Its way past midnight. Six night warriors are up. 0 active games, they are all in the 3v3 queue. The 90 seconds come and go, and again, and soon the six warriors were five.
And, of course, no one even thought about joining the 2v2 queue.
If there are no active games, can we just get a game with the best possible balance, even if it's not great?
There is already a system that makes it more easy to start a game, the longer people are waiting in line. Its some threshold based on standard deviation of rating that needs to be crossed. That threshold becomes lower the longer people are in queue. This can be adjusted, but will come at the cost of game quality. Which is a delicate trade-off.
Having the match being made immediately without waiting to see if a better match will be possible in the next few minutes, surely isnt ideal either. So the solution is somewhere in trying to strike that balance.
@stormlantern Very true for all other cases but at 4 in the morning that's as good as it's going to get. Totally an edge case ofc.
True, good chance though that some people in that queue would rather have no game then an unbalanced one. Quite a few players stop queuing for the 2v2 queue because of too many games with a lot of rating spread between players.
I'm not sure whether the current matchmaker parameters are set optimally. I believe BlackYps fine-tunes them on occasion and I personally think the current balance is set pretty well overall. However, it won't give the best results at all times of course.
It's ye olde bad game vs no game discussion.
I don't see room for improvement through matchmaker settings
Does the time-in-queue adjustment for rating threshold get adjusted by how many people are in queue/how many people are on faf? Wouldn't want to wait the same amount of time to get a larger rating threshold in peak hours vs dead hours.
I dont think there is. But "people in queue" doesnt seem a good metric for this. Since you can have 10 active 3v3 games with only 2 people waiting in queue. "Total players online" might be a bit better for this, but still doenst really seem like a necessary addition to me tbh. It doesnt seem obvious to me that during quieter hours you want to settle for less balanced games sooner. But Im interested in what others think
@stormlantern Better metric would take in account both players in TMM queue and TMM games. Total online isn’t a very good metric because people might just be zombies with loaded clients or generally not be interested in TMM.
Agree with what SainseRow said. All players currently searching and in a tmm game is a good variable i think. Maybe that even makes it possible for high rank players to find games?!
What if everyone decide the maximum balance diff they want to be matched on. A simple slider.
Personally I don't care that much if our team gets crushed, it's a learning experience once in a while. Also sometimes it is kinda fun to dominate. Better than no game.
Others might find it frustrating and boring, and would rather go to bed. I respect that.
@sainserow said in Higher tolerance when matching few people:
@stormlantern Better metric would take in account both players in TMM queue and TMM games. Total online isn’t a very good metric because people might just be zombies with loaded clients or generally not be interested in TMM.
What do you think about this? Would this at all be possible? (Matches being made earlier when fewer people are in TMM queue + in game of that matchmaker.)
I'm not sure if it is needed or at all useful, but before I discuss it with the matchmaker team, Id like to know whether it is an option at all.
It is possible, I just don't think it is a good solution. We could increase how fast the time bonus for waiting in queue increases if people feel it takes too long to get to the maximum. But the suggestion to do it based on active people seems to me to increase complexity a lot for marginal benefits.