Something I don't understand: on the one hand, people on HN seems to enjoy this kind of articles. On the other hand, they upvote any article who decries interviews where candidates are asked things that involves knowing some theory.
So, if we think knowing theory is useful in cracking hard problems, why is it wrong to asses its knowledge in an interview?
HN's readership isn't homogeneous. There's room for both points of view.
I've done a lot of useful work in feedback systems without ever really grokking Laplacian notation and the notion of complex frequency in general. A lot of the actual numerical methods used in real life boil down to a few canned formulas. But I know enough about the underlying theory to appreciate where the canned formulas come from, and fully intend to sit down some day and go through the whole process. Articles like this are interesting if only for the occasional gems in the comments, such as John N.'s pointer to Maxwell's 'On Governors' paper that I'd never run across before.
At the same time, I don't see much upside in making hiring decisions on the basis of whether someone can regurgitate a bunch of textbook math. I'd rather spend the interview talking about control problems the candidate has dealt with personally, how they were handled, and what the candidate learned from them.
While I don't claim to have the answer to your question, I think there's a jump you're making that may not be so straightforward and somewhat explain things.
You say "people on HN seems to enjoy this kind of articles" which seems reasonable, given the comments here. But then you jump to "we think knowing theory is useful in cracking hard problems".
Going from the first to the second is not quite so clear. That is, someone may enjoy such an article and even learning some theory, but not necessarily because they think they will directly apply it. People sometimes just enjoy learning stuff or reading about it and then forgetting it.
You also make a second jump, because other factors may be involved. Maybe the theory asked in the interviews is completely unrelated to the things involved in the job. The job may not even require cracking hard problems. These are frequent occurrences -in my experience, at least-, and clearly seem compatible with thinking that knowing theory is good in general.
Because programming is rarely about cracking hard problems. That's actually why the people HN enjoy the kind of article, it takes them away from the drudgery of their job.
Programming and engineering are neighbors, so it is interesting to see what's going on over there. But many engineering analysis tools are not really useful for normal programming jobs (unless you happen to be working on engineering software, control systems, or something like that).
So, if we think knowing theory is useful in cracking hard problems, why is it wrong to asses its knowledge in an interview?