I'm glad I'm not the only one who thinks the MM needs work.
The real danger of the current system is that match results can be taken out of your hands, regardless of how good you are. As Skyfaller put it,
Skyfaller, on 11 January 2014 - 03:54 PM, said:
You can be the world's best MWO player and if your team scatters all over at map start and dies noobrushing the entire enemy team it is NOT your lack skill that lost the map yet that is how it gets counted.
This is dangerous for the health of the game. If results are largely random, then there is reduced incentive to improve by continuing to play the game.
This is partially caused by the Elo system. For those unfamiliar with where Elo came from, check this out:
http://en.wikipedia....o_rating_system
Elo was created to rank chess players and it's very effective in doing so, because the results of a match are determined solely by one individual, to whom the Elo score is attached.
The problem comes when the competitors are different with every match. MWO addresses this by averaging,
http://mwomercs.com/...79-matchmaking/
which is a (far too) crude way of handling the random-team problem. Think about how this impacts the best and worst players on a team. Assuming the competitive teams have equal average Elos, a win will bring up the Elos the same amount for all players on the team. So, the noobrusher gets a boost to his Elo that he probably didn't deserve, while the veteran ends up doing the dirty work. Conversely, a team noobrush followed by a loss will lower everyone's Elo, even if the veteran plays a smart game.
This becomes important in the *next* match. In the first case, the noob has an over-inflated Elo, so his team's average is higher than it should be. In the second, the veteran has a too-low Elo, so his team's average is artificially low. You could argue that the errors will average out; for every underrated player on a team there will be an overrated player to compensate.
I won't get into the math, but hard core statisticians would tell you that this perfectly balanced combination is very rare with only 12 players. (Send me a PM if you ARE interested in the math.)
The bottom line is an Elo system where an individual is rated based on a group's performance and the group is constantly changing is bound to be inaccurate.
This Elo system is probably better than purely random matchmaking, but when non-skill-related factors are added (e.g. tonnage imbalances), I can imagine that it's not by much.
1
..... 
