When people started asking these questions, the whole point was to see how people reason about a new problem they haven't solved before. There are almost no work problems that require you to regurgitate something verbatim you saw on leetcode before.
It's seriously making me question the intelligence of these interviewers. Although at the same time I realize it's mainly just to arbitrarily whittle down the applicant pool to a smaller number you can interview in person.
Yup. It's absolutely crazy to me to see the amount of people who are apparently doing hiring telling others here in the comments (and elsewhere on HN, etc) that they should "grind leetcode" while also NOT admitting that this is absolutely stupid. Like I'll give you a pass as an interviewer / hiring manager if your hands are unfortunately tied and you can't change your companies hiring processes.
But if you're an interviewer or hiring manager who is telling people to "grind leetcode" because you legitimately believe this is a good way to find good talent ... you need your head examined. All you're doing is filtering for people who've grinded leetcode long enough to memorize solutions. This has ZERO correlation to software development ability.
it is very sad. i personally would much rather someone walks through their problem solving approach and identifies shortcomings/edge cases than to regurgitate a memorized algorithm.
it's really easy to identify code that is not optimal at code review time. it is far more challenging to have a conversation with a algorithm-regurgitating robot that their entire approach was wrong because they misunderstood the abstraction or that their code wasn't needed because Larry is refactoring that part of the system to a separate service already.
There are so many leetcode problems most people can't fully memorize them all and many people just don't.
I actually think interview questions that are more qualitative are worse. I have to freaking guess the "design philosophy" of the interviewer and cater to his viewpoint which is often pointless or just flat out wrong.
Especially today with Copilot and GPT-4, if I need to merge two sorted lists with optimal time complexity, I'll just ask GPT-4 and it'll give me the leetcode-optimal solution adapted to my language and variable names. Leetcode is less relevant than it has ever been.
When people started asking these questions, the whole point was to see how people reason about a new problem they haven't solved before. There are almost no work problems that require you to regurgitate something verbatim you saw on leetcode before.
It's seriously making me question the intelligence of these interviewers. Although at the same time I realize it's mainly just to arbitrarily whittle down the applicant pool to a smaller number you can interview in person.