Imagine asking to prove a relatively difficult theorem. That's a similar type of question and it's a waste of time during an interview. Once you know the proof (know the algorithm), the idea might seem trivial, but coming up with such idea (inventing the algorithm) took people possibly a very long time in the first place.

You shouldn't expect it to be possible during the course of the interview for those who don't know it already, it makes no sense to expect that.

At best, the question will check if someone memorized such stuff. But I don't see a lot of value in that.