I see this argument all the time, but I can't find any other place that it comes from other than disappointment from those that didn't or can't pass those interviews. (Disclaimer, outside of college internships I've never interviewed for a FAANG SWE position nor have I ever worked for one).

Is it an objectively good measure of being a software engineer? Hard to say honestly. I doubt you'll ever find a truly great measure that you can test for in an interview. When I was interviewing candidates for my company, did I ask those leetcode algorithm questions? Not really. Maybe at most one basic tree traversal question (probably would fall under leetcode "easy" if I had to guess, but honestly the kind of thing a student would learn in AP computer science in high school). Most questions were system design and problem solving with a coding challenge (building something simple, not solving algorithmic puzzles). So by evidence of my own actions, I don't believe that they're the optimal questions for screening engineers.

That having been said, I don't understand why people are upset by these interviews. Who cares? If you really think it's suboptimal, then other companies who have "better" interviewing practices should be better at identifying undiscovered talent and hiring them. Better for you if you're hiring in those cases. Let FAANG fail on their own hiring practices. FTR I don't think they're that bad either, they just filter for a bunch of left-brained people who are good at math. Maybe they do make good engineers also. And if results are anything, clearly it's been working for FAANG for the past decade so who's to say that they shouldn't keep doing it?

> Look, if I can ‘crack the coding interview’, then I can certainly crack ‘how to do as little work as possible and stack paper to the ceiling while my stock vests’.

This is a reach (to put it mildly) and unfairly paints people who are good at algorithms as inherently unmotivated and whose primary goal is to cheat the system without any evidence. Are you saying another talented developer who isn't good at algorithms could not or would not hack the system as such? I don't see any reason to expect either to be the case. Hacking said system does not require you to be able to prove the runtimes of a Van Emde Boas queue, it just requires some common sense that any human being has.

> I wonder when the last time was that Mark or Sundar actually wrote any code they pushed to prod.

This is pure ad hominem and unrelated to whether or not these questions are good screening questions. I certainly hope that Mark or Sundar are not wasting even a millisecond of their time writing code and trying to get a PR out to production. It's one of the absolute worst uses of their time. But while we're on the topic, Mark literally built the first version of Facebook (to be fair, probably in a bad hacky way) and Sundar was a product manager so I certainly don't expect him to write code.