Agent 0 Fortune, on 17 October 2013 - 08:52 AM, said:
I'm going to try to expalain this a little better, using the examples you posted. If you have 1000 players on and 100 in the appropriate Elo range, then the matchmaker can grab 24 for of those that are all adjacent making a instant match were every player is the same Elo rating.
Whereas if there is a group involved, the players on teh same group may not (are probably not) in the same Elo bracket, and the matchmaker now has to find another group with similar composition, which it cannot, so it grabs pugs from all over the Elo bracket to match Elo scores.
This is why you see so many matches that end 12-0, 12-1, 12-2; and why you see such disparity in tonnage. Pre-made groups force the matchmaker to broaden it search criteria out of the gate, because premade groups can bring in varied Elo pilots and can force a weight class disparity.
To simplify this evenfurhter. If you have ever bought a variety back of anything, you know there is always some things in the variety pack that you don't want, and if the option were available to you it would always be more beneficial be able to purchase the variety items individually and pick just the items you want. this is the same dilemma the matchmaker faces every match there is a group (which is to say almost ever match).
As for where I came up with this conclustion, it is simple, I read the dev posts, where Matt Craig specifically states, Premade groups are harder to match, and all the adjustments to the matchmaker algorithm to make closer matches were undone.
There are a number strawmen arguements being set up here, so I will try to reiterate the pro position clearly.
PUGs want a seperate queue to:
1. speed up the matchmaker
2. get closer Elo matching
3. get closer weight class matching
4. more variety in mechs and tactics
We don't believe that pre-made teams are inherently evil (I do play premade teams as well).
We don't believe that pre-made teams are responsible for lopsided matches, this is an Elo balance issue putting high and low Elo player s together in the same match and failing to match weight class.
Again, I completely disagree that it will speed up anything. Better matches? Probably not, but perhaps. I think you drastically underestimate what halving the size of a queue will do. Additionally, when he says "harder to match," I suspect he means harder to find a good match (not takes a lot longer to match (at least now that they told the matchmaker to stop being picky)).
Because of the varying compositions of premades, you will indeed get massive tonnage imbalances. But for Elo it just averages their ranks and looks for something around that.
Based on what I've seen, it really doesn't care about tonnage very much, particularly when premades are involved, and Elo is a relatively easy thing to match. I'm not saying it results in particularly balanced fights, but it certainly results in faster matches.
You're completely ignoring the huge effect a smaller queue would have, particularly at odd hours when not a lot of people are on. Halving the player pool would have far greater an impact than the removal of whatever premade calculations are being done in the matchmaker.
Add that to the fact that it would basically kill matchmaking for premades. Without tonnage limits, it would be impossible to build fair matches during off-peak times. And how would it deal with odd group numbers? It would take forever to set anything up, if it even had enough groups of the proper size to make a match at all.
I just see absolutely no logic behind the assertion that giving the matchmaker fewer options to pick from will result in it speeding up. And again, speed is all I care about at this point. I'm often waiting 3 or 4 minutes to find a match these days, and it's killing me.