I've been coding for over a decade including things 1000x as complicated as that (think embedded systems from the ground up). I could easily take 20 minutes or even fail doing such a task because I'm literally never asked to code things in a professional environment with 3+ strangers breathing down my neck and judging me based on some simplified fizzbuzz.

I'm convinced white board / live coding interviews don't do much other than test for how you perform with a group of strangers pressuring you to jump through hoops publicly on which in a short time your entire value is judged (including determining say if you'll have money for daycare next week), which for most software professionals basically happens never except during an interview.

I have never been in an interview with 3+ strangers. Is that common?

It's extremely common beyond the new grad / junior level, from my experience.

I don't understand what's the value in having 3 people in the room looking at you solve a coding problem, at any level of seniority. Sounds like it's just wasting the time of 3 engineers when 1 could perform the task.

Maybe so (though I don't really believe you couldn't) but I gotta evaluate people somehow. And my job is to prevent false positives. False negatives are less damaging.

I could bikeshed on this, but I think this is missing the overall point. It's not that nothing is measurable, obviously some tech skills are measurable and obviously some people are going to do better on those things, but those things are less measurable as you move up the ladder, and less measurable for leadership skills over technical skills, and less measurable among groups of people with relatively similar levels of experience. I'm not saying nothing can be measured so you might as well hire the mailperson to the principal engineers job. The point is that you can have a fairly large group of equally qualified people for a job.

20 minutes might just be interview anxiety, especially if it’s a first question in the interview.

Then they have 70 minutes to recover. It never happens, though.

Being able to perform under stress could be a metric he values.

"Hey, the website was down for a few hours. Technically I could fix it in a minute, but have a downtime anxiety".