[VLM]

There is a slight mistake in the article toward the end where the interviewer asks what good the BJS statistics are and the response is polite waffling and its wrong. What its actually extremely useful for is numerical prediction of how many beds you'll need in 10 years so build X prison beds per year etc, based on greater population growth and "repeat customer" rates. I'm not talking about conspiracy level stuff or isn't the prison industrial complex evil signalling, but just predictive numbers. The meta problem that the article does overall get correct is trying to apply "how many beds to build" statistics to "what fraction of prisoners turn their lives around" is totally useless as the interviewee properly states. In that way the BJS stats are awesome for construction budget projections while being simultaneously totally useless for pontificating on cultural problems. The dude interviewed basically gets this correct in long format, only screwing up this one question.

No one else has caught the obvious startup analogy of freemium model whales, AFAIK. Most people playing freemium games do not pay in. On the other hand a huge fraction of freemium revenue comes from whales. Therefore you can make all kinds of weird inaccurate statements about the revenue vs population distribution. Like if all your revenue comes from whales, all your customers must be whales, right? Or all your revenue comes from repeat customers therefore most customers are repeat customers. Strange logic like that.