A familiar product, advertisers believe, sells better when given new packaging.

Sentencing on grounds of potential future criminality has been a familiar product in criminal justice. Proponents of the traditional rehabilitative ethic have long held that, while curable offenders should be treated, those who are bad risks should be identified and isolated from society. In the last decade, however, predictive sentencing lost some of its cachet. Researchers raised doubts about the accuracy of forecasting techniques-particularly about their tendency to “overpredict,” that is, erroneously to classify persons as expected offenders. Questions were raised as well about the ethics of gauging convicts’ punishments on the basis of their expected future choices. In part because of these doubts, sentencing reform efforts, such as the guidelines developed by Minnesota’s sentencing commission have tended to emphasize the seriousness of the defendant’s criminal conduct while limiting the role of predicted future criminality in determining the sentence.

Now, however, the idea of predictive sentencing is being revived. Several RAND Corporation prediction studies have received considerable attention in our profession and one such study, by Dr. Peter Greenwood, has obtained nationwide publicity. In the Greenwood study, the idea has been strikingly repackaged: instead of being called merely the new, improved predictive sentencing, it has been renamed “selective incapacitation.” Substantial advances have been claimed in techniques for identifying repeat serious offenders. The Greenwood study also purports to demonstrate that selective incapacitation strategies may lead to significant reductions in crime without increasing the total number of offenders incarcerated,” thus promising to deal simultaneously with the intractable problems of serious crime and prison overcrowding in America.

How valid are these new claims? Does selective incapacitation really offer new solutions? Or is it a shopworn notion now decked out in flashy new guise? These are the topics of our article. We will address the following issues:

1. Traditional statistical methods of predicting individual criminality were rather crude and error-prone. Have the new prediction devices, such as Greenwood’s, significantly enhanced predictive accuracy?

2. Traditional prediction studies sought only to measure the likelihood of recidivism and did not address overall crime rates. The Greenwood study, for the first time, offers supposed evidence that predictive sentencing can reduce the incidence of crime. How valid are such claims, and how sound is the research on which they are based?

3. What of the ethics of selective incapacitation? Have any new answers been offered to the moral doubts surrounding the sentencing of convicted criminals on the basis of the crimes that they are expected to commit?

4. What of resource allocation issues? Traditional prediction studies did not address the question of sentencing policies’ impact on prison populations. Greenwood does, and claims that a selective incapacitation strategy is capable of alleviating prison overcrowding in a way no nonpredictive strategy can. How valid is this claim? Is there convincing evidence that selective incapacitation calls for any less use of prison resources than, say, a desert-oriented strategy?