Thread by @danluu

One thing I've wondered about for a long time is why I fail interviews at such a high rate, e.g., see danluu.com/algorithms-int….

People who've mock interviewed me have a variety of theories, but I don't think any of them are really credible, so I'm going to wildly speculate.

The most suggested reason people have is that I get nervous and that's the problem, which people think because I do fine in their mock interviews.

That's a contributing factor, but I only get nervous because I've failed so many interviews and I didn't used to get nervous, so

there must be at least one other cause.

Another explanation that's consistent with the evidence is that when I say something "stupid sounding", people who mock interview me (who know me) assume it isn't stupid whereas interviews assume it is stupid, e.g.,

https://twitter.com/danluu/status/1447268693075841024

My feeling is that this comes from my being an experienced engineer with a non-traditional background, which means there's a lot of stuff that people expect "all programmers" know that I don't know, which makes me sound like a fake programmer, as in danluu.com/culture/#appen…, e.g.,

I've never written production code that talks to a DB nor have I ever written production code that issues an RPC or calls an external API.

This kind of thing elicits the strongest reactions in design interviews. Once, I asked someone what MVC was, which didn't go over well (I'd

heard of MVC before, but I've never worked on anything where that would be necessary knowledge).

Something like that comes up in every "design a webapp" interview, which has been the vast majority of my design interviews.

I think I've only had three design interviews that didn't leave the interviewer with a negative impression. The three interviews were "evaluate a fundamental new disk technology and how that impacts the datacenter", "design an ad exchange", and "design the chat UI for X"

At the time I was asked the first question, I'd never done work in a datacenter, but evaluating performance has been something that's been in my wheelhouse since I was in high school, so that one was an easy one.

The other two questions were fine because, in both cases, the interviewer was prepared to interview someone with no relevant knowledge (non-ads and non-frontend/UX folks) and didn't assume the candidate having a gap in their knowledge w.r.t. the topic meant they were an idiot.

I think this is very unusual; when people say "design interviews let me see how someone thinks" what they unknowingly mean is "design interviews let me see how much relevant knowledge someone has, and if someone has the relevant knowledge, I'll evaluate how they think".

Although people often say they're looking for reasons to pass a candidate and not for reasons to fail a candidate, many people (unknowingly to themselves) view missing some "standard" knowledge as a reason to fail someone and, in fact, are looking for many reasons to fail someone

BTW, I think design is one of the areas where I add outsized value at work.

When asked to review a design, I frequently find issues that would cripple or kill a project if unaddressed (I also frequently don't find anything, another side effect of my background, I think).

A few times, I've been unable to convince the implementing team that an issue I thought would be fatal was actually serious and they went forward with their design and the issue has been fatal every time, which gives me some confidence I might be right when I make that call.

Another component thing goes badly is whiteboard coding, but probably not how most people would expect.

I'm extremely out of practice at this now but, in 2013, I did some competitive programming problems for fun.

Competitive programming problems are so much harder than what you get in interviews that, if you can do semi-easy competitive programming problems, a standard whiteboard algorithms question should be solvable at the speed at which you can write down the solution.

That didn't appear to really help me, though. At Palantir, the interviewer literally ran out of questions (IIRC, he had 5 or 6) and seemed embarrassed.

Palantir walks you out of the interview early in the day if they fail you early and they walked me out after that interview.

Another failure mode was that people would accuse me of having seen and memorized the question because I answered it too quickly. That wasn't the case (I was very up front in cases where I'd heard the question before), but people literally didn't believe me.

Many folks advised me to pretend to struggle with the question and then have an a-ha moment since, when people say that an "interviews show you how someone thinks", that's what they're actually looking for, not how you really think

But I don't like the idea of lying, so I didn't

I actually seemed to do a lot better in algorithms interviews a couple years later, when I was moderately out of practice and would struggle a bit instead of just being able to write out answers at the speed I can write on a whiteboard.

The last time I had coding interviews (4 years ago), I was out of practice and failed a number of interviews because I struggled too much.

It's much harder to actually struggle the right amount than to fake struggling the right amount, which is why so many of my friends fake it.

Another thing that I believe counts against me is that I'm candid and honest in behavioral interviews. I know what you're supposed to say but I refuse to say that instead of the truth on general principles, which I suspect has cost me a number of jobs.

I've been on the other side of the table and seen how hiring managers and other folks react to real candor to their questions instead of saying the thing "everyone" knows you're supposed to say.

I've tried to stand up for this but have generally been outvoted.

A concrete example is that the hiring manager panned a candidate because, when asked, said some negative things about their employer, which allegedly indicated that the person was negative, could be difficult to work with, didn't have a growth mindset, etc.

I pointed out that I was on the loop because I'm considered easy to talk to & work with and I answered the same question v. negatively. The HM said I must've shown a positive, growth oriented, mindset and my manager (also on the loop) slacked me with, basically "you did not".

I think that biasing towards hiring people who game their answers and say what "you're supposed to say" and away from people who are honest is a poor hiring strategy but, by revealed preference, we can see that most hiring managers want to hire people who are dishonest.

Two people who know me independently pointed out that the remaining 20% of this might be why I fail interviews at such a high rate, calling out the same specific mannerisms. I wish I'd had these friends 20 years ago, but

https://twitter.com/danluu/status/1477001191548542980

I'm not sure it's worth the effort to do anything about it at this point.

The remaining things would be fairly high-effort to change and, at this stage of my career, the majority of companies I talk to will give me an offer without putting me through a normal interview.

I think this is multi-causal, but this bit would explain why, e.g., when I smoke an algorithms question, people are often utterly convinced that I memorized the question even though competitive programmers also smoke algos questions and generally believed to be genuine.