Guild Wars 2

So, we have new WvW match-up system. Which has a randomization factor to shake things up. Which, based on what’s happened in WvW so far, was WAY TOO BIG:

For example, jsut 24 hours (or less) into the matches:

RIng of Fire 83,000, the other two servers combined: 9,500
Dragonbrand 48,000, the other two servers combined: 23,000

And there are more lopsided matches. I just picked a couple of them to illustrate the point.

The consequences of these brutally non-competitive matches has been to drive a lot of the WvW people out of WvW on the overwhelmed servers. After all, when you can’t hold the spawn towers your own borderlands… Why bother? Which, is currently what’s happening on many servers as people get tired of having NO CHANCE.

I know my otherwise very-heavy WvW guild (on Stormbluff Island) has already packed it in. We’re not a high-tier server (#13). We can’t cover like a high-tier server. And we’re so completely outnumbered that we can’t possibly win. Dragonbrand was fielding single-guild zergs bigger than our entire SBI zerg.

So we have come to the obvious conclusion that we should not waste our time and resources on a completely hopeless matchup. Which gets to the flip-side of the coin — Dragonbrand players get punished by our non-contesting.

Let’s face it, if the people losing don’t play, the winners may be racking up the points, but they don’t flip the towers, keeps & camps as they’re not changing hands… Which means no event XP for them, no karma for them, no WvW XP for them, no badges for them, no fighting for them. Nothing. Just sitting in WvW with their thumbs in their rear-ends hoping that next week it’ll be better…

I guess ANet really should have thought it through. Put a limiter on how far up or down a server could move and a ranking differential test as well. Maybe three spaces up or down with no more than a 6-slot difference between the top and bottom server. That way you’ll limit the crazy match-ups and give people at least a chance to be ‘respectable.’

But pitting the #13 server (SBI) against the #5 server (Dragonbrand) or the #25 server (Arborstone) against the #12(Gandara)… Or some of the other crazy matches…

Yeah, that’s not going so well. After all, there was a REASON for those huge differences in ranking…

There is a limiter! A tier 8 server can’t play a tier one. Someone on my server (FC) posted a link on how it works and that there are percentages for each server. There were just a few unlucky matches! I for one can say the IoJ-FC-DR is a decent match up!

Herschie.6380:

There is a limiter! A tier 8 server can’t play a tier one. Someone on my server (FC) posted a link on how it works and that there are percentages for each server. There were just a few unlucky matches! I for one can say the IoJ-FC-DR is a decent match up!

The chances a tier 8 server can face a tier 1 server will rise every single week no matter what anyone on those servers does. It will happen if the math stays the same.

Seems very few people here actually read the dev post about the new matchmaking system and understood the maths behind it. There is an adjustable limit which prevents servers with significantly different ratings from playing each other, so you won’t be seeing T8 servers or even T3 servers playing T1 servers anytime soon.

oxide.8324:

Seems very few people here actually read the dev post about the new matchmaking system and understood the maths behind it. There is an adjustable limit which prevents servers with significantly different ratings from playing each other, so you won’t be seeing T8 servers or even T3 servers playing T1 servers anytime soon.

What you aren’t understanding is that the ratings aren’t going to be “significantly different” for very long. Higher tier servers will give up points to any lower tier server they are matched with. All server ratings will move closer to one another no matter how the matches play out.

Dont limit it, my server is severely outmatched this week and we are having a blast! Makes things so much more interesting.

Maladon.5760:

oxide.8324:

Seems very few people here actually read the dev post about the new matchmaking system and understood the maths behind it. There is an adjustable limit which prevents servers with significantly different ratings from playing each other, so you won’t be seeing T8 servers or even T3 servers playing T1 servers anytime soon.

What you aren’t understanding is that the ratings aren’t going to be “significantly different” for very long. Higher tier servers will give up points to any lower tier server they are matched with. All server ratings will move closer to one another no matter how the matches play out.

I didn’t speculate over what would happen in the longer term, only the short term.

Either way, lower tier servers won’t come remotely close to beating higher tier servers when they do get matched up, and the higher tier servers won’t give up many points. Even if it does become a problem Anet can easily adjust the base variation to make lower tier servers get matched up less often with higher tier servers to fix it.

This empty debate bores me.

Someone please prove or disprove (mathematically using the parameters in Glicko 2) the following statement:

Let the current time t =0. Let X={x1, x2, .. , xn} be the set of realms in the Glicko rating system. As t becomes large, the average rating difference between any two xk in X will be lower under the current matchup system compared to the previous matchup system.

The argument that the new system will cause server ratings to move closer together seems to be based on the idea that higher ranked servers generally bleed points to the lower ranked server in their matchup.

For this to be the case, the expectation would have to be for lower ranked servers to do better than Glicko expects them to. Whether or not this is true is dependent on how the scores actually correlate with the rankings, so the parameters in Glicko 2 alone aren’t enough to prove anything. You would need to analyse past results.

Murderous Clown.9723:

The argument that the new system will cause server ratings to move closer together seems to be based on the idea that higher ranked servers generally bleed points to the lower ranked server in their matchup.

For this to be the case, the expectation would have to be for lower ranked servers to do better than Glicko expects them to. Whether or not this is true is dependent on how the scores actually correlate with the rankings, so the parameters in Glicko 2 alone alone aren’t enough to prove anything. You would need to analyse past results.

Very good! What seems to be embedded in this problem is there exists some difference in the rating distribution between the previous and current systems—an artifact of the restricted nature of the previous matchmaking system. It follows (without proof) that a wider variety of matches will move the ratings to this true^ rating distribution. We can, of course, include such assumptions in a proof like this; these assumptions are actually most likely necessary.The question becomes: will this movement in ratings lead to better matches, on average?

^It isn’t actually the true distribution because we would need a very large amount of time and an unlimited matchmaking system to get that information into Glicko. Not to mention changes that occur to each realm over time.

Murderous Clown.9723:

The argument that the new system will cause server ratings to move closer together seems to be based on the idea that higher ranked servers generally bleed points to the lower ranked server in their matchup.

For this to be the case, the expectation would have to be for lower ranked servers to do better than Glicko expects them to…

Lower ranked servers gain rating points based simply on being matched with higher ranked servers because it isn’t supposed to happen randomly. If a high ranked server and a low ranked server are matched up the glicko system will attempt to resolve the ratings delta because matchups are supposed to represent parity in the system. Only after the delta is resolved will the ratings move towards actually being accurate. Unfortunately this will never happen in randomly seeded matchups every week.

Maladon.5760:

Unfortunately this will never happen in randomly seeded matchups every week.

Partially random*

Your comment has made me reread how glicko 2 works as well as how Anet was/is using it and think about it for a while.

Maladon.5760:

Lower ranked servers gain rating points based simply on being matched with higher ranked servers because it isn’t supposed to happen randomly.

It isn’t? My understanding was this not happening was precisely the problem with the old system. Even if this point were true, I still don’t see how it causes lower ranked servers to gain rating points. Maybe I’ve misinterpreted you so some clarification on what you mean here would be welcome.

Only after the delta is resolved will the ratings move towards actually being accurate. Unfortunately this will never happen in randomly seeded matchups every week.

Why not?

I must have spent at least the past hour thinking about this last sentence. Figuring out what this means about the effectiveness of the ranking system at varying disparities of rating, considering whether the glicko formulae and/or Anet’s sine transformation would be responsible.
To put it simply, the only way I can see this to be true is if there is a systematic error in how the expected score (the score that results in no ratings change, denoted as E in the glicko page) is calculated. Thing is, I’m not convinced it’s possible for one to exist, though I need to think about that part further.

If you have justification for your statement, please provide it to save me from blowing my brains out.

should of randomized tier 1+2, then 3+4, then 5+6 and then 7+8. At least it would have been closer population wise.

Im on gunnars and Im loving the week with piken. Just set the limit so that we cant face german servers (except for dzag, they’re ok)

The new system is better for accurately assessing a server’s rating.

Getting 100% accurate server ratings won’t ever do anything to balance a match. In fact, the more accurate the ratings, the more obvious it will become how the scoring system creates snowball effects which result in huge score differences even with nearly insignificant population differences.

Either Arenanet mans up, admits the colossal structural problem with its scoring system, and starts working on changing it, or the problem of lopsided matchups will stay forever.

I’m still amazed at the refusal of WvW devs to understand such a basic concept as “people like balanced matchups and find blowouts boring”.

Well let’s compare the difference between #1 and #3 in each tier.

http://mos.millenium.org/na

tier 1: 2
tier 2: 4
tier 3: 8
tier 4: 3
tier 5: 3
tier 6: 5
tier 7: 5
tier 8: 7

The average is 37/8=4.625 ranks. So on average you are facing someone around your rank. Those 8 rank differences does happen but they are rare. Wait till next week and you will most likely face someone closer.

The server ratings will become closer with the new system. The math says that, in the long run, the ratings should even out. However, as close the ratings may be, it doesn’t change the fact that the WvW populations are vastly different. All the servers could end up being within a small range in ratings, but will still be vastly different in game.

This is because of the effect of the OLD system which encouraged server stacking and wvw population concentrations. This made it so that the difference between consecutive tiers could be huge. Thus the ratings shook the servers down into tiers with similar populations with very little movement in rankings.

The new system is trying to address a symptom of the old system, stale matchups, without fixing the core problem which created the symptom in the first place: the uneven distribution of WvW players. The system should be first fixed in a way as to direct player behavior towards the ultimate goal of evening out the WvW populations. The dynamic matchups will happen as a natural result if the majority of the coverage/populations are similar.

Fixing that problem is a much larger task. And keep in mind that the players (despite very vocal denials) clearly supported uneven populations under the old system through their actions.

dodgycookies.4562:

The server ratings will become closer with the new system. The math says that, in the long run, the ratings should even out. However, as close the ratings may be, it doesn’t change the fact that the WvW populations are vastly different. All the servers could end up being within a small range in ratings, but will still be vastly different in game.

This is because of the effect of the OLD system which encouraged server stacking and wvw population concentrations. This made it so that the difference between consecutive tiers could be huge. Thus the ratings shook the servers down into tiers with similar populations with very little movement in rankings.

The new system is trying to address a symptom of the old system, stale matchups, without fixing the core problem which created the symptom in the first place: the uneven distribution of WvW players. The system should be first fixed in a way as to direct player behavior towards the ultimate goal of evening out the WvW populations. The dynamic matchups will happen as a natural result if the majority of the coverage/populations are similar.

Fixing that problem is a much larger task. And keep in mind that the players (despite very vocal denials) clearly supported uneven populations under the old system through their actions.

I am tired of hearing about uneven population, people just need to stop. There is no magical fantasy land where people log in and log out at the same time and ArenaNet won’t and shouldn’t force people when/where to log on. We’re not victims of an unfair system, we’re all band wagoners. Face it, if all the servers were even population across all timezones, people would transfer off the 1 server that lost by 1 point to the server that won by 1 point.

At least in this new system, the top 3 tier 1 servers can stop being in a non-stop recruitment race that just leads to people havin sissy fits and abandoning their server.

Sororita.3465:

should of randomized tier 1+2, then 3+4, then 5+6 and then 7+8. At least it would have been closer population wise.

a simple, easy, logical solution. there you go. end of discussion. Sororita wins all. everyone go away cuz sororita just won. gg. oh wait…did I use the L word? logic…hmmm. sorry Sororita…this will never happen. its too perfect. hashtag sadface

dodgycookies.4562:

The math says that, in the long run, the ratings should even out.

I’m must be missing something here because people keep making claims like this as if they’re obvious and accepted facts.

Please can someone explain (restating it in other words doesn’t count) this idea that the rating system is pushing servers towards in even rating. I accept the current ratings are artificially extreme because of stagnant matchups but I don’t think this is what you’re getting at.

styx.7294:

I am tired of hearing about uneven population, people just need to stop. There is no magical fantasy land where people log in and log out at the same time and ArenaNet won’t and shouldn’t force people when/where to log on. We’re not victims of an unfair system, we’re all band wagoners. Face it, if all the servers were even population across all timezones, people would transfer off the 1 server that lost by 1 point to the server that won by 1 point.

At least in this new system, the top 3 tier 1 servers can stop being in a non-stop recruitment race that just leads to people havin sissy fits and abandoning their server.

The large WvW guilds don’t transfer to win, they transfer to have even match ups and good fights. If they transferred to win, large guilds would actually move down the tiers where they could blow out opponents.

What my point was that the old system encouraged people to congregate into tiers of similar populations to find balance, which naturally resulted in stale matchups and bloated ratings. The new system fixes the stale matchups and ratings at the expense of balanced matchups, which in the opinion of many was a bad tradeoff.

dodgycookies.4562:

styx.7294:

I am tired of hearing about uneven population, people just need to stop. There is no magical fantasy land where people log in and log out at the same time and ArenaNet won’t and shouldn’t force people when/where to log on. We’re not victims of an unfair system, we’re all band wagoners. Face it, if all the servers were even population across all timezones, people would transfer off the 1 server that lost by 1 point to the server that won by 1 point.

At least in this new system, the top 3 tier 1 servers can stop being in a non-stop recruitment race that just leads to people havin sissy fits and abandoning their server.

The large WvW guilds don’t transfer to win, they transfer to have even match ups and good fights. If they transferred to win, large guilds would actually move down the tiers where they could blow out opponents.

What my point was that the old system encouraged people to congregate into tiers of similar populations to find balance, which naturally resulted in stale matchups and bloated ratings. The new system fixes the stale matchups and ratings at the expense of balanced matchups, which in the opinion of many was a bad tradeoff.

Maybe they don’t transfer to win but most of them are only transferrin when they’re losin.

They said they’d lower the deviation if they thought it necessary and we’ll see, probably they might do that.