[wongarsu]

Assuming perfect information, Jim's skill being transferable, and Jim's performance eval being objective, you'd expect that the company would profit from transferring Jim and other top performers to their profitable products, and cutting the worst employees from those projects (after all, even a department making profit is likely to have some employees on the low end of the performance bell curve).

Of course that isn't as easy because of morale, team cohesion, performance evals rarely being comparable across teams, and people being not as fungible as the above suggests. Not to mention all the work this takes, in a time when you probably have other worries. So maybe it's not "really weird", just "not immediately obvious"