Guild Wars 2

Thank you for not resetting. Please keep this up.

Hugs, from your fanbase.

I really wish they would modify the points however. BG is beating us by 100k points and loosing rating. No matter what your rating is you should never go negative if you are beating a team by 100k. It should just be less positive points the higher your rating.

IMO the system should be modified to if you win you go up in points, if you loose you go down. On IoJ we got roflstomped for 8 weeks straight in t2, we had no right to be in that tier but the system kept us in it.

MODIFY the Glicko system please!!!

Also the gulf of points difference between tier 8 and 7 may never be breached.

Reset all tiers to have 150 points difference between them,
Tier 1
1st 2100
2nd 2050
3rd 2000
Tier 2
4th 1950
5th 1900
6th 1850
……
Tier 8
22nd 1000
23rd 950
24th 900

Would have the desired result of a reset without the months of volatility.

i was anxiously anticipating the reset. because something must be done. glicko is obviously not meant to accommodate 200 rating point gaps.

guess i shouldve been more vocal. all the whiners have ruined the hopes of ebay, bg, and t8.

Thank you, Anet!

But thanks to people bandwagoning the upper tiered servers, instead of increasing the competition in this game. They should just merge/close the low tiered “dead” servers.

They will merge the lower tiers in a year or so from now probably. No way this game can sustain 48 different servers.

Yep. Having taken Anet to task for showing no signs of listening only yesterday, I’m very pleased to be proven wrong in this case. Thanks guys.

insanemaniac.2456:

i was anxiously anticipating the reset. because something must be done. glicko is obviously not meant to accommodate 200 rating point gaps.

guess i shouldve been more vocal. all the whiners have ruined the hopes of ebay, bg, and t8.

People being vocal was why they decided to reset the ratings in the first place… then a lot of the same forum trolls decided to get on and they changed their minds again. I enjoyed the “this isn’t fun” thread pointing at the 100k point differences where the winning server was losing points and players were getting annihilated by the same servers week in and week out. Today’s announcements were a really sad day for me in regards to this game and a huge let down. I just hope that they can manually adjust some of these match ups because obviously BG and t8 need to be fixed. Might be a good idea for the developers to hop onto one of those servers and see for themselves.

Honestly, we didn’t even see how the ratings would work out in 2-3 weeks before speculation and whining shot down the idea. Makes one wonder if they had any “new math” in the first place or we had just called their bluff.

Balking at wvw ratings because of a few vocal people on servers is a huge step back. I don’t know why you’re thanking anet because now there will be a lot of players stuck in the same grind against the same opponents who grossly outnumber them… and we never really got to see how things would work out. What’s to say the proper tier 1 servers will not rapidly move up under the “new math.” Very close minded imho.

They have to review the ranking calculation, this is a fact. Servers that are overpopulated and are blowing the enemy should go up faster I agree this would create more balance for servers with fewer WvW population.

But reseting now doesnt make any sense… now that transfers are not free anymore it is a matter of few weeks to each server get to his right tier.

With a reset, the three T1 servers would have been split up… one in T3, one in T4 and one in T5.

The next week would not have been fun for the other servers in T3, T4 and T5.

VOLKON.1290:

With a reset, the three T1 servers would have been split up… one in T3, one in T4 and one in T5.

The next week would not have been fun for the other servers in T3, T4 and T5.

The thing that’s being pointed out by folk from low-pop, screwed-over servers is: it’d be better to get shafted for one week, mayhap two. HOWEVER, we’re now (sans reset) back to where servers who get population-steamrolled are catching it in the bung for MONTHS. A couple of weeks of a hopeless fight, versus months … which is more logical, and going to keep people invested in Wv3?

Solution? Fine, keep yer current maths – but take a direct hand in cases where it’s an obvious blowout. Seriously, gents – as others have said – come down to Ferg or ET for a week, then SF for a second week. At that point, Devs, you will definately see the problems of our current system. Firsthand.

albotelho.2931:

They have to review the ranking calculation, this is a fact. Servers that are overpopulated and are blowing the enemy should go up faster I agree this would create more balance for servers with fewer WvW population.

This is how their current math works, I never knew why they were thinking about revising it in the first place. in the current system the score differential and previous rating are used to calculate a rating increase.

so if A server is facerolling from t5 due to population increase they will jump up places quicker by having a higher score differential but a lower intial rating. if they do it again in their new matchup the same thing will happen.

Agreed, I am very glad with the new decision. Keep it up guys!

Advent.6193:

VOLKON.1290:

With a reset, the three T1 servers would have been split up… one in T3, one in T4 and one in T5.

The next week would not have been fun for the other servers in T3, T4 and T5.

The thing that’s being pointed out by folk from low-pop, screwed-over servers is: it’d be better to get shafted for one week, mayhap two. HOWEVER, we’re now (sans reset) back to where servers who get population-steamrolled are catching it in the bung for MONTHS. A couple of weeks of a hopeless fight, versus months … which is more logical, and going to keep people invested in Wv3?

Solution? Fine, keep yer current maths – but take a direct hand in cases where it’s an obvious blowout. Seriously, gents – as others have said – come down to Ferg or ET for a week, then SF for a second week. At that point, Devs, you will definately see the problems of our current system. Firsthand.

Do you really think things wouldn’t ultimately settle into pretty much the same grooves they are now? With transfers locked down now by cost things will be much more stable and servers will settle into ranges appropriate for their WvW population and coverage.

I am happy with their decision. Yes, there may be some hurt feelings on the lower tiers, but they are down there for a reason. The glicko system may not be perfect, but at least it is pure. It is Math.

There are a set of rules in place (Glicko). Use those rules and make it work for you as much as possible. Having an arbitrary person swoop in and manually adjust numbers and massage numbers is even worse. The team that it does not favor (because anytime you have someone adjust things there is a loser and winner), would be even more outraged. They would want justification for every single individual point argued for and against.

The math is pure. It may not be perfect, but it is consistent. No one has arbitrarily adjusted numbers during the months since launch. It has all been player driven within a mathematical formula.

We have controlled the numbers. The power is in our hands. Even considering transfers, it was still our decision, not theirs. This is an immense privilege that they gave us. If you are in a tier 8 server, it is because the community at large decided that you should be in a tier 8 server. We all know that the server transfers took the decision from the world’s hands, but with free paid transfers off the table the power will be wrestled away from the community and placed back into the control of the world.

Go have fun in WvWvW and stop worrying about what rank you are in. Even if you have just one camp on one map. Own that camp and defiantly display your flag in the face of your opponents. And when they descend like locusts, stick out your tongue and teleport away if you are worried about the repair bills.

Signed.

Reset was not necessary. All that is necessary is perhaps tweaking the math behind the ratings a bit to make it easier for servers to move up and down.

fishergrip.4082:

Signed.

I agree, Reset was not necessary. All that is necessary is perhaps tweaking the math behind the ratings a bit to make it easier for servers to move up and down.

SniffyCube.6107:

Balking at wvw ratings because of a few vocal people on servers is a huge step back. I don’t know why you’re thanking anet because now there will be a lot of players stuck in the same grind against the same opponents who grossly outnumber them… and we never really got to see how things would work out. What’s to say the proper tier 1 servers will not rapidly move up under the “new math.” Very close minded imho.

If you actually understood how the Glick-2 rating system works you wouldn’t have posted that. Resetting everyone would have caused horrendous oscillations in the rankings that would have caused blowout after blowout for weeks.

And if you had been able to understand what Anet wrote in their announcement you’d notice that they stated that the “new math” (actually just different weighting factors in the Glicko-2 formula) was unable to make any rating improvements in the trial runs they experimented with. ANet got to see how the new system would have worked out and they didn’t like it … neither would you.

Anet wasn’t just caving to forum complaints, and at least in this case they weren’t being close minded at all. They were simply being realistic … and you aren’t.

fishergrip.4082:

Reset was not necessary. All that is necessary is perhaps tweaking the math behind the ratings a bit to make it easier for servers to move up and down.

Well, there are other fundamental flaws with using the Glicko-2 system in the first place (examples below), but yes, adjusting the factors in the formula (the “new math”) for things like how heavy the weighting for prior match scores or how heavy the penalty for score variability would in theory make the rankings more responsive without introducing the ridiculous step function instability that a reset would have caused.

Remaining fundamental flaws:

-1. The Glicko-2 system was never meant to be used for tiered competition. It was intended to be applied to competitions where every participant plays a continuum of opponents in the general vicinity of his own rank. A tiered system locks opponents into a bracket from which they cannot escape if they are evenly matched … even though they may be better or worse than the participants in adjacent tiers.

-2. Anet’s version of the Glicko-2 system assumes that Green versus Blue, Blue versus Red, and Green versus Red are all independent matches. This is patently invalid on several levels. Two servers may, either through agreement or simple match dynamics, both end up attacking the third server a disproportionate amount of the time. Losing servers may suffer loss of morale and play much more weakly as the week goes on. Both are perfectly legitimate occurrences in a three faction match, but neither represents an independent result for two team pairings.

Cactus.2710:

-1. The Glicko-2 system was never meant to be used for tiered competition. It was intended to be applied to competitions where every participant plays a continuum of opponents in the general vicinity of his own rank. A tiered system locks opponents into a bracket from which they cannot escape if they are evenly matched … even though they may be better or worse than the participants in adjacent tiers.

i see this problem cited constantly any time there is discussion about glicko flaws.

and every time i see it, its wrong. we just call our matches tiers. tiering hardly applies at all, as the definition here is vague and mutable. what makes a tier? well the 3 servers in the match of course. we dont define it based on ranking, which is what we actually should do. like 1001-1200 is tier 7. 1201-1400 = tier 6. etc.

because of this, the actual issue of having a low population of possible competitors gets confused with whatever youre saying. matchmaking repeatedly creates matches based on the most even matches, always resulting in little change in opponents, because there are so few opponents to choose from. a server doesnt get stuck with bad matches because of the tiers, it gets stuck because there are 0, 1, 3, or not 3N-1 equal opponents. and the system requires 3N-1 equal opponents for such a low competitor population.

insanemaniac.2456:

Cactus.2710:

-1. The Glicko-2 system was never meant to be used for tiered competition. It was intended to be applied to competitions where every participant plays a continuum of opponents in the general vicinity of his own rank. A tiered system locks opponents into a bracket from which they cannot escape if they are evenly matched … even though they may be better or worse than the participants in adjacent tiers.

i see this problem cited constantly any time there is discussion about glicko flaws.

and every time i see it, its wrong. we just call our matches tiers. tiering hardly applies at all, as the definition here is vague and mutable. what makes a tier? well the 3 servers in the match of course. we dont define it based on ranking, which is what we actually should do. like 1001-1200 is tier 7. 1201-1400 = tier 6. etc.

because of this, the actual issue of having a low population of possible competitors gets confused with whatever youre saying. matchmaking repeatedly creates matches based on the most even matches, always resulting in little change in opponents, because there are so few opponents to choose from. a server doesnt get stuck with bad matches because of the tiers, it gets stuck because there are 0, 1, 3, or not 3N-1 equal opponents. and the system requires 3N-1 equal opponents for such a low competitor population.

It isn’t wrong at all. Picture a Tier 6 matchup where the three servers are equally matched and battle to essentially the same score. Unless one of the servers in Tier 5 loses by a large enough amount to drop down to Tier 6, or one of the serves in Tier 7 wins by a large enough amount to climb up into Tier 6, those three servers in Tier 5 will be playing each other week after week. It’s entirely possible for the servers in Tier 5 to be losing proficiency (population, coverage, key commanders) to the point that they are actually weaker than the servers in Tier 6, but if they do so collectively the Glicko-2 system won’t drop one of them down to Tier 6 unless a blowout happens.

The Glicko-2 system was designed to be used in situations where, for example, the player or team ranked #12 played #‘s 9, 10, 11, 13, 14, and 15 … and where #13 played #’s 10, 11, 12, 14, 15, and 16 … etc That creates a constantly overlapping reference for every team that more or less properly assesses their relative proficiency. You don’t get that when you lock teams into tiers … you lose the reference to teams above and below you unless there is lots of disparity in proficiency (i.e., blowouts happen) or unless something happens beyond your control above or below you. That’s a fundamental flaw any way you look at it.

Cactus.2710:

insanemaniac.2456:

Cactus.2710:

-1. The Glicko-2 system was never meant to be used for tiered competition. It was intended to be applied to competitions where every participant plays a continuum of opponents in the general vicinity of his own rank. A tiered system locks opponents into a bracket from which they cannot escape if they are evenly matched … even though they may be better or worse than the participants in adjacent tiers.

i see this problem cited constantly any time there is discussion about glicko flaws.

and every time i see it, its wrong. we just call our matches tiers. tiering hardly applies at all, as the definition here is vague and mutable. what makes a tier? well the 3 servers in the match of course. we dont define it based on ranking, which is what we actually should do. like 1001-1200 is tier 7. 1201-1400 = tier 6. etc.

because of this, the actual issue of having a low population of possible competitors gets confused with whatever youre saying. matchmaking repeatedly creates matches based on the most even matches, always resulting in little change in opponents, because there are so few opponents to choose from. a server doesnt get stuck with bad matches because of the tiers, it gets stuck because there are 0, 1, 3, or not 3N-1 equal opponents. and the system requires 3N-1 equal opponents for such a low competitor population.

It isn’t wrong at all. Picture a Tier 6 matchup where the three servers are equally matched and battle to essentially the same score. Unless one of the servers in Tier 5 loses by a large enough amount to drop down to Tier 6, or one of the serves in Tier 7 wins by a large enough amount to climb up into Tier 6, those three servers in Tier 5 will be playing each other week after week. It’s entirely possible for the servers in Tier 5 to be losing proficiency (population, coverage, key commanders) to the point that they are actually weaker than the servers in Tier 6, but if they do so collectively the Glicko-2 system won’t drop one of them down to Tier 6 unless a blowout happens.

The Glicko-2 system was designed to be used in situations where, for example, the player or team ranked #12 played #‘s 9, 10, 11, 13, 14, and 15 … and where #13 played #’s 10, 11, 12, 14, 15, and 16 … etc That creates a constantly overlapping reference for every team that more or less properly assesses their relative proficiency. You don’t get that when you lock teams into tiers … you lose the reference to teams above and below you unless there is lots of disparity in proficiency (i.e., blowouts happen) or unless something happens beyond your control above or below you. That’s a fundamental flaw any way you look at it.

the glicko 2 system is designed to have multiple matches in a small rating bracket. not 1 match covering a 300 point rating span. its meant to have a population of hundreds or thousands of competitors. not 24. not 50. at least 100.

my point is you may as well call “tier 6” “match 6”. the meaning would be no different. with only 8 concurrent matches, there is no room for opponent variation when server participation differs so drastically from server to server. giant rating gaps are an unfortunate and probably unforeseen side effect of only having 24 competitors, and wouldnt be an issue with a few hundred competitors to fill in the gaps. blaming so called tiers for a lack of competitors is whats wrong.

as an example of a fully fleshed out rating system: league of legends. there are millions of competitors. the ranking system is merely a flavor, riot chose a modified elo system as opposed to glicko. but a 1700 player does NOT get matched against a 1500 player because there are enough people playing that you can always matchmake 10 1500ish players. it only runs into issues with 2300+ players, they always get matched up against each other with oddball 1800+ players tossed in to the mix when matchmaking forces a match to be made with insufficient closely rated players. because the population of 2300+ players is really small. just like the population of servers in wvw. when you only have 24 players, weird things happen.

insanemaniac.2456:

Blaming so called tiers for a lack of competitors is whats wrong.

That’s simply semantics. Call them tiers, call them brackets, call them divisions, call them whatever you want … the root flaw is having so few participants in any one bracket.

You do realize that in a twisted sort of way both of us are identifying the same flaw in ANet’s use of the Glicko-2 system, right? I say it’s flawed for them to be using it to rank only three teams playing against each other versus other teams playing elsewhere, and you say that the system would work fine if there were more teams playing each other. We’re both looking through the same telescope … just from different ends.

Very unprofessional of Anet to fall back on their publicly announced decision

Vodka.3958:

Very unprofessional of Anet to fall back on their publicly announced decision

Very wise of them to listen to the Majority of their community<3. TY Anet for not reseting ratings^^

VOLKON.1290:

With a reset, the three T1 servers would have been split up… one in T3, one in T4 and one in T5.

The next week would not have been fun for the other servers in T3, T4 and T5.

Honestly, the top 3 tier servers could be mixed up into whatever combo you can think of, and in two to three weeks they would be on top again. Not because of numbers but because of skill.

And people migrate to places of competition. One reason why came from a mid tier server to a top one because I wanted to WvW only and I didn’t get the same level of play as I do now.