Stop rules are fine, we'd probably go insane if we didn't use them. But as soon as somebody questions the validity of a stop rule, it should cease to be a stop

Stop rules are fine, we'd probably go insane if we didn't use them. But as soon as somebody questions the validity of a stop rule, it should cease to be a stop rule and become subject to rational examination. Once you are satisfied that it is indeed valid, it becomes a stop rule again... until next time. In other words, we should be comfortable with switching between two modes: a "production" mode where we rely on our stop rules to get things done efficiently, and a "debugging" mode where we can step back to examine our source code. Failure to switch modes like this leads to assholery as exemplified by the various examples mentioned in the article. The most dangerous stop rules are the ones you don't understand. Most non-IT business executives have no understanding of how encryption works. It's magic to them. Therefore, they are incapable of stepping back and examining it. They segfault as soon as they try to switch modes, so after a while, they don't even try. Most DRM schemes are the same. It's a mantra that they chant without understanding. And from the point of view of consumers who suffer from the consequences, sufficiently serious incompetence is indistinguishable from malice.