Guild Wars 2

Ok wait! before I get hammered to oblivion by the math loving hordes hear me out first, If what I have heard is correct that WvWvW uses the Glicko system, that means that the game is using a system that is designed to chart individual progress to make judgments about unpredictable masses.

Ok. Wait. put the bladed slide rulers and extra sharp pencils down for a moment and let me make my point before you math me to oblivion. WvWvW is not really a team sport, in the sense that you have established teams or even sides. The people hitting the field this week may not be the people who were playing all last week, in fact, the “teams” could change hour by hour when it comes to WvWvW.

Now, before you chew me too hard, and try to explain to me, that is what the math is supposed to accommodate and figure out, the truth is, simplicity is really better for dealing with massive things like this.

Even more then that, if you try to adjust your math to say “Oh they are supposed to lose, and that means team X needs to win by more to really “win” then you are removing a teams ability to win or lose of their own accord, and you end up with a stalemate of some teams stuck forever facing each other, but with always the same outcome, one always loses and the other always wins, boredom and tedium quickly set in, and it kills the WvWvW Dynamic of actually being able to just win and lose.

Now, If you had a means to judge and gauge every individual player in one world and then pit them against the individual players of the other worlds, you could very well make predictions based on who just stepped into the field that night and how it may end within the hour. But that is not what the math is doing at this point, it is treating each world like it’s own team, but not taking into account that each world is not a team, it’s a collection of individuals, who may or may not play today, they are trying to draw and design a coherency around something that simply does not have it.

Unlike the military, WvWvW is entirely volunteer based, the thrill is it’s own reward for us, as it’s not for money or fame, that is for sure. We are out there, pretty much fighting for the joy of it, but you can’t predict or treat people like that as a “team” the arrival or removal of people is whimsical at best and outright volatile at worst.

new players join, other quit, etc, etc, So, really, what needs to be done is removing the math. Simply put, stop trying to brain something as mindless as war.

Treat it like a Sport. Each “Area” is worth X points for every 15 min held, at the end of the day/week, the team that won that round of 3 moves up a tier, and the team that lost moves down.

No complex math, no “Oh by you need to win by a larger margin to really win based on this Glecko model” just make a win a win and a loss a loss and stop trying to over-think WvWvW.

Now what would removing the “complex Math” mean for WvWvW in the end of things. It would mean simply that every week, except for the top tier and the bottom tier, Each world would face 2 new opponents that were around their skill bracket.

For example:

in a 8 Tier system.

Tier 1 – 1st and 2nd place, stay, 3rd place moves to Tier 2
Tier 2 – 1st place moves to Tier 1, 2nd Place Stays, 3rd Place Moves to Tier 3
Tier 3 – 1st place moves to Tier 2, 2nd Place Stays, 3rd Place Moves to Tier 4
Etc, etc, etc, all the way Down to the List till Tier 8.
Tier 8 – 1st place Moves Up, 2nd and 3rd Stay.

This would keep words cycling, but also keep people close to where they belong without any form of stagnation, or would there be top or bottom tier traps.

Ergo – A world that belonged in the Tier 5 or 6 area, would move between then, maybe on a good week move up to Tier 4, and on a bad one move down to 7, but it would very easy move back up to 5 or 6 without this complex system trying to predict how well they should do.

This method would in fact, equalize the servers faster, and keep the fights fresh and new, and would establish a system where if some servers took a dive in population or some took a spike, they would move around swiftly and accordingly, as opposed to being stuck in point treadmills forever.

So yah. My feeling is that WvWvW was over analyzed and that needs to go away, and a simple method of making winning and losing mean more in the here and now should be put in.

I’m okay with how the raw ratings numbers are calculated, but the matchups shouldn’t be determined by simple ordering.

Just take the existing ratings and determine the matches semi-randomly as long as the standard deviation between each server’s ratings doesn’t exceed some point of imbalance.

Yes this might result in a couple bootstomps, but the higher potential point difference of the ‘green’ server means more potential mobility should the red or blue server take first place.

I would prefer a combination of the current ranking system and total WvW population over the course of the week to determine final scores & rankings.

The whole point of the highly detailed points system is so that we don’t get stuck for a week at a time in the kittenty wvw matchups that your volatile suggestion creates.

It’s no fun for anyone when the only WvW action is a server farming another server’s spawn with 30 ballistas.

Why would you even complain anyway? It’s not like you have to do the math. Just get out there and cap and defend.

Now that the free server transfer madness is over with I think this system seems pretty legit. First place moves up one, second place stays in the tier, and third drops a tier. Is there a reason this isn’t how it is?

Harbinger.9645:

Now that the free server transfer madness is over with I think this system seems pretty legit. First place moves up one, second place stays in the tier, and third drops a tier. Is there a reason this isn’t how it is?

None that I can think of.

Eliyahu.1467:

The whole point of the highly detailed points system is so that we don’t get stuck for a week at a time in the kittenty wvw matchups that your volatile suggestion creates.

It’s no fun for anyone when the only WvW action is a server farming another server’s spawn with 30 ballistas.

Why would you even complain anyway? It’s not like you have to do the math. Just get out there and cap and defend.

How would this proposed system do any of what you said? In fact…if you ask the lower tiers how the current point system is working out for them they will be happy to let you know exactly how the current system is pitting the same three servers against each other week after week after week with the same situations you describe and 0 tier movement. At the very least this new system will allow tiers to move more fluidly based on actual wins and losses.

First of all, the OP does seem to understand why the Glicko-2 system is a poor choice for something like WvW. It has exactly the problems he describes for the reasons he describes, and why ANet can’t seem to grasp that is beyond me.

That being said, the alternative described by the OP has been suggested many times here before and will result in more unbalanced matches. There is simply too much disparity in WvW player populations from top to bottom servers within GW2. You can’t have balance and variety from the same system. It will definitely change the matchups more often, but they won’t necessarily be more fair ones.

100% agree with the OP. This should be the system implemented. I know some people are worried about it creating uneven match-ups. The beauty of this system is that if you get blown out by a server, you only have to face them for one week!

Let’s be honest, playing the same server every week is boring. Getting blown out by one server every single week is even more boring. Sadly, that is the case now in 5 of the 8 NA servers. (I’m defining getting blown out by the first place server ahead by at least 30k points after the weekend).

Let’s get more exciting match-ups!

Harbinger.9645:

Eliyahu.1467:

The whole point of the highly detailed points system is so that we don’t get stuck for a week at a time in the kittenty wvw matchups that your volatile suggestion creates.

It’s no fun for anyone when the only WvW action is a server farming another server’s spawn with 30 ballistas.

Why would you even complain anyway? It’s not like you have to do the math. Just get out there and cap and defend.

How would this proposed system do any of what you said? In fact…if you ask the lower tiers how the current point system is working out for them they will be happy to let you know exactly how the current system is pitting the same three servers against each other week after week after week with the same situations you describe and 0 tier movement. At the very least this new system will allow tiers to move more fluidly based on actual wins and losses.

In the proposed system a server that barely wins or barely loses will change tiers and then likely get dominated and dominate respectively. This results in extremely unbalanced matchups essentially every other week.

The points system is good, but the overall ratings need to be tightened up a bit to help them settle with the new player populations.

style.6173:

100% agree with the OP. This should be the system implemented. I know some people are worried about it creating uneven match-ups. The beauty of this system is that if you get blown out by a server, you only have to face them for one week!

Let’s be honest, playing the same server every week is boring. Getting blown out by one server every single week is even more boring. Sadly, that is the case now in 5 of the 8 NA servers. (I’m defining getting blown out by the first place server ahead by at least 30k points after the weekend).

Let’s get more exciting match-ups!

Exactly!

In end, not only would it quickly give people a feeling of where they belonged, and would also account for things like vacations, school is out so populations on some servers rocket, others die down, and this system would adapt and quickly. in fact it is simplicity of the system that keep it vibrant and adapting.

Lets say a Server that took a large influx of people due to say, holidays, would not be stuck in some endless point cycle, it would move up, and then when the influx left, they would drop down, and not have to worry about falling to far or climbing too high.

Also, it would build tension as more evenly matched worlds faced off, not knowing who was going to win and who was going to lose, for tightly matched servers, it could very well be anyone’s game till the last hour for some servers and that is also what builds thrill as opposed to being able to predict what the match-up will be a week in advance.

But, in the end, you would never have to deal with the same severs week in and week out, and might even be thrilled to face “Those guys again”

Sometimes, complex is not the best way, and in this, when you have such a complex beast as WvWvW, maybe it can only really be tamed by the most simple yet enduring methods.

Eliyahu.1467:

In the proposed system a server that barely wins or barely loses will change tiers and then likely get dominated and dominate respectively. This results in extremely unbalanced matchups essentially every other week.

I have no idea why anyone would trade dynamic match-up for monotone conformity, but I will say you possess too much ego in regards to the tiers below you, and too much fear in regards to the tiers above you, and no matter where you are, that is a bad place to be.

Eliyahu.1467:

In the proposed system a server that barely wins or barely loses will change tiers and then likely get dominated and dominate respectively. This results in extremely unbalanced matchups essentially every other week.

Experience has shown that this is largely not so. And besides if servers keep winning or losing they will go up or down eventually and the same issue will arise, with more static populations now it is unlikely that there would be a large change to the population in the meantime.

Personally I dont think its much of an issue now. There was an issue before where it was taking Glicko too long to adjust to the large server hopping. Now with (expensive) server transfers, we aren’t going to see so much server hopping.

I completely agree with this thread make the fights simple if you lose you go down a rank if you win u go up a rank . My server has fought the same people for a month now we always win 1st or 2nd place yet we keep having to face the same servers week after week. Also I might add one of the servers always loses every match up always places 3rd and yet we have to keep fighting them week after week because somebody has done some smart math that sucks kitten If you win you don’t go up if you lose you don’t go down so really wtf is the point? Someone’s gonna say its not about winning its about playin and having fun and I say to you good sir BS Im here to win and have fun doing it not stay on this HAMSTER WHEEL TO NO WHERE!!!!

winner moves up and loser moves down is a TERRIBLE idea. There are huge difference between most tiers and you’d just end up with massively unbalanced matches every single week.

Its a bad idea be cause the second best server would always soley focus the third placed server.

Catisa.6507:

winner moves up and loser moves down is a TERRIBLE idea. There are huge difference between most tiers and you’d just end up with massively unbalanced matches every single week.

When you accumulate or lose enough points you move up or down, whats the difference?

And you’re overstating the difference. As I said above the experience shows there isn’t that much difference between adjacent tiers.

Math and I have never got along .. I suggest we pick matchups based on the server’s uniforms. Pink vs purple, yellow vs red, green vs orange .. /nod

Catisa.6507:

winner moves up and loser moves down is a TERRIBLE idea. There are huge difference between most tiers and you’d just end up with massively unbalanced matches every single week.

This is not the argument I would take against this.

As you and another have said, the proposed system would result in some bad matchups.

What you guys must realize is the current system has resulted in “massively unbalanced matches every single week.” Currently over half of the servers in NA are playing in matches they know with near certainty they can not win.

The whole premise is that there currently is not balance within tiers, that there is not balance between tiers doesn’t exactly counter the OP’s argument.

Okay so with no math you end up with things like:

T8: SF moves up and HoD moves down, do you really thing HoD won’t kitten slap the other two the same as SF has been doing?

T6: NSP drops and DH moves up, Do you somehow think NSP is going to beat GoM and HoD by less then DH has been doing? No in all likelihood, NSP will make T7 even more lopsided then it was. Compound this with a T8 moving to T7 and you essentially have a match up of 1 t6, 1 t7 and 1 t8 … that’s a disaster of a match up.

The problem isn’t the ranking system, its population/coverage differences. Anet really should have had separate WvW servers and strictly maintained population balances between then, then maybe we wouldn’t have big differences between tiers and some tiers with a good match but one that’s stale and old (personally I am sick of both enemy servers I have been playing again but I know any change and the match up will only be worse, so you end up with a choice, new faces or better match up).

Catisa.6507:

Okay so with no math you end up with things like:

T8: SF moves up and HoD moves down, do you really thing HoD won’t kitten slap the other two the same as SF has been doing?

T6: NSP drops and DH moves up, Do you somehow think NSP is going to beat GoM and HoD by less then DH has been doing? No in all likelihood, NSP will make T7 even more lopsided then it was. Compound this with a T8 moving to T7 and you essentially have a match up of 1 t6, 1 t7 and 1 t8 … that’s a disaster of a match up.

The problem isn’t the ranking system, its population/coverage differences. Anet really should have had separate WvW servers and strictly maintained population balances between then, then maybe we wouldn’t have big differences between tiers and some tiers with a good match but one that’s stale and old (personally I am sick of both enemy servers I have been playing again but I know any change and the match up will only be worse, so you end up with a choice, new faces or better match up).

Pretty much this I don’t know why the people of T8 seem to think just playing another server would make it fun and add variety. It will add variety thats for sure and the people of T8 would maybe have fun fighting someone different for 3 days then it would suck.

Any server that experienced kainengs rise to T2 I doubt would call it fun. Thats a different server right that all the tiers between 3-7 never faced I doubt any one really found that fun even if it was only for 1 week.

It sounds like it would be fun until you face it after reset weekend it sucks. The difference between a T2 server and T3 is night and day ask people of Fort Aspenwood how their recent visit to T2 was. My server was in Tier 4 for a while before we dropped and Crystal Desert for whatever reason dropped down to Tier 4 not a fun week at all.

Tighten up the rating a bit more but otherwise I think it is better than the proposed system from the OP.

Catisa.6507:

you essentially have a match up of 1 t6, 1 t7 and 1 t8 … that’s a disaster of a match up.

No. That is not “Essentially” what you have, that is exactly what you will have and it Sounds like fun, lets make it happen.

You will have:
The Worst of T6
The Holds their Own of T7
The Best of T8

Going Against Each other, and that is exactly what we really should have been doing all along, enough with the over braining what should be simple and direct.

We beat you this week, we move up.
You lost this week, you move down.
You held your own, you stay here.

If we see you again in 2 weeks great, if not, oh well. Welcome to direct conflict where we don’t try to maths victorious into defeats, or losses into wins.

Nuff said!

Ungood.3054:

Catisa.6507:

you essentially have a match up of 1 t6, 1 t7 and 1 t8 … that’s a disaster of a match up.

No. That is not “Essentially” what you have, that is exactly what you will have and it Sounds like fun, lets make it happen.

You will have:
The Worst of T6
The Holds their Own of T7
The Best of T8

Going Against Each other, and that is exactly what we really should have been doing all along, enough with the over braining what should be simple and direct.

We beat you this week, we move up.
You lost this week, you move down.
You held your own, you stay here.

If we see you again in 2 weeks great, if not, oh well. Welcome to direct conflict where we don’t try to maths victorious into defeats, or losses into wins.

Nuff said!

Yea your not biased at all. I know you guys in T8 are tired of SF I read the Anet doesn’t care thread but your idea is terrible.

So my upcoming match would look like

Ehmry bay vs Borliss pass vs SBI

Currently SBI is the worst of tier 3 higher than us but they lost a ton of people so they are falling. They might make it to T8 so you guys have someone new to fight same with IOJ.

That would be boring thanks to the math SBI will take our spot in tier 5 and we move to tier 4 next week. We would roll SBI and Borliss Pass it would be the same as the last 3 weeks in which we have rolled DR and IOJ except the servers would be called SBI and BP.

oZii.2864:

Catisa.6507:

Okay so with no math you end up with things like:

T8: SF moves up and HoD moves down, do you really thing HoD won’t kitten slap the other two the same as SF has been doing?

T6: NSP drops and DH moves up, Do you somehow think NSP is going to beat GoM and HoD by less then DH has been doing? No in all likelihood, NSP will make T7 even more lopsided then it was. Compound this with a T8 moving to T7 and you essentially have a match up of 1 t6, 1 t7 and 1 t8 … that’s a disaster of a match up.

The problem isn’t the ranking system, its population/coverage differences. Anet really should have had separate WvW servers and strictly maintained population balances between then, then maybe we wouldn’t have big differences between tiers and some tiers with a good match but one that’s stale and old (personally I am sick of both enemy servers I have been playing again but I know any change and the match up will only be worse, so you end up with a choice, new faces or better match up).

Pretty much this I don’t know why the people of T8 seem to think just playing another server would make it fun and add variety. It will add variety thats for sure and the people of T8 would maybe have fun fighting someone different for 3 days then it would suck.

Any server that experienced kainengs rise to T2 I doubt would call it fun. Thats a different server right that all the tiers between 3-7 never faced I doubt any one really found that fun even if it was only for 1 week.

It sounds like it would be fun until you face it after reset weekend it sucks. The difference between a T2 server and T3 is night and day ask people of Fort Aspenwood how their recent visit to T2 was. My server was in Tier 4 for a while before we dropped and Crystal Desert for whatever reason dropped down to Tier 4 not a fun week at all.

Tighten up the rating a bit more but otherwise I think it is better than the proposed system from the OP.

Yes because beating both servers by their scores added together +30,000 pts is fun and exciting and should lower their ranking. Get a clue before you post. Point is, bad or good, the math is broken if someone can beat another server group by that much and have their elo lowered. That’s the root of the problem, no matter how well you do, you’re stuck in t8… and although we might not know if HOD can absolutely and utterly crush us like everyone thinks, have you ever thought that you might have no clue what’s going on because you don’t play this tier or t7… like some of us who are posting… let’s say… this msg here

also, not saying I agree with the person posting, but any and all creative ideas are welcome at this point… this is frustrating everyone involved

oZii.2864:

So my upcoming match would look like

Ehmry bay vs Borliss pass vs SBI

I don’t think this is right.

From the current matchup, if you are on Ehmry Bay, your tier would do the following:

Ehmry Bay will move up and trade places with SBI.
Yak’s Bend will trade places with Maguuma.
CD will move into 11th place in T4.

your new matchup (from green up red down) would then be

Maguuma – Green
Crystal Desert – Blue
Ehmry Bay – Red

*note: this is the same predicted matchup next week under the current system.

and you might say you have no chance against Maguuma.. but as it is you guys are mopping the floor in a tier that has no chance against you and Maguuma is getting stomped in their tier so.. your new matchup would be anywhere from about the same to better.

oZii.2864:

It will add variety thats for sure and the people of T8 would maybe have fun fighting someone different for 3 days then it would suck.

So you admit it would be more fun.

fun fighting something different for 3 days = 3 days of fun

3 is greater than zero.

Same thing I posted in another topic I’ll post in this one, as my argument is the same:

https://forum-en.gw2archive.eu/forum/pvp/wuvwuv/Bring-Sorrow-s-Furnace-out-of-Tier-8/1421612

Kerithlan.1659:

My only problem with the “you win, you move up; you lose, you move down” plan is that it generates poor match-ups more often.

I totally agree that we don’t want servers like Sorrow’s Furnace hopelessly eating other servers for months, but at the same time I don’t want to see servers like Ehmry Bay getting paired up with Anvil Rock, or servers like Sanctum of Rall getting matched up with Fort Aspenwood.

Additionally, a simple winner goes up/loser goes down system breaks up competitive tiers that probably should remain together. Borlis Pass/Anvil Rock/Northern Shiverpeaks have been neck and neck for several weeks now and it’s almost a toss-up who’s going to win each week. I don’t think we should arbitrarily break that up for the sake of change if the match-up is healthy, exciting and fun.

Lot of good points from all sides tbh, and a lot of good pros and cons. Wish I could +10 almost all these points

The bit about breaking up healthy tiers is probably the best argument against it(winner up loser down) one could make.

The only problem being there are so few healthy tiers. (and no matchup change will help that)

This is the problem that would have to be addressed somehow before any matchmaking system can work to everyone’s liking.

I think it’s worth a shot. The current system is dreadful.

I think it is becoming apparent that Glicko causes stagnation in the long run. This is not something that can be fixed by tweaking numbers: the system itself, by comparing your performance to your history, promotes stability in a setup where server populations are not going to mass shift ever again. It should be clear that Glicko is going to give us crystallized tiers over time – T8 is crystallized, and in turn the same fate will happen to T1 (already happening), then T2 and T7, then T3 and T6.

We do not want crystallized tiers. When we say that, we agree that there will be some imbalanced fights every now and then. That said, winner moves up and loser moves down is going to create a lot of imbalanced fights. NA T6 is fighting it out perfectly fine alone, and the proposed simple approach would wreck it.

But, the premise is a system that encourages changes. So we need to start from this, and tweak. On a related note, every sport league I know that has more than one tier (football or anything else in europe) employs this kind of system. With good reason.

So let’s say Winner Up Loser Down (WULD) is better than Glicko. We want WULD to not break such a magnificent setup as is NA T6. Very well, let us add a constraint: the more balanced a tier is, the harder it is to enter in it. Let’s call D the difference in points between the winner (1st) and the loser (3rd) of a certain tier. Let us decide arbitrarily that a tier’s ideal D is 50k points. We call this magic number M.

The tweak would work like this: a server in T7, to enter T6, must have a D(T7) > M – D(T6). What does this mean? It means this: if the difference between first and last in T6 is only 10k points, the winner of T7 will need to have 50k-10k = 40k points of difference from the last team in T7 to go up.

In T1, currently, D is about 35k points: the winner of T2 would need 15k more points than the last in T2 to move up.

In T3, currently, D is more than 50k points; this means that any winner from T4 would move up.

Get it? Now all that’s left is figure out a good M, possibly a variable one according to some convoluted rating system. Now is the time to over-engineer with ratings, since we now have a proper system.

The only problem with basing it off of final score is that you’re comparing the scores between two different match-ups. I understand that it doesn’t seem like a problem to some folks, but to me it is.

The reason for this is that you’re essentially comparing apples to oranges. I understand that all servers and all tiers are all comprised of the same exact collection of mechanics. I also understand that population is the only real x-factor (contributing variable size, skill, coverage, and so on).

What needs to be understood is that a server scoring big in one tier doesn’t have any bearing on the tier above or below them. Let’s use Sorrow’s Furnace as an example to explain this. Sorrow’s Furnace would move up in your model, and would win out over Henge of Denravi by quite a lot. This would boost them up to Tier 7 (as is desired), but the consequence is that Henge of Denravi isn’t necessarily the best fit for Sorrow’s Furnace’s previous opponents.

Henge of Denravi is in a predominantly healthy match-up right now. They’ve been very close with Gates of Madness for a while now, and in fact their server seems to have a relatively good balance. Their score is lower not because Sorrow’s Furnace is necessarily better than them, but rather because Henge of Denravi has better competition while facing Gates of Madness and Darkhaven than Sorrow’s Furnace does against Ferguson’s Crossing and Eredon Terrace.

The point of the current system is to ensure that servers are eventually and ultimately paired up with other servers with a similar type of play. The system you’re proposing ruins the current Henge of Denravi/Gates of Madness balance and doesn’t necessarily correct the issue Sorrow’s Furnace is experiencing.

Another problem with the system you’re proposing is that it causes servers to drop down more slowly than in the current system, and likewise to rise more slowly than the current system. Disallowing servers from moving multiple tiers (as your model framework does) creates more poor match-ups than might otherwise show up using the glicko ratings.

For example, a couple of weeks ago SBI dropped from Tier 2 to Tier 4, completely skipping Tier 3. Currently, SBI is slated to drop to Tier 5. In the system you’re proposing, SBI would have lingered in the upper tiers for much longer than would otherwise be necessary as they would be forced to go through Tier 3 before reaching Tier 4. Ultimately the results would be the same, but your model would have slowed this progressive decline toward SBI’s new home by at least a week.

Likewise, there would be no positive gain in terms of server advancement. Servers scoring big wins would be prevented from skipping tiers they don’t belong in and would be forced to spend weeks facing servers they don’t belong against. Sure, this wouldn’t matter much for servers like Kaineng that visited each tier on their way up anyway; but for servers such as Sanctum of Rall that did skip tiers on their way up this would mean an additional week wasted.

The glicko system, for all of its faults, is simply superior to any potential, “winner goes up; loser goes down,” system because it allows for these rapid changes to take place. I completely agree that Sorrow’s Furnace needs to move out of T8 and into a better position, but I don’t believe removing the glicko system is necessarily the best way to go about that.

It might be easier to simply tweak the existing glicko formula to not require such exponentially larger victories in order for servers to move around than anything else. As was previously stated in this topic, simple is often the best answer. If viewed in the bigger picture, the simple solution is to tweak the math that already exists, and not overhauling the entire system.

In the end and no matter what happens, Sorrow’s Furnace moving up will ultimately break T7, which will potentially break T6, and so on. Getting new players into Eredon Terrace and Ferguson’s Crossing seems to be a more ideal solution in my mind than overhauling the rating system, but that’s a pipe dream (unless ArenaNet makes transfers to those two servers free or something).

Actually, I think the current situation is misleading you a bit.

First, I have to ask you a question. You state, let’s assume it to be true, that T7 is a good matchup. I will counter-assume for the sake of discussion that SF’s true strength is somewhere in T5-6. The true power ranking then is such that only two servers are worse than the T7 ones. This means that in the true power ranking, one of the T7 servers actually belongs to T8.

Now, is the better system the one that gradually freezes and leaves SF hopeless in T8 for the sake of maintaining an even pairing, or is it the one that breaks the good matchup because the three balanced teams are lined up wrong with the tiers? Yes, my system would break T7, and I think it is the correct thing to do.

I’ll give you an analogous, still hypothetical, example: JQ is the best server. SoR, SoS, and BG are of equal strength. Should we have these three fight together in T2 forever and leave JQ pounding on kaineng/TC, either in T1 or T2? No, definitely not. JQ should be stable T1, and the other three teams should battle it out. Glicko will freeze T1 as JQ/SoS/SoR, it is already happening, and it is the nature of the algorithm. Tweaking won’t fix things.

On the speed of moving up/down: it is true that my system makes you move up/down for only one tier per week. Being on SBI, I can tell you that it has been around two months since we last won a matchup, and we have ended up third like seven of those times. With my system, we would have had plenty of time to reach T5-6 where we belong now. With Glicko, our super high rating forced us repeatedly into artificially high tiers as we had to be pummeled repeatedly in order to lose enough rating to go down. Then all of a sudden we cross kaineng and maguuma and drop two tiers in once, but well, it’s been very late. With my system, SBI would have stabilized a lot faster. Same goes for Kaineng which has been winning for about the same time.

Scores in different servers are apples and oranges, correct. With Glicko, those playing with apples will never see an orange again though. With the playerbase freezing in the servers now that server transfers are expensive, server strengths will freeze, and Glicko ratings, which are designed to handle well power changes, will instead freeze the server ladder. My system guarantees variety and a “win some, lose some” environment.

As an addendum, the history of SBI from our last win (17 November).
24/11: third by 60k points, we stay T1 (red)
1/12: second, we stay T1 (blue)
8/12: third by 40k, we stay T1 (red)
15/12: third by 30k, we stay T1 (red)
22/12: second, we stay T1 (blue)
29/12: third by 40k, we stay T1 (red)
5/1: third by 120k, we stay T1 (red)
12/1: third by 100k, we go T2 (green)
19/1: third by 90k, we stay T2 (blue)
26/1: third by 220k, we stay T2 (red)
1/2: third by 410k, we go T4 (green)
9/2: third by 100k, we stay T4 (red)
currently: third by 80k, going to go T5.

Three months to move down three tiers so far. Glicko is not fast for moving rapidly dropping teams.

The problem with that argument is that you ignore some facts in favor of others to support your argument.

Again, I’d point to the fact that Kaineng only spent one week in each tier under the glicko system as counter-evidence to the claim that they’d have risen any faster under the system you’re proposing. In fact, your system does not allow for potential shifts to take place where Kaineng would have been able to skip tiers, whereas the glicko system does (as is evidenced by Sanctum of Rall’s jump from T5 to T3 in October).

Additionally, your proposed system solves a handful of problems while simultaneously creating an equal or greater number of new problems. In my mind, that isn’t a solution, but rather an exchange. It’s shifting the issue rather than correcting it. In this case, it’s attempting to correct the problem of poor match-ups for some servers by creating more frequently poor match-ups for all servers. That is a sub-par approach in my eyes, and one that we should not be pushing.

SBI has indeed lost the last eight match-ups, but under your system there would be much more volatility and instability for all servers. If we are to assume that SBI belongs in T5/T6, and if we assume that 8 weeks ago is when SBI’s fall should have begun, we’d have ended up with an incredibly different scene.

Instead of the 7th week seeing SBI in T1 again, you’d be in T2 and Blackgate in T1. Sure that sounds fine on paper, but an additional consequence would have been Tarnished Coast and Isle of Janthir switching places again. We’d have seen Fort Aspenwood and Maguuma swapping places, we’d have seen Dragonbrand and Ehmry Bay swapping places, and we’d have seen Henge of Denravi and Kaineng switching places.

None of those are necessarily healthy match-ups: IoJ had begun to implode (which is why TC originally moved up) and would have been no match for SBI and SoR; FA would have absolutely mopped the floor with T4 while pre-transfers Maguuma would have been smashed by TC; and Henge of Denravi would have begun playing horseshoes over T8/T7 several weeks sooner than they currently are.

The system you’re proposing ultimately pairs up servers that don’t belong together and ensures that servers that belong together won’t be together, while the current system at least makes attempts to group relatively equal servers together. Let’s make up some servers to demonstrate this point.

9. Cavalon

10. Istan
11. Snake Dance
12. Mehtani Keys

13. Urgoz’s Warren

Let’s say that Cavalon lost this week, Snake Dance won this week, Mehtani Keys lost this week, and Urgoz’s Warren won this week. The new model then would look like this:

9. Snake Dance

10. Cavalon
11. Istan
12. Urgoz’s Warren

13. Mehtani Keys

Looks good, right? There’s a problem though. We now have servers from three different tiers facing up against each other in a single tier, which all but guarantees a blowout win for Cavalon in a stabilized environment. This is because coverage and population are the greatest factors in determining ladder placement.

Under your system, we’re pitting vastly larger population servers against one smaller server and one much smaller server more often, which is decidedly bad for competition. Even though glicko creates stagnant match-ups, it also pays attention to population shifts and moves servers accordingly long-term, which is something that your system is incapable of doing.

In essence, your system attempts to over-complicate a simple problem with a much more volatile and inconsistent solution. I’m not necessarily defending glicko as the end-all-be-all, but it is a far reach above what you’re suggesting.

A balanced match-up between three servers is what I would consider a healthy “Win some, lose some” environment, as we can see with T6 right now. It is very likely for any of the three servers to ultimately win week-by-week, and we can see a similar scenario right now between Blackgate/Kaineng, Henge of Denravi/Gates of Madness, Vizunah Square/Elona Reach, Piken Square/Seafarer’s Rest, Gandara/Abaddon’s Mouth, Miller’s Sound/Drakkar Lake, and many more.

Forcing servers to have blowout matches more often isn’t healthy, and creating blowouts for the sake of having blowouts is likewise unhealthy. I am not arguing against a better balance for any tier, but rather attempting to explain why your solution is actually poised to create more problems more often. I hope it’s a bit clearer now.

mucco.1867:

As an addendum, the history of SBI from our last win (17 November).
24/11: third by 60k points, we stay T1 (red)
1/12: second, we stay T1 (blue)
8/12: third by 40k, we stay T1 (red)
15/12: third by 30k, we stay T1 (red)
22/12: second, we stay T1 (blue)
29/12: third by 40k, we stay T1 (red)
5/1: third by 120k, we stay T1 (red)
12/1: third by 100k, we go T2 (green)
19/1: third by 90k, we stay T2 (blue)
26/1: third by 220k, we stay T2 (red)
1/2: third by 410k, we go T4 (green)
9/2: third by 100k, we stay T4 (red)
currently: third by 80k, going to go T5.

Three months to move down three tiers so far. Glicko is not fast for moving rapidly dropping teams.

Anecdotal evidence in SBI’s case, especially when the volatility of the end of free transfers and the mass exodus is taken into account.

SBI was originally a good match-up for TC and SoR before half of SBI moved to SoR in Week 3 (19 Jan), and only lingered for a single week before dropping to Tier 4. Additionally, once there SBI was paired up with another falling server (Crystal Desert), and so understandably was slowed down.

It isn’t as dire or as sensational as you’re making it out to be. SBI had an appropriate rating drop for the circumstances it has experienced.

In your system, neither Crystal Desert nor SBI would have moved along at an appropriate rate (as both servers are falling). In fact, Crystal Desert would have incorrectly moved up a tier and into a poor match-up, which frankly is a bad exchange: as servers drop into better match-ups, it shouldn’t cause other servers to move up into worse ones. The same is true of the opposite scenario: as servers move up into better match-ups, it shouldn’t cause other servers to move down into worse ones. That’s what your system causes.

One other problem with your proposed system: it makes blue servers switch positions even less frequently, as they’re sandwiched between much larger and much smaller servers.

You are indeed a bit misled by the current situation.

The first thing you do is saying my system would have created more volatility: this in a period in which everyone and their mother bandwagoned around. Being volatile is not exactly inaccurate in this case! When there are big power shifts, Glicko does predict the trend correctly. Nobody disputes that. I dispute that it is slow, but not that it is correct. I think it is very ill-suited for our situation.

We are going to have monolithic servers with very small power changes for possibly years to come. With Glicko, we are guaranteed that every matchup will freeze, eventually, because it is the nature of the algorithm. My system creates unfair matchups from time to time, but it corrects itself within a week or two at most.

Let’s move on to your r9-13 example with my system, because I’m not sure you grasped its essence – it’s not straight WULD, it is mitigated, and it is essential to mitigate it. If SD wins by little, and as such the gap is small, it won’t have enough to make it to the next tier unless Cavalon crashes badly; likewise, if r10-12 was a close matchup and as such the gap is small, Urgoz would need a landslide victory.

I would believe that a pummeled Cavalon would be more than happy to spend a week winning some in the lower tier, just as Urgoz would be happy to pit themselves against stronger forces, for once. If they won by enough and the other tier was imbalanced, they get one chance. But if r10-12 is truly balanced, it will be hard for Cavalon/Urgoz to enter it. In the long run, Istan/SD/MK will be the T4 matchup three times out of four.

With Glicko? 5 years straight of Istan/SD/MK. Are you sure about that?

Without going too deeply into it since I’m into my third beer and will soon become incoherent:

I understand that your system is artificially gated through score comparison mathemagics, but what’s missed is the irrelevance between scores between tiers. A huge score in one tier is not necessarily representative of an improved match-up when the winner moves up (see: HoD vs SF being a good match until Kaineng happened, pushing SF down).

Many players ultimately don’t care about the score (especially in Europe) and would rather be paired up with opponents that they can have competitive and fun encounters with/against. Yes the score can be and is a motivating factor, but to solely determine placement based off of scores ignores the long-term ramifications such a system would create.

Servers will never become locked into match-ups for too long because there will always be shifting populations, even with free transfers ending. Old players will eventually retire, new players will come in, and each server will be in constant (albeit smaller) population flux. The glicko system is simply superior because it can account for such fluctuations.

It may not be the best, but it’s definitely superior.

No, 500k’ing one tier goes not guarantee you’re better suited for the next tier. But it gives you the right to try yourself in it, doesn’t it?

Glicko will freeze all ratings. It’s how it works. Look at T1 and T8 both NA and EU. T1 is 100k points ahead of T2 already, and the gap is getting larger. T8 is hopeless for two teams already and in NA the third team might never succeed in getting out. If you look at the formula, you will realize this locking is going to expand toward the center of the ranking gradually until nobody can move anymore. Glicko is paralyzing.

Populations will not shift. Some servers might drop down, and they might do so very slowly (see SBI). But the population flow is expected to be uniform across servers. Therefore, Glicko is “superior” in the only case that you want rankings to never move again for the eternity.

assuming minimal population shifts, the goal of the rating system is to eventually settle down into the same matchup forever. this gives the most balanced matches score-wise. a close match makes for a good match.

this is the intended goal. the eventual freezing of ranks is a GOOD thing. it means you will have meaningful matches every week. who cares if its the same 2 servers, if the matchup is competitive and all 3 servers have a good chance of winning? the rankings aren’t there for you to be #1 in T1, its there to setup a bracket in terms of population so you can eventually get matched with a similarly populated server, so you can have close matches.

with your system, you’ll have 2 new opponents every single week, with many imbalanced matches. because every tier except the first and last will have 2 servers seperated by 2 tiers. a T2 server drops to T3, and a T4 server goes up to T3. you have the T2 server dominating both, rinse and repeat next week, where another T2 server drops and another T4 server rises. one sided matchups is not fun.

I agree whole-heartedly with the original poster. He really seems to understand things in a way that most people (even those at ArenaNet) don’t. The current system leads to stagnation, and I know for myself at least, what really appealed to me when I first heard about WvW was the idea of facing different servers each match…getting variety and keeping things fresh, facing new tactics and different players and rising to the challenge. If stagnation lead to good match-ups, I might agree that the current system is best…but it only leads to good match-ups in a few tiers…everywhere else it cements bad match-ups into permanent, depressing quicksand.

The psychological effect of the current system on players is very negative. It allows bad feelings between servers to fester and grow for weeks and weeks, and turns the experience into petty bickering and a lot of rude and negative behavior. For all but the winning server, player populations shrink as people get tired of seeing the same thing and completely assured outcomes.

Moving servers around and not allowing bad feelings to build up would probably raise the level of play, and the level of civility and sportsmanship in all matches. What was the point of not showing names of players on other servers, if ArenaNet planned on sticking us in a permanent match-up with two other servers? Before long, we all recognize each other and it becomes personal…and that’s supposedly what they were trying to avoid.

The current system is worst in the bottom tier, where a giant hole is created by the math, making servers appear much worse than they are and forcing endlessly lopsided matches. Worse losses lead to less people playing WvW leading to even worse losses. At least in a system like the original poster describes, players on losing servers would know that each week they would have a fresh possibility of a different outcome, and likely the WvW participation would rise and make matches more competitive.

Of course servers who are winning week after week don’t want any change at all in the system. They don’t want a reset, they don’t want match-ups to change more often. Why? Because then the outcome isn’t assured and every server will have good weeks and bad weeks. That would be infinitely more fair and more fun than what we have now.

mucco.1867:

Therefore, Glicko is “superior” in the only case that you want rankings to never move again for the eternity.

I believe that this is a dishonest and misguided belief, as it assumes stagnation over time. Again without going too much into it, population simply does in fact move over time.

Anything from holidays to regional lockouts to new games coming out to sudden popularity surges for GW2 (such as expansions) will change populations. It is short-sighted to believe any one server will be locked in position for greater than 6 months, or to believe that the end of free transfers will significantly dent population flux.

Additionally, to only look at the 500k win blowout is to ignore the other half of the equation. For an example of what I’m talking about, look at NA T2 and T3 this week vs last week. Kaineng winning and FA losing improved Kaineng’s experiences at the cost of FA’s experiences. FA went from a great match-up in T2 to what is more or less a blowout in T3, while Kaineng went from a T3 blowout to a competitive T2.

I wouldn’t call that an improvement, but rather an exchange. Your system creates more exchanges, not necessarily balance.

Except one-sided matchups will be the norm. Ask T8.

Glicko will freeze everyone in their current position, eventually. If in T4 green constantly does twice as many points as red+blue, and every week the results are the same, in a couple years their ratings won’t shift one point; despite it being a very imbalanced setup. If the intended goal is to freeze the ladder into whatever (usually bad) form it has at the moment, Glicko works perfectly.

With my system, you will NOT have two different opponents every single week because it’s not straight WULD. How many times do I have to repeat it?

The problem is: Glicko works a bit like a pump, that moves points from the bottom to the top.
When a server goes up a tier, it has higher points than the one that come down.
By that the upper tier gets an infusion of points that increases the mean-rating in the tier and the lower tier get a reduction of points that decreases the mean-rating in the tier.

Matches in a tier, are more or less neutral to the mean-points of the tier, what servers get as plus the other servers loose as minus, i.e. the sum of all plus and minus in match is more or less 0, and therefore the mean of a tier doesn’t change much.

In all tiers beside the top and the bottom tier this is more or less balanced, because they have a plus from the lower tier and a minus to the upper tier.

But the top and the bottom tier are imbalanced, top only gains points and bottom only looses points and thats why the borders to them become more and more impassable.

I think to avoid that there must be an external reduction of points in T1 and infusion of points in T8 (or a cap, after the rank is determined every server above 2000 pts is caped to 2000 and every server below 1000 points is capped at 1000).

Fozzik.1742:

I agree whole-heartedly with the original poster. He really seems to understand things in a way that most people (even those at ArenaNet) don’t. The current system leads to stagnation, and I know for myself at least, what really appealed to me when I first heard about WvW was the idea of facing different servers each match…getting variety and keeping things fresh, facing new tactics and different players and rising to the challenge.

The psychological effect of the current system on players is very negative. It allows bad feelings between servers to fester and grow for weeks and weeks, and turns the experience into petty bickering and a lot of rude and negative behavior. For all but the winning server, player populations shrink as people get tired of seeing the same thing and completely assured outcomes.

The current system is worst in the bottom tier, where a giant hole is created by the math, making servers appear much worse than they are and forcing endlessly lopsided matches. Worse losses lead to less people playing WvW leading to even worse losses. At least in a system like the original poster describes, players on losing servers would know that each week they would have a fresh possibility of a different outcome, and likely the WvW participation would rise and make matches more competitive.

Of course servers who are winning week after week don’t want any change at all in the system. They don’t want a reset, they don’t want match-ups to change more often. Why? Because then the outcome isn’t assured and every server will have good weeks and bad weeks. That would be infinitely more fair and more fun than what we have now.

I was in favor of the ratings reset and still am. I am not in favor of oversimplifying the system to the point that it becomes much more problematic than the current environment is.

mucco.1867:

With my system, you will NOT have two different opponents every single week because it’s not straight WULD. How many times do I have to repeat it?

I’m not saying every single week. I am saying more frequently, which is an honest evaluation.

mucco.1867:

Except one-sided matchups will be the norm. Ask T8.

Glicko will freeze everyone in their current position, eventually. If in T4 green constantly does twice as many points as red+blue, and every week the results are the same, in a couple years their ratings won’t shift one point; despite it being a very imbalanced setup. If the intended goal is to freeze the ladder into whatever (usually bad) form it has at the moment, Glicko works perfectly.

With my system, you will NOT have two different opponents every single week because it’s not straight WULD. How many times do I have to repeat it?

except the current rankings aren’t balanced yet. so they won’t freeze where they are now. the system is self correcting over time. any imbalances will iron itself out. in the ideal case, the end result is all tiers have close matches, and their ratings don’t change. this is when you have balance.

Kerithlan.1659:

I was in favor of the ratings reset and still am. I am not in favor of oversimplifying the system to the point that it becomes much more problematic than the current environment is.

Resetting the system just delays the inevitable, unless you change the system. After a reset, a few months later you are in the same situation of heavy stagnation. Maybe individual servers’ positions would be shifted from where they are now, but it would be just as bad, and just as stuck.

Is this really what people want? To play the same two servers every week forever? I’d much rather have variety, even if there are some unbalanced matches. Sometimes they would go your way, sometimes not…much more fun and requires effort on the part of players, rather than a guaranteed, predictable outcome every week for most servers leading to players getting bored or frustrated and no longer logging in.

With my system, different matchups will happen as much as you want: it is sufficient to tweak the M value.

wads: you do not fully grasp Glicko. Glicko is not self-correcting, as it suffers from a problem similar to greedy algorithms stumbling on local maximums. T8, I will repeat, is the proof. NA T1 and T7 are already becoming like that. Ratings WILL freeze if servers play like they played the last weeks. The end result is that curbstomping servers will keep curbstomping, and fair matches will stay fair matches.

Ratings don’t change != all close matches. In fact, If a server goes 500k ten weeks in a row, its rating will not change from like the fourth week onward.

I think that there’s good and bad with a WULD system.

The good is that a the red server in the T2-T7 gets to see the tactics used by the green server 2 levels above them as well as the blue server one level above them.

The blue server gets to see the tactics of the green server from the level above.

THE BAD:

It will end up that all tiers have the same tactic….SMASH RED
Green wants to wipe red twice as fast as they wipe blue
Blue wants to wipe red as much as they get wiped by green
Red will just try to hold on but won’t be able to match up.

The matches will probably be over before the Oceanics get to play in their prime time.