Performance cannot be evaluated in a vacuum, and likely has as much to do with your environment as it does your experience and talent. A _high_ performance company, if such a tech company even exists, would do everything in their power not to lose institutional knowledge. Throwing out a low performer is the wrong tact if you have more products and teams then you know what to do with it, it is much better to shuffle that employee into a new environment a couple of times to see if anything sticks.

Because it encourages pathological behaviour.

Amazon's attrition goal is an open secret everybody knows. Once everybody knows that someone else's success is your failure, people start backstabbing eachother. They stop helping eachother.

The problem is that the attrition target is a quota that all of us had to fill.

This assumes an organization can effectively identify who are the underperformers of a company. I've yet to see this proven out.

More concretely, imagine you're on a team of 4 and management says the lowest performer by some metric will be cut at the end of the quarter and replaced with some other random person.

Does this sound like a healthy environment? It creates perverse incentives where folks become more concerned about outperforming the next person than actually working toward the primary goal of shipping a product.

If your goal is to whittle a group of people down to one, then stacked ranking is the right tool. However most, if not all, organizations are created to achieve larger tasks. And this requires the cohesion of a team. If you think of only individuals, then you'll miss the forest for the trees.

How high performing was the stack ranking regime of Ballmer-era Microsoft? Thought it was industry consensus that stack ranking led to a toxic, highly political organization propped up by perverse incentives. This is history repeating.

Stack Rank is bad is because mid to low performing developers will put a lot of effort to not get fired. That means they're not fully putting all their focus and effort on the company's goals. Their views will focus on short term gain/success because they expect to eventually get fired. They'll try to get an accomplishment notch on their resume and preferring speed over quality and design. Hopefully their managers will see their accomplishments over the other brown nosing under performers and won't be around for the fallout of their low quality work. That's only for the ones who have hope. The hopeless ones who knows they can't make it won't care about the quality of their work and would focus more on finding another job.

If you have a whole team of high performers, you still are going to have to fire one of them. The high and low performers are not evenly distributed across teams because good people do their best not to take bad jobs.

1. Is it actually good to fire the lowest performers? In some cases, yes, if those performers actually add negative value by creating more problems than they solve. But if the simply are less productive than others, but still add positive value overall, maybe less so. You may lose institutional knowledge, team cohesion, or other intangibles that do not show in metrics.

2. Does your system actually correctly identify the lowest performers? If you believe that, you have more faith in metrics than I do.

3. Does every team have lowest performers that should be fired? I think that is the real toxic component here. I'm sure that all of us have occasionally met an underperformer that the company would have been better off without. But an automatism that every year, a blood tribute of N% shall be paid leads to all sorts of perverse incentives.

Sure. But is "Rank and Yank" the best strategy for that?

https://en.wikipedia.org/wiki/Vitality_curve