this sounds suspiciously like something a organization theory / game theory team thought up. Incidentally, we know that amazon hires such people from math and

this sounds suspiciously like something a organization theory / game theory team thought up. Incidentally, we know that amazon hires such people from math and econ departments for big bucks. Attrition is good for exploration and against other "stagnation" phenomena (textbook March/Levinthal early 90's). And if you are willing to throw aboard any ethics, then a "usable period" (pre burn-out) for some percentage of workers is a pretty good boost to performance. A secret focus list is probably some "well" thought out mechanism (as in mechanism design) tuned precisely to kick out x% of people while setting incentives to self-identify (slackers vs. normal workers vs. workaholics) by generating uncertainty or ambiguity, thus generating some ideal workforce distribution: people who are doubtful are incentivized to just leave right away, people who are extremely competent are moderately pressured, people who are willing to grind themselves into dust for amazon do "okay" for as long as they are useful, and "normals" are retained probabilistically to the degree that their contributions are needed that quarter. Much has been written about how these models don't reflect human behavior. Sadly, as a normative forcing function, they work out just fine. edit: in the short run, until such articles as op appear