Memorizing algorithms doesn’t really make sense to me. Even the trivial ones have edge cases that you might forget of you try to implement it from memory. You aren’t going to ship it before double-checking a textbook or Wikipedia, right? And who wants the naive implementation anyway? Look up the definition to have a really 100% solid description of the steps, so that you can use that as the starting point for optimization (assuming a tuned library doesn’t already exist).

It feels like a strange feedback loop or something. People memorize algorithms to pass interviews, instructors help students memorize them to help them get jobs, then once they get into the jobs they start asking interviewees about the questions to check if they were paying attention in class, or something.