Document Type



Thousands of sex offenders in the United States are being held indefinitely under civil commitment programs. The analysis in this Article suggests that none (or precious few) belong there. Specifically, in a large dataset, an instrument as good as the one most widely used by experts (the "Static-99") could not identify even one sex offender who met the legal standards for commitment. Supplementing such instruments with additional information does not appear to improve matters, so the failure of the instrument is profoundly disturbing. There are three possible responses to this failure: (1) improve instruments to meet existing standards; (2) lower the existing standards; or (3) abandon sex offender civil commitment. This Article focuses on the first response, identifying and correcting flaws in the most widely-used instrument. But the greater significance of the Article is to reframe the debate around the other two potential responses. Can we predict the future well enough to justify the indefinite detention of "dangerous" people?