Most tech companies I've worked for in the past years have had about 30-40% new people every year through a combination of natural turnover and growth.
You're right that if a company never grows and nobody ever leaves yet they still try to isolate 6% to fire every year, that's a bad practice.
However, nobody gets hiring perfectly right. When you're hiring hundreds of people every year into a department, you're going to get a couple people who don't work out.
I think people on HN get nervous around these discussions because they imagine companies as static collections of individuals, but at a high growth company you actually have a lot of new people coming through all the time.
It's also bad when companies never fire anyone. If you just let everyone stay until they quit, you accumulate a lot of people who aren't working out while their high performing coworkers get frustrated and leave. I've seen this situation too and it's not good.
If your hiring quality is uniform, there’s only about a 5.6% chance the new guy gets fired next year. Instead they’ll go after someone else. For an average tenure of 3 years that’s closer to 1/6 chance of ending up on a PIP. Higher if they hand out extra PIPs in order to hit the quota.
It is often clear that some teams are about 50% underperform while others are close to 0%.
I've worked on many high-performing teams where you couldn't find 6%, or 1-in-18 of people who shouldn't be there.
However, I've never worked for an entire fast growing company where I couldn't readily pick out 1-in-18 people who were clearly underperforming and everyone knew it.
Hiring is far from perfect and these companies are growing quickly. Combined with turnover, you're bringing in a large number of new developers every year. Some people are really good at Leetcode and interviewing, but don't actually like to do work at jobs, for example.
IMO, the problem comes when these arbitrary thresholds are applied at low levels like individual teams. If you take a group of 18 people and declare that 1 of them must be fired every year no matter what, it's terrible. On the other hand, if you were to look at a large department of 1000 people and went through everyone's performance closely, I don't think it would be hard to identify 60 people who weren't working out.
The other problem is that some people have become really good at gaming this system. Every time I've worked at a company that does arbitrary culling of employees, there was always a contingent of old timers who had mastered the system years ago and never found themselves up for consideration to PIP. Either they were friends with management, had some arcane knowledge that they refused to let anyone else access, or they were really good at buttering up their manager come performance review time. I can think of one guy who was always fighting with everyone and refusing to do work, but would become the sweetest, most helpful hero a couple times a year when some high profile issue came up. He'd work 16 hour days for a couple weeks, emergent triumphantly with some solution, then disappear again for another 6 months. Untouchable, yet clearly the lowest net performer on the team.