How Microsoft's ruthless employee evaluation system annihilated team collaboration.

I'm very skeptical of fairly limited experiences being used to make universal pronouncements.

I'm sure this was the experience for many individuals and teams.  I know for certain it was pretty normal and not worried about for others.  I knew a lot of MS employees in that era, though I worked at a different giant tech firm with vaguely-similar procedures.  I was senior enough, though an IC rather than a manager, to have a fair bit of input into evaluations of my team and division, and I  saw firsthand the implementation and effects of this theory.  It was dysfunctional and harmful for some teams, but more often (in my experience) somewhat effective at actually improving the makeup of teams.

There is a brutal and unpleasant truth in hiring and employment decisions, which is that there is a wide variance in contribution among similar-seeming employees.  Much of this is illegible and extremely hard to objectively measure individually, but it's visible to coworkers and line managers.  For a long-term improvement of a very large organization, it's a reasonable theory to replace the least-productive employees with a new hire (random sample, likely to be better than bottom).  Brutal and unpleasant, but not wrong.

I don't think the policy itself was particularly effective at motivation, but it wasn't terribly harmful either.  The top and middle very much understood the theory, and it was just part of life. 

Details, of course, matter.  A fixed percentage imposed on a small group is going to be wrong most of the time.  I've heard MS, at times, did exactly that - 10% is way too high to be immediately fired (unless it's 10% to get special attention, with the reasonable expectation that more than half will improve and stay).  

Seen this wherever there is rank and yank. People write cool but obfuscated code. Never update documents and. Ignore requests for help from colleagues.