I was about to rant at length about the stupidity of using puzzle questions for software developer interviews when Google took care of it for me:
Google did the math and came up with two conclusions: The infamous puzzle questions had zero correlation as a predictor of success, and the questions themselves merely served to make the interviewer feel smart.
Really, the head of Google HR actually said that. He was unknowingly quoting what I’ve been saying for years.
I also inevitably hear interviewers say, “I like to ask these kinds of questions because they give me insight into how candidates think”, so I go a step further: That’s a total crock. Unless you have a Ph.d in cognitive psychology or maybe educational psychology, you’re flipping the bozo bit. Those questions do give me insight into how the interviewer thinks, that’s for sure.
I’m a little biased because not so long ago I aced a technical interview with a very knowledgeable CTO. The MIT and CalTech-educated CEO called later and ambushed me with puzzles. I couldn’t answer them as well as he wanted, so I didn’t get the job. I’m pretty sure I was the CTO’s choice, too, but the CEO nixed me. The irony: Their web site was ugly and broken.
I think they were too busy solving puzzles.
P.S. By the way, the answer interviewers are looking for with the manhole cover question: They’re round so they don’t fall into the hole no matter what angle you turn them. The correct answer, however, is actually historical: Pipe manufacturers could use existing pipe molds to forge them, so they were cheaper and faster if round.