Roadkill, on 11 July 2015 - 05:23 PM, said:
Incorrect, both statistically and as implemented.
Elo typically generates a Bell curve, but the width of the curve varies based on the exact implementation. It is not statistically correct to say that "almost all players hover at around roughly the median Elo".
It's also not correct as implemented in MWO, because PGI has said that people have hit the Elo cap that they put in place. Elo in MWO is behaving normally.
As stated by the devs directly, Elo doesn't work. For players who sit in solo queue almost exclusively, their Elo scores tend to huddle together in the middle, regardless of personal skill. The reason is in how Elo works. It's not designed to be able to tell you how a particular player on a team plays... only how a team as a whole plays. Without a system to account for individual performance, it's impossible to determine what influence individual players have on the outcome of a match.
Thus any change made to a player's Elo at the end of the match is, fundementally, a guess. If you were playing with the same group of people consistently, that guess might be better, but playing with a random group of people whose Elo scores are also fundementally a guess means your end-match Elo result might as well be entirely random. In fact, it's easy enough to say that all 24 players in a match will have a fundemetally random result, and extrapolated over many many matches, Elo scores will average out for nearly all players right at the middle of the bell curve regardless of skill.
This is basically what the devs have been finding. They've come out on record to say Elo is broken. And by design, it is. Because it's being used to judge something it's not designed to judge.
0
