Hacker News Reader

Comment by yzmtf2008 | Hacker News Reader

yzmtf2008

In my experience, doing leetcode to such an extent is in general useless for job searches. Companies are not stupid, and your ability in solving the technical problem is only a very small part of the interview rubric. I’ve done 5, maybe 10 leetcode problems and never had an issue interviewing.

View on HN

Having been on several different company hiring boards and reviewed literally thousands of interviewer feedback posts, let me assure you that interviewers happily ask leetcode questions: even when they've been told expressly not to ask leetcode questions by their company.

The fact is that most software engineers just don't think interviewing is an important part of their job, so there's no impetuous to improve. They were probably asked leetcode questions when _they_ were interviewed, so that's the established norm.

Leetcode questions also have well defined paths for success. Coming up with your own untested question is not guaranteed to be as productive in the interview. And once you’ve developed and used your question enough to be tested, it’s up on leetcode anyways.

Goodhart’s law is unbreakable.

Counterpoint. My Google interview was entirely whiteboard leetcode problems. There wasn't a single non-technical aspect of the whole interview.

Same for Amazon- hours of it. That was the last time I went to an interview without first asking about the interview format. Had I known what I was walking into, I would have laughed at them and turned the interview down.

In fact, I did exactly that to a company somewhat recently (small fish pretending to be FAANG rockstars) though without the laughing bit.

Lucky. I found the culture fit and architecture interview rounds to be the most difficult.

It may be only a very small part but not knowing how to invert a binary tree may well knock you out of the process. The process is designed to cull as many people as possible as early in the process as possible. It is accepted they might miss a few good people but it is more important that don't hire a bad one.

You don’t need leetcode to know how to invert a binary tree. You need solid foundation on algorithm and data structure, both of which are still relevant today.

I find it funny how inverting a binary tree is used as an example of a bad leetcode problem. This problem seems more like a fizzbuzz test, anyone who knows recursion should be able to solve it on their own.

There’s a difference between being able to solve the problem given some thought, and being able to answer and program 2-3 of these problems in 45 minutes. The latter is what Facebook expects.

No idea about 'bad', but https://duckduckgo.com/?q=invert+binary+tree turns up nothing but interview prep sites.

But your slow reasoning and problem solving approach will be compared to another candidate's fast and savvy 'I've memorised this already' approach.

I don't know a single person across my org that will ever have to invert a binary tree. It's a hilarious problem that helps people who know it's a kind of problem asked, and hurts those who walk into it blind.

I should have put a smiley after the bit about inverting a binary tree. :) It is a common joke. The rest is true though.

I think the point of fizzbuzz & inverting binary trees type of questions is to filter out people having issues solving them blind

> Companies are not stupid, and your ability in solving the technical problem is only a very small part of the interview rubric.

I've not interviewed at a single company in the bay area where this is true.

It’s maybe a small part but the first part—-the technical phone screwn. Maybe I’m in the minority but I don’t write semi obscure algorithms for my day job so I need explicit prep for TPS.

There are a lot of top paying jobs (e.g. FB) that you would never get into without LC grinding.

Reply on news.ycombinator.com