Yeah pretty sure this is BS. Bock talked extensively about interviews in his book. I recall him saying that after 4 interviews there were clear diminishing returns. If the interviews had no value, the diminishing returns would have been total at one.
I reread, and his claims were more specific than I allowed: brain-teaser-style questions have no predictive value; neither do "unstructured interviews", because the interviewers generally form an impression in the first ten seconds, that impression has no predictive value, and the interviewers then generally spend the rest of the interview confirming that initial impression.
However, structured interviews--interviews that consist of standardized domain-specific questions about relevant skills--do have predictive value. The problem with them is that interviewers don't like doing them.
I conflated the specific kinds of interviewing he deprecated with "the technical interview", I assume because those are the kinds of technical interviews I've seen. In thirty years or so, I've never seen the kind of structured interview Bock praised. Still, conflating the different things was my mistake. Thanks for correcting me.
