Calculating chance of re-offending

A long New Yorder story describes prosecutor psychologists who try to predict sex crimes, like psychics in the movies:

The prosecution’s expert, Amy Phenix, a forensic psychologist who makes her living testifying at civil-commitment hearings around the country, maintained that the stories John had told Indy-Girl were true, because they were “consistent with his patterns of sexual arousal.” She drew heavily on John’s admission at his probation-violation hearing, in 2005, that he did not have “enough control.” She said that John had roughly a 24.7-per-cent chance of reoffending within five years, based on her scoring of the Static-99. Phenix co-wrote the coding rules for the Static-99, which has been cross-validated on different samples of sex offenders more than sixty times. She predicted that if John was released he would “reinforce his deviance” by looking at child pornography, but “ultimately that will be insufficient and he will, in my opinion, then seek out children for sexual activity.” She explained, “It’s just almost like an accident waiting to happen.”

Claiming 24.7% should be obviously bogus science. No psychologist can predict anything with that sort of precision.