Fixing the WvW ranking and pairing flaws

#1 - April 5, 2013, 4:13 a.m.
Blizzard Post

Dear ArenaNet,

You have mixed two essential principles when you designed the WvW rankings: ranking and pairing. These two things are intrinsically different, yet the pairings should be loosely linked to the rankings, not strongly as it is now.

In short, Glicko 2 is a ranking system, but not a pairing one. So please, don’t use Glicko 2 to make your pairings and allow unbalanded pairings (and get rid of that classification in tiers).

It’s a fact not really well known over here, but Glicko 2 works better when the same opponents don’t fight every week. It is sane to have the lower ranked guys sometimes play against middle-ranked guys as a kind of “poll”, just to check that the lower guy has not improved his skills.

Therefore please use Glicko 2 for ranking, but find another system for pairing. For instance a fixed rotation with the calendar known in advance, or a dynamic match-making where you put half random so that a server competes with other servers in a 6-8 servers sliding window.

Glicko 2 would still work properly. It would even work better! That system ensures that you get/lose points according to your rank, so you might still move up or down while respectively losing or winning.

But what about the fun caused by such an imbalance?
Well, I don’t know for you, but zerging is fun. One hour. Top. So why zerg if you stomp? This system would ensure that zerging is really useful two or three servers have the same ranking. If we take a planned calendar, it wouldn’t happen often. If we take a sliding window approach, it happens more. On the other hand, what about the server being steamrolled? Well it has been seen before, so I’ll write it, that server will gain some tactical skills that zergs don’t have. They will adapt and they will become more commando-like and still get steamrolled, but they may get more points.

Also, for this to work correctly, add points for defense ArenaNet. Really, do it.

#4 - April 5, 2013, 8:22 a.m.
Blizzard Post

It’s a fact not really well known over here, but Glicko 2 works better when the same opponents don’t fight every week. It is sane to have the lower ranked guys sometimes play against middle-ranked guys as a kind of “poll”, just to check that the lower guy has not improved his skills.

We are aware of this fact and we are trying to decide what we actually want to do about it. It’s more complicated than just that though as there are aspects of scoring that are not reflective of the quality of individual servers as well. Until we address the underlying scoring issues, it’s just a temporary solution though. To that end we are actively designing things that we hope will, if not solve, at least ameliorate the scoring issues.

#7 - April 5, 2013, 8:33 a.m.
Blizzard Post

It’s a fact not really well known over here, but Glicko 2 works better when the same opponents don’t fight every week. It is sane to have the lower ranked guys sometimes play against middle-ranked guys as a kind of “poll”, just to check that the lower guy has not improved his skills.

While that makes sense in theory, it would result in matches that are total blowouts and where most of the player base drops out early in the match.

Which would not be fun and not really give you valid information. Also, every blown match results in a few players quitting WvW altogether.

That’s our current assessment as well, which is why it’s going to take more work to decide what we really want to do here. Speaking for myself, I think we need to change some of the mechanics of scoring to more accurately reflect what we want people doing in the game, which would hopefully get us tighter matches. There are any number of possible ways to deal with this however, which makes it a paradox of choice to a certain degree. Rest assured we have top men working on it! And me. :P

#10 - April 5, 2013, 8:53 a.m.
Blizzard Post

What we would really like is a rough ETA. We’ve been told you were looking in to it before, and yet some of us have been in the same match-up for months. If it’s going to be a while for the solution fine, but do something temporary soon please

EDIT: Holy quaggan, I just read in the other thread about this that you ARE doing something temporary, and the word “today” was even used! You sir, are a rockstar!

I’d love to give you an ETA, but there isn’t one. It’s one of those things where we have to get some solutions, design through the problems, scrap all those, start again, get better solutions, then implement them. Depending on the nature of those solutions, they could be quickly implemented or they could require some time. The promise I will make is that as we get a clearer understanding of the time frame I’ll make sure everyone knows. Specifics are always going to be hard to come by, but I can at least give you all that much.