Recent example: I was asked to do a density map with about ~20 minutes remaining in an interview. I could immediately see the performance issue of the naive solution (re-summing the same positions), but I needed some time to think through an efficient solution. But because of the time limit and the fact that I had someone watching my every step (i.e. there was more performance than problem solving going on), we agreed that I would take the naive approach and simply sum all of the positions as quickly as I could. When I wasn't asked back for a second round, it was obviously in part because I had resorted to such an ugly and inefficient solution.
I don't blame the interviewer—he was actually one of the best I've ever interacted with—and I actually think the problems selected fit the role very well. But at the end of the day I don't feel that format gave a very good snapshot of my abilities. (Case in point, I looked up solutions afterwards and an intuitive approach I had thought of was very close to the optimal solution, but I didn't feel I had time to go down that route.)
I don't think there is a perfect interview format. They all have downsides. But this is definitely a tradeoff you're making with timed algos.
Are you hiring for a fire department? The Navy Seals? Engineering section on a nuclear submarine?
Maybe, just maybe, things would improve if you had a few people who said 'no way' when someone tries to crank up the drama on your team instead of just grinning and bearing it.
Capitalists don't reward the quiet workers. They exploit them. Billionaires and wannabees do it remorselessly.