The fact that there are more than 0 interviewers who ask questions like this is pretty damn sad.
Since then I've stopped asking questions like this. Trick questions and logic questions I think are a bad metric. A person can be great at sudoku but be a horrible developer in terms of writing maintainable code. Instead, I tend to just talk about the macro-level stuff. "What have you built? Can I see it? How long did you maintain it? What was the trickiest part? I imagine it had performance issues; did you spend time optimizing it?" I do ask some tech questions. For example, for web developers I ask "Describe how an XHR request happens in detail." But that is as deep as typically go.
Kind of like if you got an interviewer that responds with a blank stare if you talk about version control and says they just make copies of the directory, or something like that.
I can't fanthom how would one come to the conclusion that there is a difference between the two otherwise.
If an interviewer makes some crazy assertion, then I pretty much flip into the mode of "how do I get out of this interview ASAP, as there is no way I'm working with this idiot". Whereas in the office, I'm the guy that sends the really difficult mass email that calls out his boss, if it is really needed. Engineering is not interviewing, and vice versa; why would you expect identical behaviors in both?
And yup, I am self selecting out of the situation where the interviewer is not an idiot and just 'testing' me. I'm fine with that, because I'm also selecting out of a job where people aren't willing to treat me with respect.
Interviewing is largely voodoo; we shouldn't be inflicting cognitive pain on people based on our random selection process. The only measure that has ever been shown to be predictive is past performance. So let's talk about my past performance, and leave this amateur hour psychological testing in the trash bin where it belongs.
Don't try something because you sit down and think "hey, I am going to create an artificial situation, and then guess at what a good candidate would do[1]." Then see if he does that.
[1] Which is often "what I (imagine I) would do."
Or maybe that's just me; nervousity in presentations or interviews or even talking is IMHO an indication of not knowing what you're talking about. And knowledge is confidence.
I've interviewed a number of people who were incredibly confident in their language, voice and body language who clearly had no idea what they were talking about.
Similarly, I've interviewed people who clearly knew way more about the subject matter of the job than I did but who were incredibly nervous during the interview either because they were intimidated by the idea of working outside of academia, real/preceived language limitations, or because they were just incredibly shy and awkward.
I've generally tried to look for people who fit the sweet spot in-between: confident and sociable enough to be reasonably comfortable during the interview and humble enough to not be brash. The more "customer/client" facing the position, the more I care about the confident and sociable part.
Personally at the end of the interview (as the interviewer) I would refer back to any of those sorts of questions and say.. "actually I was trying to catch you out, you did good there", or some such.
This not only stops them doubting themselves, but also clears the air if I was to come back to them and offer them the job. Who knows, maybe the interviewee ran off to SO and categorically proved me wrong.