The later is too.
Its just renaming the same method and introducing a "objective" metric that is gambled again.
And this is where "dogmatic use of a rule instead of a guideline" comes in. Stack ranking down to the level of every team has bad results.
But it is equally true that in a large organization (and the only places that would even consider stack ranking have hundreds or thousands or tens of thousands of employees, not 10), you just don't have nobody but high performers. It's not possible.
I've discussed stack ranking before on HN, and I feel like there are a lot of people here who want to believe that you never need to terminate anyone. That's not true.
There also seem to be a lot of people here who believe that the fundamental failure mode of most organizations is being overly quick to fire. That's also not true -- though there are exceptions.
You have to understand that firing an employee is extremely stressful for most managers. So much so that lots of managers will very willingly live with a terrible performer who's a huge drag on their team rather than terminate that employee. And so we get management doctrines that are aimed at forcing managers to actually terminate their problem employees.
I'm not a fan of stack ranking, though I've also never worked at an extremely large organization. But it's not a fiendish plan created by mustache-twirling villains in order to cause people pain for sheer sadism. It's a deeply imperfect practice that should be looked at with a clear understanding of what motivates it. Dismissing it as sociopathy speaks a lot more to the commenter than the practice.
But even in those organizations, the "stack" doesn't consist of thousands of employees, as the unit on which stack ranking is done is lower-level units (often very low-level), not the whole organization.
But also: If we're talking about an organization where it is impossible for a manager of, say, a 100-200 person department to figure out who the 10-20 worst people are across his entire department -- where he's helpless and just has to say, "Well, I don't know, I'll punt down to my lower-level managers," that's prima facia evidence of a dysfunctional org. What is the doctrine that you feel is going to lead to better termination decisions in that org?
Stack ranking is based on forced rankings of employees by managers who are familiar enough with them to theoretically be able to provide an ordinal rank of their performance. There is no way this works in large groups, it only works in small subunits.
If you have actual objective measures of business-meaningful performance, you have to throw away a lot of the meaningful data to reduce it to a ordinal ranking and use it for stack ranking, which is why organizations that do have such measures use them for more useful accountability than is provided by stack ranking.
Once people realize that making other people look bad is in their interests the organization is doomed.
You need to get rid of poor performers but that should be based, as far as you can possibly manage it, on an objective standard.
And, as people have pointed out, that means that it can be gamed, and can be applied very inflexibly an inappropriately. But the virtue is that it gives people relatively little opportunity to get out of engaging with it at all.
Stack ranking usually takes as it's key input management subjective evaluations of staff performance, which is why the subject units are usually small suborganizations. It's mechanistic in what it does with that input, but it absolutely is not usually based on purely or predominantly objective measures.
Organizations which spend the effort to develop meaningful objective performance measures usually also spend the effort to find better uses of them than inputs to rank-and-yank systems.
Amazon has this same problem.
You can try to consider cross-team measurements, so as to fire the 'bottom 10%' across the company, perhaps preserving good teams. But how do you compare the value of an IT performer with an embedded firmware designer with an OS driver supporter with a backend schema architect? Their criticality to the company can differ wildly.
Further, after a cycle or two, teams should be stable and have suitable members. But the scythe keeps reaping, destroying what it was trying to build.
Yes, applying a strict stack ranking mechanism to an arbitrarily small team does not have good results. Agreed. That's what I mean by saying, it has bad results as a dogmatic rule. But the goal is not to reduce your workforce by 10% every year, either.
And if you can't hire because of budget constraints? Or you had a 5 person team that was gelling well, but now you've tossed one to the curb and replaced them with someone new (and burdened the remaining team with the interview process in the interim).
Any "fire x percentage" mandate is going to suck when you're looking at a team of high performers. Or, the manager is just going to hire a dud for the express purpose of having someone to fire the next go around. I've seen that happen more times than I want to think about.
That is far more detached from reality than stack ranking is, and far more destructive than stack ranking is to a large organization.
Stack ranking is exactly intended to deal with the fact that hiring a spotty, error-prone process. It says, "No, you can't only hire high performers, whether by having brilliant hiring processes or by training low performers. What you need to do is fire the low performers and roll the dice again. Because eventually the dice will roll high."