With the exception of a very few prolific offenders, sex offending is not a high rate activity. Even recidivist offenders will commit only a small number of offences in their careers, and these may be separated by intervals of years. Because of this, anyone setting out to predict reoffending by sex offenders will do best if they simply assume that none will reoffend, in which case they will be right more often than not. But such an approach, of course, would be criticised for being oversimplistic. Sex offenders have a history, and there is a common belief that if we know enough about an individual's past we should be able to predict his future with great accuracy. This has led some workers to claim that if the right variables can be discovered and plugged into a risk assessment algorithm, then the resulting desktop prediction of risk will outperform any competing clinical method.