress is at best incomplete and at worst harmful to pressing policy goals. Harcourt is no armchair legal economist, with graphs and equations, divorced from a contextual understanding of real-world complexity. Harcourt’s analysis instead finds sure footing in a detailed understanding of practical social and cultural dynamics.

The target of Against Prediction is what Harcourt calls the “actuarial method” in criminal law. This is the idea that policy makers can look at large data sets about regulated population and, via complex statistical analysis, discern from the data predictions about the future behavior of the particular individuals. As Harcourt explains, the actuarial method in law arose first in the parole context in the 1920s to determine which inmates should be granted dispensation from the criminal sentence.

At the heart of actuarial method is the presumption of a connection: The fact that certain traits have been associated with criminal behavior in the past mean that people with those traits will in the future be more likely to offend. In the name of efficiency, the actuarial method thus jettisons the presumption of individualized scrutiny that most lawyers associate with the rule of law.

Today, nevertheless, the actuarial method is pervasive in the criminal law and elsewhere. For example, sentencing guidelines rest on an actuarial bedrock. “Sexually violent predator” state statutes, which the Supreme Court authorized in Kansas v. Hendricks as a form of civil commitment for certain former felons, also use actuarial calculi. Even prosecutorial decisions by the Internal Revenue Service are based on a complex algorithm called the Discriminant Index Function, which predicts the likelihood of a particular person violating the Tax Code.

After detailing the historical rise of the actuarial method, Harcourt makes three substantial assaults on its validity: one economic, one in terms of social costs, and the third ethical in dimension. The heart of his book, spanning almost two hundred pages, comprises this tripartite argument. A short review cannot do justice to the complexity or nuance of these ambitious arguments, but a summary of those claims is essential for understanding this book.

First, using racial profiling as his example, Harcourt shows that the actuarial method often rests on untenable assumptions about the “relative elasticity” of particular groups to profiling, i.e., the extent to which being profiled (or not) changes their behavior. This rather complex economic argument boils down to a simple proposition: When law enforcement resources are deployed scrutinizing one group particularly closely, others will have the incentive to violate the law more. Profile African-Americans on I-95, for example, and drug cartels will happily employ Hispanic or Caucasian couriers. Or, to pick an example that Harcourt elaborates only in passing in this book, profile Arabs at airports and terrorist groups will increasingly pump resources into finding cadres of a different ethnicity (something we know is already happening with al-Qaida).

Harcourt’s second argument is that profiling leads to a disproportionate share of a profiled group being swept into the criminal system, which imposes costs not only on those individuals but on whole communities. It also provides false confirmation of stereotypes that animated the profiling in the first instance. Because we, as a society, have given up on rehabilitation as a goal in punishment, it means that profiled groups are more exposed to crime-inducing environments, such as prison. The net result, explains Harcourt is a “ratchet effect,” which has the profiled group becoming increasingly over-represented in the offending population.

The final argument Harcourt makes moves beyond the realm of economic and social effects and into the world of moral philosophy. The actuarial method, argues Harcourt, is a problem of a scientific mentality that wants to know, to categorize, and to predict with increasing rigor and completeness. Harcourt traces these impulses back to the beginning of actuarial science in the first half of the century and then argues that this will to know and to control has usurped our other moral instincts about punishment. We have lost the desire to use punishment to rehabilitate or to express a shared concept of justice. Prediction thus warps our sense of the just.

Pulling these strands together, Harcourt gives a detailed case study of racial profiling in the highway search context. Yet, as he emphasizes, the critique of Against Prediction focuses not so much on the use of racial categories, but on the very idea of prediction based on past behavior that has become part and parcel of the administration of the criminal law in the United States.

In lieu of prediction, Harcourt makes a counter-intuitive plea for randomization in his final chapter. This means discarding actuarial methods of prediction, and it means making greater use of deliberate randomization in police procedures.  These recommendations will strike many readers as literally bizarre. But by combining a historical account of the development of actuarial methods in the criminal law, and by detailing three incisive critiques, Harcourt, is well-positioned to challenge deeply entrenched norms, and to re-orient the reader’s presumptions. Not all readers will walk away from Against Prediction convinced. Few, however, will be unmoved by the forceful and challenging ideas it contains.

As debate on profiling and terrorism grows sharper and reaches the courts, as it did in the U.S. Court of Appeals for the Second Circuit’s recent decision on subway searches, Harcourt’s book will remain essential reading for those who wish to look past the chestnuts of stale debate on crime and policing, and to see with fresh eyes the problems of the criminal law.

Aziz Huq is the director of the Liberty and National Security Project at the Brennan Center for Justice at NYU School of Law.

Copyright 2007 New York Law Journal

Bernard E. Harcourt
Aziz Huq