I sort of agree that if I wouldn't re-hire a person for a role then they should be let go: after all once someone has been trained in and is gelled with the tea

I sort of agree that if I wouldn't re-hire a person for a role then they should be let go: after all once someone has been trained in and is gelled with the team, they should be better at the job than anyone coming in from outside. In that sense this is just a dramatic reframing of standard performance evaluation wisdom. But I also find it a very strange thing to say, because the biggest challenge of hiring new people is that you really never know how they'll perform until you actually start working with them. In my case I've always hired for smaller organizations with a smaller pool of applicants, so I prefer to just roughly gauge for Spolsky's test of "smart and gets things done" as well as I can and then turn them loose to see what they can do. In large, well-known companies with huge hiring funnels false negatives matter much less and they are incentivized to screen very aggressively on whatever axes they can. However in either case, the existing employee has a known performance profile, and the new hire is unknown regardless of how good they look on paper. So I just find it an obtuse apples-to-oranges comparison to frame performance reviews this way.