I actually enjoy those puzzles, but agree that they are irrelevant to the work. In fact, the very few times they are relevant is something of a special occasion! "Hey look, I have to design a slightly novel algorithm!" Most of the job is, of course, figuring out why things went wrong, based on imperfect evidence, which requires a good mental model of the entire system plus the ability to make changes to test and eliminate hypotheses. Very few, if any, technical interviews capture the essence of this act.