Guild Wars 2

First, can we agree that the ranking system for WvW is pretty bad?

That’s why I decided to give my best shot at designing a new ranking system. My design philosophy was the following :

1. No randomness
2. Should be simple (so no more glicko)
3. If you win by a lot you should move up
4. If you lose by a lot you should move down
5. Don’t break perfectly balanced match up (unless #3 or #4)

You can see the live demo over here : http://coveragewars2.com

How does it work?

The only thing that matters to determine if you move up or down is your score relative to your opponents for the current week. Preceding weeks do not matter at all.

For a server to move up, his point minus the point of the weakest server in the tier above him must be 37. Note that 37 is just a number that I chose for this example. You can adjust it to make it easier or harder to move up/down. An higher value will make it harder, while a lower value will make it easier.

The formula to calculate the point of each server is the following :

server_point = ((server_score / average_score_in_tier) – 1) * 100

What do you think?

TLDR: Just check the demo it was a nice visual table.

ranking system for WvW is pretty bad?

there is no ranking system, look here:

https://forum-en.gw2archive.eu/forum/game/wuv/how-anet-s-linked-server-ranking-system-works/first#post6274364

Good Ranking of what ?

Ranking to indicate what server is better, is pretty useless atm because the server with the most active players will easy outrank any server that has obvious less active players
So you might as well count the number of active players to get the current rank.

The only way any ranking system will give real and usefull information regarding with server (accumilated players on a server) are better that other server is when you force the servers have about equal number of players at all times that other servers.

Forcing (approx) equal number of player can be done, but we wont like it.

Because you still need to determine which server should fight against which one.

Would you want the strongest server in the same match up as the 2 weakest servers? That would be pretty boring for everyone.

This thread is a prime example of why we need a new ranking system :

https://forum-en.gw2archive.eu/forum/game/wuv/Anet-Seriously-STOP-Screwing-with-Glicko/first

Just to note that SBI would have moved up last week with the system I describe.

This is good. But I think it could be better. For example I don’t think this produces enough variety.

Take T1 for example. There are basically four T1 servers; TC, DB, BG and JQ. They should switch in and out of T1 on a regular basis regardless of how big the victory in T2 is. Heck I would even say throwing Mag into T1 for a week would be good.

And I don’t think it should be CD that goes down to T4. It should be a rotation between CD, YB, SoS and SBI. Maybe you can put some factor in the equation to create different matches if the scores are within some boundary.

Johje Holan.4607:

This is good. But I think it could be better. For example I don’t think this produces enough variety.

Take T1 for example. There are basically four T1 servers; TC, DB, BG and JQ. They should switch in and out of T1 on a regular basis regardless of how big the victory in T2 is. Heck I would even say throwing Mag into T1 for a week would be good.

And I don’t think it should be CD that goes down to T4. It should be a rotation between CD, YB, SoS and SBI. Maybe you can put some factor in the equation to create different matches if the scores are within some boundary.

It’s a good point. Let me explain.

Algorithm details

The algorithm objective is to create balanced match up. As long as the match up are balanced it will not change them but as soon as a match up is unbalanced enough it will switch server around in order to find better match up.

Its strength is that it would adapt very fast to population changes. For example, if a tier 4 server received a massive bandwagonning, in only 3 weeks, that tier 4 server that is now overpopulated could reach tier 1.

Currently, the number of point required for a switch is 37. It’s still quite hard at the moment to climb or fall a tier with that number so it could be reduced to improve the match up variation. I think it should be reduced a bit at the moment.

To answer your question

There are 2 things that could be done to increase match up variety.

1. Reduce the number of points required to climb a tier
2. Introduce a random number

About #1, you cannot reduce it too much or it will transform into one up one down.

About #2, this is similar to what Anet is doing at the moment but with the glicko rating. For example, it could help Maguuma to get thrown into tier 1 but I think they still need to get a bit stronger before they enter that ring. About CD, it could definitively push YB down instead of them for 1 week with a lucky roll.

But, do you like randomness? Personally I’m not a fan but it’s easy to add.

Why have a ranking system with the current population anyway? As someone already stated population [coverage] affects the score the most in a 24/7 match. You can try to artificially inflate/deflate scores all you want, but it won’t change the amount of people playing.

Instead, abandon the 24 hours system, and just segregate servers based on time-zones. Probably in increments of 6 hours as an example. This means all points stop outside the timer by some manner. If there are enough match-ups of the same timezone, you can then consider having a tiered ranking system for given timezone.

1. Run a poll of the population asking their time slot they prefer to play.

2. Create servers that represent data from poll:

For example, assuming we’re using the NA servers:
1 match up for Sea Timezone (3 servers)
1 match up for EU Timezone (3 servers)
1 match up for OCX Timezone (3 servers)
3 match ups for NA timezone (9 servers)

3. For those that play for multiple time-zones, you can allow 1 choice in each other time-zone outside the primary one you chose. It should cost gems to transfer within the same timezone however.

This would mean for the so called “dead” time-zones will now have a higher chance of finding people instead of k-training empty towers or doing running simulator for 50% of the raid. If those are activities you like to do, I’m sorry that’s not the audience WvW should cater to imo.

There is a concern over of destroying “server” communities. I can somewhat understand that fear for those communities that are still even alive. You can argue the idea of ‘community’ would probably shift from server, and more on to guilds. Juxtapose that against the current problems of coverage (and bandwagoning).

TLDR: It’s easier to have meaningful rankings as a match is shorter in duration, e.g. make time-zone segregated servers representing populations based on a poll. This most definitely isn’t a new suggestion. Unfortunately, searching function on the forum is janky.

@Chinchilla Is one of the goal of your change to make WvW competitive? If yes, you are not going far enough.

Indeed, WvW cannot be competitive until you remove the 24/7 factor but you also need to remove the 4 maps factors as there are too many players required to make it controllable.

For WvW to become competitive, they need to make it guild based, on one single map for a period of 4 consecutives hours max and probably only on the night the guild is available to raid. So basically PvP but in a WvW setting.

My point is that what you are suggesting is such a big change that if they were doing it, I would prefer that they go all the way and make a real competitive mode to WvW.

Obviously, the casual mode will need to continue. So one competitive and one casual.

But back on topic, the ranking system I’m suggesting is just to fix the glicko mess in the easiest possible way.

I’m too lazy to go into the numbers and stuff, but it looks fairly fine except for little match variety.

Especially tier 1+4 would end up with 2 static servers and 1-2 that sometimes changed back and forth. And in case of T1 could possibly end up stagnant for 2-3 weeks in a row. On the positive side, even if say T3 is stable and competitive, the big server in T4 doing well enough could force a rotation, so I think it has some good sides.

It can create some interesting human interactions though. Like if 1-2 servers are just sick of another server they’re playing against, they can either focus it, or just take a break for a week and let them score run away, to get rid of them for a week. So expect some servers to game this.

Regarding competitive, I agree that if they want to actually make a competitive mode, go all the hog way. Make a GvG (like) mode, that functions like a PvP match on WvW maps. Pick one map (EBG, ABL, DBL, EOTM) and get together 2 or 3 guilds/teams of 40 each or more. And set off 2-4 hours and play. That is the only way to make it actually competitive.

Once they do that, normal wvw is likely going to get mashed into a EotM system to cement it as casual once and for all.

Chinchilla.1785

(That sounds like something I suggested once, forgot where)

Iirc I suggested that server split system as a way to use with the linking, it would sort of end up as only 1 tier having full time coverage, but essentially everyone would also be a part of those “off hours” servers, even if they didn’t play there.

Basically anyone that ever played later than NA prime, into NA night, would also be part of one of those 3 servers. So you could continue to play, but everyone at that time would be forced into 1 match-up.

Not certain exactly how that would work, but I suspect (Strongly) that it would have to function as a server transfer, and thus you would have to get thrown out of the game every 6 hours, to change servers for 5-10 minutes. Plus a whole slew of other problems, like different map states, and 4 auto throw outs a day etc.

Unless they worked some magic with the linking system, to temp swap everyone onto the other time servers, dunno how that work since I’m not a ANet engine dev :p

But for all practical purposes, this means, EotM during all non-prime hours. Still kinda like the idea, but not certain if I would want to see it implemented. Sorry for babbling.

Gudradain.3892:

@Chinchilla Is one of the goal of your change to make WvW competitive? If yes, you are not going far enough.

Indeed, WvW cannot be competitive until you remove the 24/7 factor but you also need to remove the 4 maps factors as there are too many players required to make it controllable.

For WvW to become competitive, they need to make it guild based, on one single map for a period of 4 consecutives hours max and probably only on the night the guild is available to raid. So basically PvP but in a WvW setting.

My point is that what you are suggesting is such a big change that if they were doing it, I would prefer that they go all the way and make a real competitive mode to WvW.

Obviously, the casual mode will need to continue. So one competitive and one casual.

No, my intention is to make the game mode more engaging across all time zones. Could that lend to competition? More than likely, but right now, we have mismatches in coverage fighting other mismatches on the NA servers. If you played in T2, you would understand that JQ is Sea heavy ktraining and carrying that server up the ranks against servers that lack Sea time zone. Irony is JQ’s NA is abysmal compared to the T1 counterparts it gets compared to. Overall, I wanted to reduce running simulator/ktraining and give a better chance that people will engage other people. Instead of people transferring around looking for the “promised land.”

Furthermore, splitting between casual and competitive hurts the population growth for both (Eotm vs. “Normal WvW”). Eotm was intended to be for overflow, but it devolved to where lower levels go due to poor up-level scaling formula, and community disdain for said up-levels in “normal WvW” mainly due to said formula (amazing how circular that issue is…). Basically, the current Normal WvW has a barrier to entry problem with community,stats,tutorial,etc so newer people stick to EOTM, eluding to a division in Casual/Competitive WvW wouldn’t work for the current population, or be a long lasting solution. Not saying those are the only issues with Eotm.

Gudradain.3892:

But back on topic, the ranking system I’m suggesting is just to fix the glicko mess in the easiest possible way.

The thing is this, you cannot really discuss score changes without mentioning or being concerned for match making. Each one affects the other. It is better to view it as eHarmony (not sponsoring them intentionally). It is about making the best match ups based on similar qualities. Right now, TC doesn’t have the quality Sea timezone compared to BG, DB, and JQ. So you could say, TC is not marriage compatible with those three during that timezone. As already stated, JQ doesn’t have NA timezone to be successfully matched with TC, BG, or DB.

Imo, if you rectify match making first, scoring will be easier or will conform on it’s own. In the case of time zone segregation, you will drastically remove the effect of coverage. If you just try to fix the janky glicko ranking system, it is more of bandaid (I think WvW has enough of those).The bandaid will not necessarily give quality match ups across the board, nor is it a sustainable system, and especially for mass exodus of guilds who look for “better fights” while leaving said server in bad match up for a couple of weeks (do I need to give examples of that?). Such a thing has happened often for sea timezone (more than NA ironically), but I am sure NA people are more familiar with a certain “alliance.”

TLDR: Go back and just read the Bold

Remove server from the equation.

Players matter and what they do. As do guilds and informal alliances. You ever wonder why a game dating back to 2001 never had these issues surrounding servers? It didn’t because the focus was placed on players, guilds, and alliances.

Sure, they had population and wall clock coverage issues – any game with a durational score will. But they didn’t allow transfers and focused on the player. Ranks with structure and caps, abilities mapped, and of course weekly statistics by:

Player
Class
Guild
Alliance
Realm Ranks

This is getting so boring in that every other post is CD this, glicko that, I demand, about time population numbers, and other nonsense. We all agree it is broke but none of us are the developers or designers who can produce change.

Only show in town boys and girls.