I got explicitly failed from a first round interview because I took too long to solve a graph traversal problem. I reasoned my way through and got to the optimal solution, the interviewer confirmed that it was the optimal solution, then said I "should have recognized a variant of Dijkstra's algorithm" and that I had failed for not knowing it in advance.

Personally, I think developing it from scratch shows more promise than memory, but a lot of employers are looking for the inverse right now, IME.

edit: and this was not a FAANG, this was a startup with a single digit number of employees!

Did that experience actually reflect what you'd be doing on the job, or was it just another bullshit interview experience?

Absolutely no relevance. And even if there was, I'd like to see a business where having to take an extra fifteen minutes to reference an algorithm textbook is the difference between success and going under.

What kind of time pressure do you put on candidates? Do you expect them to sit there live for 45 minutes and complete something that reflects what it would be like to work there day to day?

Depends on the interview or the level but I prefer open-ended situations like "imagine a local small business that sells widgets. They want a backend that keeps track of inventory and a frontend that displays widgets and prices to customers." Or whatever the idea is. Some candidates only need that much and begin asking questions, designing systems, writing code, whatever they want. Others need more guidance and I'm happy to continue defining how this app should work or how big the audience will be or any other details. To me it's about exploring their experience with the whole process holistically and letting them dive into their subject matter expertise naturally. From this idea, there is no "completion" of the exercise as no one is actually going to write a backend and frontend in 45 minutes.

But I think you learn an enormous amount about an engineer here. How will they use this time? What will they focus on? The details they choose to spend time on, how did they turn out?

I've seen Principals take the whole 45 minutes on system design asking extremely detailed questions and building a system far superior to anything I would have engineered myself (interviewing up level). I've seen Jrs hash out a simple database structure in 5 minutes in a comment block and spend 40 minutes writing a simple react app that shows this hypothetical data. Obviously this principal wasn't hired for their front-end knowledge and this Jr wasn't hired for their system design, but still, very valuable to see how they do it.

Do you provide feedback to rejected folks so they understand what they should try to improve if they re-apply? The lack of feedback from most places completely puts me off from considering them again, even if that’s what they do to everyone.

I am willing to provide a short communication some times providing information but the reality is usually a lot less interesting than people think.

Reality: I want to hire 1 person, we get say 5 past all the stuff before me, we interview all 5, and then we choose the one that we think fits the best culturally, personality wise, skill wise, etc. So our reply is usually "hey you were great, we wouldn't interview you if you weren't, we just found someone we liked a little bit more". Which isn't very satisfying but you have to think about it from my perspective too. How do you choose between 5 great candidates?

Ultimately, luck is playing a massive role here https://www.youtube.com/watch?v=3LopI4YeC4I

I definitely failed at one or two places on the behavioral / experience questions, but a few told me explicitly I need to improve my technical chops (I was grateful for the feedback). I felt like nobody was unfair at all, and that I just need to refresh my knowledge in the fundamentals. Basic DS & algos, concurrency, programming language trivia, etc.

Startups are kind of hard to get an interview at these days, you ideally should know someone already working in one, who can vouch for you. I don't feel like they have a low bar either, just less rounds.

My Facebook phone screen many years ago consisted of an HR-sounding woman asking me three questions, the first two I answered correct (according to her), and the third one I answered correctly (I checked afterwards) but didn't use the term that was apparently written down in front of them, was told 'wrong, it's <term>', was thanked, hung up, and got a rejection email a few days later.

In that case I think it's pretty clear I was rejected based on getting one answer incorrect. The phone call wasn't even that long at all.

And while other interviews have been longer, I suspect I've been rejected a few times based on me struggling with one particular Leetcode problem.

One guy didn't even let me finish it beyond five minutes of trying to figure out the recursive function, just rejected me on the call. That guy had a huge ego and I already wasn't certain I even wanted to work for him anyway (but I didn't have any other offers yet so I gave it a shot anyway), so not a huge loss there.