Hey everyone,
I wanted to address the idea of moving world linking to monthly instead of every 2 months, since it is being brought up more and more frequently.
Yea it has definitely been brought up a lot.
The team isn’t opposed to this idea; we actually think it would be beneficial to move to monthly because it would allow us to iterate faster on how we are calculating which worlds should be linked. However, the main reason for not doing this right now is the matchmaking algorithm, Glicko. Each time we shuffle worlds via world linking it takes about 4 weeks’ worth of matches before Glicko begins to reliably match make those new worlds into balanced matches. If we did world linking monthly, Glicko would not be able to create balanced matchups.
I agree, that glicko is a terrible system when you rapidly change matchup since there is no easy way to combine server scores to create a new score, and in addition it does not account for rapid player movement between servers (which happens regardless of whether you link or not, but…)
Our next priority poll is going to be asking if players would rather have us work on adding rewards to skirmishes (and possibly other feedback items being collected from this thread) or replace Glicko matchmaking with a 1-up 1-down system (wherein the winner moves up a tier and the loser moves down a tier.) The 1-up 1-down system should work better with monthly linkings than Glicko, so we are most likely going to hold off on 1 month linkings until that system is in.
I think that reworking how you make matchups is something that is definitely a good priority, since the glicko system essentially means that right now, we have 4 weeks of matchups don’t make sense. For servers who are on the bad end of this deal it means they can’t necessarily play wvw very well because they are continuously stomped until the winning server can finally balance out of that matchup.
This suggestion system sounds like it would move faster, and give obvious and visible changes to who is/will be ranked against who which I really think could work out well. The actual impact of such a change is really hard to determine though.
Another possibility we could pursue is 1 month linkings, but use the Glicko offset system to guarantee the matches. Alternatively, we could manually change Glicko ratings to what we believe they should be for each world. Either option would force worlds to start out closer to being in the correct tier and thus give better matches faster. These options are contentious, so even if everyone on the forums seemed to like this idea it would be something we would poll on.
These both seem like patch-work solutions to the problem of glicko rating. I don’t think that the glicko adjustments will work very well especially on a faster pace.
Match making is incredibly hard, but I think that currently it’s really problematic because so many matches can end up being really bad for a lot of players.
Thanks for clarifying your teams thoughts on this matter, and I really appreciate how much effort and thought has gone into a lot of these changes over the past month.
