Earlier quoted context omitted.
Companies can self-consistently seek to hire the best and fire the worst from the the hired-pool. 1. Hiring is a noisy process. You can tune your hiring test to lean more towards false negatives or more towards false positives, but you cannot (at this juncture in history) eliminate a large amount of bias (error) from evaluations of candidates. 2. Regression to the mean ensures that any hiring process that sets a "hig…
Knew the HN SV crowd would find a way to spin this as "This is good for "
Whatever my beliefs or situation though, it's not hard for anyone to understand how a medium-to-large scale organization with a high hiring bar could make a rational decision to "hire the best" while still making an effort to release the least valuable developers in the company.
An alternative strategy is Google's. Google tunes their hiring process to produce almost exclusively false negatives rather than false positives (they very deliberately reject qualified candidates far more often than they hire under-qualified candidates). With this hiring strategy in place, it wouldn't make as much sense for Google to "weed out" developers.
And for the record I think 20th-century impersonal, statistical business thinking is toxic, but that doesn't mean I don't see its use or that I know of a better way to do things.