Some small updates today on stuff that is in progress.
Matchmaking Skill Gap
As many of you have noticed, we’ve increased the allowable skill gap in most playlists. We did this because while the old settings were tighter (usually only 1 major Rank apart), there were parts of the population that were unable to find matches over unreasonable spans of time. Some players would search all day and not find anything.
This is partly due to us also enabling party restrictions. The restrictions make it less likely to face a full party. Facing a full party isn’t always necessarily “bad”, but the range of possible skill gap that can happen when you face a full party is much larger than when not, so the restriction does give a slight improvement to non-full-party qualify of life.
But it does effectively shard the population, reducing who you can match against, and means we have to, again, widen the allowable skill gap so people can play at all.
The changes let everyone matchmake in a timely manner, but they do allow larger gaps in the players you will see. The gaps are currently like this:
HCS, Team Arena, and Slayer
-
2 full ranks apart (e.g. Gold to Diamond, Champ to bottom Onyx, etc.) - Narrower because more skillful.
FFA, SWAT, Snipers, Doubles, and Breakout -
4 full ranks apart (Platinum to Champ, Bronze to Gold) - Wider because less skillful or pop a bit smaller
By “less skillful” I mean that in the data, given the same skill gap, the underdog wins more often. In HCS, for example, the underdog rarely wins.
When we do this, you can end up with a wide range of skill on your team. If you are the best player in the match, the balancer may put worse players are your team to make sure each side has a 50/50 chance of winning, making you “carry” a bit more. We understand this can be undesirable and cause backaches.
Our current compromise errs on the side of making sure everyone can play.
But, to improve this situation, we are working on a change that will allow us to potentially get back to where we were before by increasing wait times for higher-skilled players. They would wait longer than they used to but would get into closer matches, and not have to pop into lower-skilled matches as often. I don’t have an ETA on this, but it’s coming soon.
With the change, we could probably cut the skill gaps in half, and double high-skilled wait times in Focused. I’ll keep you updated.
Party Matchmaking
Even with the party restrictions, party vs. party matchmaking still has a larger range in possible skill gaps that can happen than we would prefer. It’s even larger now that we’ve reduced the availability of non-full parties, and made full parties wait longer so they end up with worse matchups overall.
To alleviate this, we are wrapping up work on a feature that lets us control the maximum allowed skill gap between teams given the best possible team balance.
Today, we can only control the average skill gap between players or parties, but this gives us no guarantee that the team balancer can find a good gap between two teams once the match forms.
This will result in matchups that are significantly closer than today when parties are involved and very few “blowouts” This is coming soon, but no ETA yet.
In addition, for the longer term, we are updating our skill system to account naturally for party-skill benefits. This would allow us to remove the hard-party restriction, and instead use a soft one that improves matchmaking quality and wait-times for everyone, without imbalanced matches.
Sandbox Tuning
From ske7ch’s post, you’ve heard about the upcoming sandbox update. I’m not directly involved on the sandbox or even the settings side. My involvement is to advise a little on the playlist management side, and keep people informed about population dynamics.
I’m investigating if the new sandbox suggest changes in how the playlists are divided up, for the better of the population.
In one world, the new sandbox tunings could allow us to find settings that both the Team Arena and HCS audiences and Pro players love, so we can merge the lists again. Merging the lists could double the matchmaking population, resulting in immediately better matches for everyone across the board.
In another world, the tunings only result in different settings in Team Arena, and we inject more variety overall in that list, making is significantly different than HCS. Creating a clearer distinction between the two lists would also be good if it reduces the overlap and interaction between those two lists, and fits the audiences better.
We aren’t going to just jam new settings into a list like HCS though. We’ll try it out in a list more like Team Arena, and see how well they are received. This will inform both additional tuning, and playlist management.
I don’t have any more info on my side of this at the moment, but stay tuned. I’ll try and answer questions here as well.