The argument of "I don't like the matches the MM puts me in, why should I have to fight against bad odds because MM" and the whole "lets not handhold bads, put on your bigboy pants" seem an odd match to me.
While not perfect, I certainly prefer the current system to the old solo-pug vs 8-mans roflstomps... which were way more common than now. I only solo PUG though so can't really speak to the problems ELO causes for group or high ELO players.
A MM should focus on 'making enjoyable matches' as opposed to 'being a statistically sound system'. Sometimes simpler is better.
3/3/3/3 is a concern (with the overweight on heavies), as is NPE, as is 'good matches with variety' overall.
If I was to design one, I would probably have it build a match by:
- Match 8 of the 12 players as 'like' players, either 'up to 50% win rate/first-100-matches' or 'above 50% win rate' (could be player overall, but would be more 'customized' if it was 'in that mech').
- Toss in 4 more players random, but favoring longest wait time.
- Build a 2/3/3/4 match with the 2 from the least-number queue and 4 from the most.
Given that basis, I think it would be fine to allow 2-man teams into the 'solo pug' queue, so long as they were also balanced (equal # groups on each side).
That allows friends to play together, mentors to play with trainees, a variety of skill matches without excessive steamrolls, and allows everyone to start new mechs without getting nuked by their 'rating'. It should also allow for faster match search.
There is a group-queue issue of 'big groups vs. a mashup of small groups' which I am not experienced in, but it seems a similar 'like-sized groups' system should apply.
Edited by MadBadger, 29 May 2015 - 04:21 PM.