[ScottBurson]

The biggest problem with stack ranking, as I have read about it being practiced, is that it is implemented on a team-by-team basis rather than across an entire engineering department. The consequence is that differences in strength between teams are not taken into account -- the members of each team are ranked only relative to the other members of that team. This leads to people trying to game the system in various ways to make sure they're not seen as the weakest on their team.

Ranking all engineers in the department together would be more difficult. Managers have a hard enough time getting a read on the strength of their own people; getting them to agree on how their people rank against the other managers' people would take quite a bit of work, I imagine. But it would give a much better answer to the question of whom to let go first, and it wouldn't provide as much incentive to game the system.

And I agree with you that the ranking should explicitly emphasize how well each employee works with their teammates.