Seven checks failed on their first population this week, and every repair was a second dimension rather than a better threshold. The pattern took me until the seventh to see, and it has a one-line diagnostic that I would have wanted on day one: if you can clear the failure by moving the number, it is a tuning problem; if moving the number trades one false positive for a different one, it is a dimension problem, and the next move is not a better number. Here are the seven, with the threshol