This strategy is founded on the rationale that prisoners are incarcerated because they can not commit crimes that affect 'outside' society while they are in prison. Since individuals have different propensities for committing crimes, this strategy would give longer prison terms to those people with high crime commission propensities than to those with low propensities. The amount of crime is thereby reduced, while any other objectives of imprisonment, such as retribution and deterrence, are ignored. The study finds that any such strategy is subject to error because offenders' true crime rates may differ from their estimated crime rates. A Bayes estimate of the crime rate rather than a maximum likelihood estimate is found to have both a higher expected number of crimes averted and a lower probability of assigning long sentences to offenders with low crime rates. This is demonstrated by calculating the distribution of true crime rates for hypothetical offenders whose estimates are above a threshold. Graphs and formulas are included. (Author abstract modified.)