Further, stack ranking mostly ignores the existence of human psychology. As soon as metric is chosen to be a benchmark for the measurement of performance, it almost immediately because useless as a benchmark for performance, because people are people. People will find out how the game works, and they will play the game. Employees will sabotage other's work to make themselves look relatively better, managers will hire whipping boys for the specific purpose of giving a low rating so they can take the sword. Sure this happens with any performance rating system, but the worse behaviors are accentuated with fundamentally broken rating systems.
How should employee evaluation be done then? It's straight forward:
* Set clear expectations.
* Get peer feedback
* Manager decides how the employee performed in comparison to the expectations that were clearly communicated to the employee based on the manager's personal knowledge of their reports work, and from that of peer feedback.
No zero sum hunger game bullshit required.