1. False positives: you get folks who do really well at DS&A yet who are really bad developers. I mean really bad. I wouldn't have believed it if I had not seen their interviews and then subsequent performance. I'd wildly guess that it's about 20-30%.

2. False negatives: you get folks who are really good developers, and yet for whatever reason, perform badly on DS&A algorithms despite practicing. I think this number is higher than false positives, probably around 50% or more.

If you're a FAANG company, you can afford to play these odds. If you're not FAANG, then you're killing yourself by requiring DS&A interviews, almost inevitably.

Both are contributing to destroying the profession for huge numbers of people, IMHO. I mean, that's good for me, because I'm almost certainly sticking around and generally do OK on DS&A interviews with adequate practice (which is a waste of time, since anything beyond a broad knowledge of the performance characteristics of various DS&As is totally unneeded for 99% of us). But it's not right, and I really don't like it.