The idea of removing the lowest performers, regardless of their ability, is a business idea from I believe the Jack Welch era at GE.
In my management classes at business school, it's taught as a cautionary tale of unintended consequences and bad management. So this policy is a well-explored bad idea from decades ago, and it seems typical of Silicon Valley to arrogant reinvent a wheel without looking at cautionary tales from the past.
Most people cannot even tell you how to measure a programmer's performance. But we're comfortable with firing the "lowest performing" developers? Lowest performing by what metric?
Honestly these "get rid of poor performance" policies are just abused to remove people who others have a personal dislike for, or similarly those who don't make enough friends to game the system.
Plus whose to say that even if we could measure performance, that it wouldn't be down to externalities outside of the individual's control (e.g. poor manager, poor team culture, bad project requirements, poorly written/documented legacy system they need to integrate with, etc).
To put this idea another way: good managers are humans, and exceptional humans aren't scalable. To solve that problem, companies try to deploy algorithms (via spreadsheets or software) that do scale, but end up being terrible managers.
In my management classes at business school, it's taught as a cautionary tale of unintended consequences and bad management. So this policy is a well-explored bad idea from decades ago, and it seems typical of Silicon Valley to arrogant reinvent a wheel without looking at cautionary tales from the past.