>Famously even if you know the answer pretty well, say because you were the original creator of the thing, but answer in a way that they can't link to the answer key, you still fail.

Which is the most ridiculous part of it. You are being examined and evaluated over trivia in a subject that the examiner likely has no idea about and where there are only 1-2 possible "correct" answers.

That's like sending a janitor to "source" surgeons by asking questions to people in white coats in a competing hospital about appendectomy.

Yet this is considered effective (and acceptable) somehow ...

What alternative would be more efficient and acceptable? A typical team of 6-8 replaces two persons per year. The number of resumes per opening is measured in thousands.

> A typical team of 6-8 replaces two persons per year.

How is this fact not seen as a complete failure of hiring practices or company/team leadership?

Internal mobility is strongly encouraged. People typically move after promotion, which is expected after 2 years. Some people don't like to move, bringing it down to about 2/year. All working as intended. Then, people moving out tend to want to join new teams, for the career growth opportunity, while you want new hires in established teams.

Note: that's anecdotal from a couple years of observation. I haven't looked into any Google-wide statistics (and if I did I would not blabber about them). But I'd be surprised if this was way off.

In an industry where the median tenure for a SWE is ~2.5 years, that's "doing no worse than the industry average".

Turns out upper management rates predictability quite high.

Are you saying a company with the resources of Google, which expects all of its employees to answer trivia and whiteboard answers in interviews - all of that brainpower and money can't invent a better process?

I don't think it's a matter of impossibility. The current process is simply a local optimum. Most practicing software engineers can pass the trivia quiz. Most time wasters can't, unless they googled for the answer key. The result is a pool of candidates that usually can understand a real interview question an engineer will ask. The system works, at the price of being annoying.

If there's only a single correct answer that the interviewer doesn't understand and an expert might get wrong, it's people who google the answers that have the best chance of passing.

I tend to ask open-ended questions. Let them talk about code, and I can probably figure out whether they're full of shit or not. And a real coding challenge proves better whether someone can code than any code question or whiteboard challenge.

But you're an engineer, aren't you? Do you think you could figure out if someone is bsing you about their experience in mRNA manipulation?

No, but that's why you need to let people's skills be assessed by people who are able to assess those skills.

Having too many resources tends to lead to sub-optimal processes. Just look at the Joint Strike Fighter program.

If you want to mechanically apply a trivia test, then at least make it multiple choice.

You will still be choosing people through a mechanical procedure, what is quite stupid, but it's less stupid than that way.

If the applicant has graduated from a university, technical school, whatever, then look at the records from their exams there. Also taking into account the average level of the university, technical school in question.

That'd be really unfair on people who were busy e.g. gaining actual work experience

Wait, seriously? That's an absurd screening question.

> Famously even if you know the answer pretty well,

Everything above would make sense if the question was "what is an inode?". But it was "what is an inode not?". That wording doesn't make much sense to ask anyone, no matter how you spin it.