Comment by acjohnson55
1 day ago
I was a contestant on Jeopardy, and led going into Final Jeopardy 2 out of the 3 matches I played. So I'm an above average Jeopardy competitor. My edge is not in pure trivia. I'm probably below average, by the standards of people who have won a match on Jeopardy. I suspect I'm above average in dealing with the competitive aspects of it and the stress management. It makes a big difference in a fast-paced game. The reality is that Jeopardy is not just a trivia game, but also a game of reflexes, decisionmaking, and stress management.
I wrote about this here: https://acjay.com/2023/11/12/everything-all-at-once-inside-a...
There are certainly jobs where you need someone who is cool under intense pressure, where what happens inside an hour matters. That's what these interviews are revealing. But I think this is a tiny minority of dev jobs.
However, I also think when people talk about hiring practices, the elephant in the room is that the true purpose our interview processes have evolved to serve is taking the candidate pool from a lot of people to something that is decidable. We can't fully evaluate 500 candidates, but we can evaluate 5. The process of weeding out 99% is designed to be inexpensive, at the cost of accuracy. The hope is that you'll have one person squeeze through the filters that is a good fit. If that happens, then all the false negatives are tolerable.
But we really, really don't like to admit this. We tell ourselves that we are running interview processes that are predictive of performance when there is actually very little evidence to support that, and plenty of reason to doubt that it's true.
If quality were paramount, I think we'd do hiring completely differently: https://acjay.com/2019/05/16/hiring-developers-is-easy/
> There are certainly jobs where you need someone who is cool under intense pressure, where what happens inside an hour matters. That's what these interviews are revealing. But I think this is a tiny minority of dev jobs.
I disagree. Every software job I've ever had has involved times where the rubber meets the road and production software has crashed, deadlines fall behind or difficult decisions have to be made under pressure. The fact that it might be 5-10% of the work at most doesn't change the fact that it's the most critical 5-10%.
That's a different thing than what these interviews effectively test for. You're typical live coding asks the candidate to do an unfamiliar task under observation, on their own, and generally without the typical resources at hand.
Even if it were a good proxy for the sort of stress that occurs episodically in the real world, I would argue that the extent to which that skillset is helpful, it is grossly overrepresented in the interview process.
Frankly, I can't be arsed to watch people stumble around for 4 hours. It's an hour for my sanity and yours.