G# and particularly Jack Welch was the one who first introduced this system en masse. Even he, IIRC, did say that, this Forced Ranking (as it was called), system should be more of a short term thing where in you clean the system of the non-performers (how can someone identify justly and accurately is a question) and do not enforce this once it has run its course (3-5 years). Everyone attributed the success of the G#'s northward share prices during his reign to this "performance culture". If there was any truth in that approach, only it has not sustained after Immelt took over.
Even I>M follows this so called "bell curve" based assessment. Every manager should strictly adhere to the skew in his department, down to the smallest of teams.
With this being essentially a HR question for decades to solve, with so many people graduating with MBA HRs and Ph.Ds, I am surprised at the lack of progress or do not know of any alternative system being implemented in a big company successfully.
The normalization which is done at these companies make a mockery of sound statistical principles. A bell curve is forced on even the smallest of teams rather than letting a pattern emerge with very large populations where the distribution is in a shape of a bell curve.
e.g., In a particularly bad year, in a sales team of 10 people the guy/ girl who sold a 50 mill USD as opposed to a top ranker who has done a 150 million is rated bad and equated with some body who wrote 1000 lines of code and his project got cancelled (Note: Project could have got cancelled for any reason, not necessarily and least likely because of the code he wrote). A person who wrote 1000 loc and showed off (read:built visibility with managers who decide his fate on the rating discussions) could easily get a better rating. The whole system goes around a) building perception b) taking credit for success c) pushing responsibility for failure onto someone else. People become paranoid. Nobody shares information with any one else which would help the other guy get a better rating. The organization suffers as a result.
Despite all this and despite Edward Deming's call decades ago to abolish this rating system as one of his 14 principles, I have not seen a successfully implemented system which replaces this Darwinistic one. Does anybody have any examples of better systems or better implementations in organizations?