Guild Wars 2

Chaba.5410:

I wonder if the new calculation considers the movement of population to different servers. Like, does it immediately remove a “WvW” player from a server’s total and add it to another server’s total when that player transfers? Or is it painfully slow to adjust to population movements like Glicko is?

Many have asked if EOTM is being taken into account.

That question needs to be answered if it hasn’t been officially – because not one of us in WvW believes that EOTM is WvW in any sense of the word.

Aerial Melodies.4938:

Jayne.9251:

Then maybe EoTM is factored into WvW population. Is yours played a ton?

I may be wrong, but I believe a dev said that EOTM does not factor into the calculations as it’s not considered part of the actual WvW matchup (if you go into EOTM and look at your inventory, it shows karma instead of badges).

If you can find that reference it would answer one question.

BUT then the question remains…

Why are low servers showing the SAME population “amounts” as what we KNOW are high servers like JQ…

This is what I saw today logging into world selection. Tier 1 Full Tier 2 full and high Tier 3 high medium Tier 4 Medium Tier 5 Medium Tier 6 Medium Tier 7 Medium Tier 8 Medium.

So according to the current population calculations tier 8 has the same amount of wvw activity as Tier 4 and into Tier 3. So really anybody in those tiers want to tell me if you can only get a group of 3-5 together on any boarderland most nights at primetime because thats what we got in tier 8.

Here is a a post from someone that a dev posted on reddit and clearly says EoTM isn’t counted towards wvw population.

https://forum-en.gw2archive.eu/forum/game/wuv/World-Population-Changes-Are-Coming/page/2#post5316590

Here’s what it said about EoTM:

No, EotM won’t be counted towards WvW. I don’t think it’s the same experience, and you can always enter it since it can have as many instances as needed, so there’s really not that same population problem as in “core” WvW.

As for JQ having massive numbers compared to full servers, the dev stated (and it’s been said in this thread too) that some servers were beyond full and would take some time before they even hit the cap of full, through attrition or what have you.

Michelangelo.1742:

From what I have seen in NA over the past week now is that Fort Aspenwood is at about the WvW activity level that divides Very High and Full. That meens Yak’s Bend and greater are heavily over-stacked.

Yes, I think this too. NA’s numbers show that rank 1-4 servers are heavily overstacked and only some sort of implosion of all four of those servers will ever make them come off Full. And HoT hasn’t even hit yet, when every casual logs into WvW. My guess is that those servers are going to be completely miserable with very high queues at the release of the new borderland map.

As for the divide between Full and Very High in T2, it is still difficult to say. Compare the EU numbers. There isn’t the mass-exodus from a tier in EU I think and the glicko ratings there are closer and more stable. So there’s some full servers losing to very high servers. Also, FA’s population is probably not close enough yet to SoS’s despite what the rating shows since SoS gains a lot of rating through TF’s night-capping including going after easier FA objectives. What I’m saying is that the off-hours population still is worth more in PPT rating than NA.

Chuck Zitto.2367:

This is what I saw today logging into world selection. Tier 1 Full Tier 2 full and high Tier 3 high medium Tier 4 Medium Tier 5 Medium Tier 6 Medium Tier 7 Medium Tier 8 Medium.

So according to the current population calculations tier 8 has the same amount of wvw activity as Tier 4 and into Tier 3. So really anybody in those tiers want to tell me if you can only get a group of 3-5 together on any boarderland most nights at primetime because thats what we got in tier 8.

Full-Very High-High-Medium-Low-Very low stand for certain thresholds. The system probably has a very low one for medium as its used the most over the different tiers.
You can’t compare servers because the system doesn’t compare, it just looks at the absolute numbers (I’d guess).
Thats why you see FSP flipping from full-very high and back again a few times the past few weeks because they’re hovering around that certain threshold.

They probably don’t want to include low from a more psychological point of view as people wouldn’t go their. Unless ofc they’d make it free to transfer to low servers.

Chaba.5410:

Michelangelo.1742:

From what I have seen in NA over the past week now is that Fort Aspenwood is at about the WvW activity level that divides Very High and Full. That meens Yak’s Bend and greater are heavily over-stacked.

Yes, I think this too. NA’s numbers show that rank 1-4 servers are heavily overstacked and only some sort of implosion of all four of those servers will ever make them come off Full. And HoT hasn’t even hit yet, when every casual logs into WvW. My guess is that those servers are going to be completely miserable with very high queues at the release of the new borderland map.

As for the divide between Full and Very High in T2, it is still difficult to say. Compare the EU numbers. There isn’t the mass-exodus from a tier in EU I think and the glicko ratings there are closer and more stable. So there’s some full servers losing to very high servers. Also, FA’s population is probably not close enough yet to SoS’s despite what the rating shows since SoS gains a lot of rating through TF’s night-capping including going after easier FA objectives. What I’m saying is that the off-hours population still is worth more in PPT rating than NA.

I am assuming that over 3/4 of the players that are returning will not be on the top 4 servers because, well, there are 20 other servers that they could have chosen. That is not to say that they will have higher queues, because they will of course. I think the real interest of HoT is how the combination of returning players, new players (who won’t be able to select those top 4 to 5 servers), and current PvE/PvP players will have on the populations and ratings of the lower tier servers during the first couple weeks of its release.

You make a good point that rating doesn’t have as significant an impact as we may have thought, especially considering the instance between SoS & FA and EB & DB. They must be correlated somehow, because otherwise their calculation system is further shrouded in mystery. They may keep tweaking it so for someone like me who likes looking at numbers and data, this is quite exciting ^^

Michelangelo.1742:

I am assuming that over 3/4 of the players that are returning will not be on the top 4 servers because, well, there are 20 other servers that they could have chosen.

Considering all the guilds and players that transferred to the T1 servers and then quit over the life of this game potentially returning to check out the new bl map… I cringe.

We could see that happening on former T1 servers too.

Chaba.5410:

Considering all the guilds and players that transferred to the T1 servers and then quit over the life of this game potentially returning to check out the new bl map… I cringe.

We could see that happening on former T1 servers too.

Very true, but that is likely a very small subet of the returning players, since I was thinking in terms of the game in general, not just old WvW players. Putting together what you are saying and what I am, T1 servers will definitely be changing to Overflowing instead of Full lol. It is quite cringeworthy to think about it that way, haha maybe Sanctum of Rall will suddenly find itself with crazy queues!

atheria.2837:

Why are low servers showing the SAME population “amounts” as what we KNOW are high servers like JQ…

What low servers?

Jayne.9251:

As for JQ having massive numbers compared to full servers, the dev stated (and it’s been said in this thread too) that some servers were beyond full and would take some time before they even hit the cap of full, through attrition or what have you.

JQ is not massively overstacked compared to other servers in the same range.

morrolan.9608:

What low servers?

JQ is not massively overstacked compared to other servers in the same range.

I think this confusion is pretty common, because we don’t even know the benchmark they set for what is full and is not full. They may have a “score” that relates to the level of activity they want on a full server. Technically a full server should be balanced with all the other full servers, because they have all attained the activity cap. However, since ArenaNet was very late in implementing this system and deemed the activity cap they desired was lower than the level T1 servers are currently playing at, there is some very large disparities between them. So that would explain why a server like Fort Aspenwood, which is quite measurably smaller than a T1 server (Glicko rating is almost 300 points below JQ), is full.

Hasn’t FA dropped to very high? I think its fair to say that FA is most probably understacked compared to the other full NA servers.

morrolan.9608:

Hasn’t FA dropped to very high? I think its fair to say that FA is most probably understacked compared to the other full NA servers.

There’s been a few brief flashes to Very High reportedly. It goes back to Full almost immediately.

It occurs to me that if FA is a test case for what the cap is then it may be set too low. How is its coverage across timezones? Looking at coverage wars may be slightly misleading due to YBs dominance.

I wonder why people think HOT will see a huge increase in wvw participation – aside from the new map, a lot of it is pve.

For every returning player is a Pvx headed to pve for the new shinies…… (Me included)

atheria.2837:

Chaba.5410:

I wonder if the new calculation considers the movement of population to different servers. Like, does it immediately remove a “WvW” player from a server’s total and add it to another server’s total when that player transfers? Or is it painfully slow to adjust to population movements like Glicko is?

Many have asked if EOTM is being taken into account.

That question needs to be answered if it hasn’t been officially – because not one of us in WvW believes that EOTM is WvW in any sense of the word.

that question has already been answered in the original post by the dev. No EotM is not counted.

morrolan.9608:

It occurs to me that if FA is a test case for what the cap is then it may be set too low. How is its coverage across timezones? Looking at coverage wars may be slightly misleading due to YBs dominance.

I can’t speak for NA/EU but we get maybe 15-25 people during OCX and maybe 10-25 during SEA depending on the day of week.

I wish FA had a force in NA prime… Would be refreshing to have a fight against normal numbers. Instead you have to bang on doors to get an entire server population to show up and fight 20 people.

Since T1 messed up YB and FA doesn’t seem to want to come off full… your only options to keep the game fun may very well be to beat Anet to the punch and transfer off.

Estimating about 170 players who played an estimated 2-3 hours daily no longer log into WvW on FA because they are no longer on the server. Are they still included in the population count?

Seems like once again the players have reacted quicker than what was anticipated by the devs.

Remember the chaos before…

End of free transfers.
WvW tournament season 1
WvW tournament season 2 (with free transfers)
WvW tournament season 3
Change in population calculation

Chaba.5410:

Estimating about 170 players who played an estimated 2-3 hours daily no longer log into WvW on FA because they are no longer on the server. Are they still included in the population count?

Estimating those players raided 5 days a week, that’s 2500 player-hours lost by FA.

Since the Population Calculation uses “active WvW players”, is an active WvW player measured in hours/day, hours/week, WvW map entries, or player-hours.

Really, the metric that should be used is player-hours and nothing else. Filter out the AFK players, although maybe you should also not count time spent in spawn so those that are spawn camped won’t be counted, since they’re really not active.

An answer would be nice, but I also want Anet to review their new algorithm in light of the player exodus from T2. Or at least I want be assured that Anet is revisiting it.

These are players who left completely, meaning they have absolutely no potential to add more hours to WvW activity and no one will replace their hours. IMHO their hours should be immediately averaged into the servers they transferred to and removed from the FA average. Both YB and SoS also lost a lot of players so there should be plenty of data.

As of writing, Piken Square (Very High) has the following queues on BL’s:

Jade Sea BL: 12
Piken BL: 2
Vizunah BL: 1
EB: 11

Now.. On Desolation our status is:
Gandara BL: No queue
Desolation BL: No queue
Far Shiverpeaks BL: Outmanned
EB: Outmanned (!)

So, how do ANet calculate World Population? Desolation has a decent night-crew, is it that making Desolation’s Status: Full?

I got a friend on Piken Square who just activated the game after 2 years of being inactive, him and I breathe WvW, and he can’t transfer to join me on Desolation which kinda sucks. We both have our whole former guild on this server so we’d both like to play on Desolation.

Edit: I posted this as my own thread (got merged here) and had hence not read any other replies.

Desolation has loads of Pve players (and to an extension PvX). It has been lead server in pve player since day one (apart from some of the language servers).

I think the system is flawed, in that even the people who just cap a centry for daily ap (could be a hell of a lot on Desolation cause of big PvX player base), is counted towards the cap. Cause indeed:

SFR: massively overstacked (please don’t try to deny it, I havea a lot of friends there, and they keep me posted, for instance, the lead ap farmer: almost all went to SFR for easy Eotm (always win) ap farming).
FSP: ‘borrowed’ a lot of desolation people, and other servers (including piken I think). They don’t have ap hunters, Eotm people, but the ‘veteran, we want wvw raids and win fights’ player base is the biggest amount EU, Pretty certain of that. (SFR is bigger cause Karma train/AP hunting sort of people).
Then comes desolation (third spot). If we purely count the Queues (that are next to non existant, low commander prescence, low raid prescence, then it doesn’t match the ‘full server status’ we own at the moment. But I admit we are close to it. My fear is old vets who rarely/barely/sometimes play are messing the system up, making it think of desolation ‘to high’.

Quit a bit behind all these servers are:

Kodash (temporary slacking? They used to be t1)
Gandara (used to be low, but growed steadely over time).
Piken square (most of time a medium server but in last tournement got 3rd spot, went down a bit, and now up again. Quite a bit below other listed servers I think (but a lot bigger then medium servers).

Jade Sea. Perhaps same player base as Kodash, but more ppt oriented (hello Old Vizunians), hence the higher Glicko.

Then come mostly Medium servers. It will be interested to see, how Elona reach does. (and to an extend Baruch bay and Vizunah square). CAuse they got the biggest Pve (from what I saw) populations among their language, always hanging in the full, even when all other server didn’t. But atm their performance in wvw is ‘normal’ (not excellent, neither weak). It’s interesting to see, how old veterans (that don’t play, or rarely play – but are a lot on those servers) are calculated into the formula by Anet.

Chaba.5410:

An answer would be nice, but I also want Anet to review their new algorithm in light of the player exodus from T2. Or at least I want be assured that Anet is revisiting it.

These are players who left completely, meaning they have absolutely no potential to add more hours to WvW activity and no one will replace their hours. IMHO their hours should be immediately averaged into the servers they transferred to and removed from the FA average. Both YB and SoS also lost a lot of players so there should be plenty of data.

A more problematic result of the slow reaction of FA bleeding players and still retaining it’s FULL status is that there will also be a slow reaction of a server receiving players and could be VERY HIGH for longer than it should.

A situation like what happened with YB could happen again. The highest VERY HIGH server is SoS and could get a spike of coordinated transfers before it becomes full. You could see the two servers that end up hopping between T1 and T2 combining and transferring to SoS to become a dominate T1 server.

Chaba.5410:

These are players who left completely, meaning they have absolutely no potential to add more hours to WvW activity and no one will replace their hours. IMHO their hours should be immediately averaged into the servers they transferred to and removed from the FA average.

Given that FA has gone from full to very high and back again I would say that is happening.

Swamurabi.7890:

A more problematic result of the slow reaction of FA bleeding players and still retaining it’s FULL status is that there will also be a slow reaction of a server receiving players and could be VERY HIGH for longer than it should.

Hasn’t FA slipped to very high since the system was changed?

server population status MUST be updated week-on-week basis and by taking the 7-day moving average , not 30-day or any long duration. 7-day is sweet spot…

Any transfer changes should be effected immediately. This isn’t like “blackouts” where people mass logoff for manipulating the algorithm.

azizul.8469:

server population status MUST be updated week-on-week basis and by taking the 7-day moving average , not 30-day or any long duration. 7-day is sweet spot…

Far more frequently than that, it appears. FSP went from full to very high to full again within a three day period.

fishball.7204:

Any transfer changes should be effected immediately. This isn’t like “blackouts” where people mass logoff for manipulating the algorithm.

Spot on, they must be taken out immediately and added to whatever server they transfer to.

morrolan.9608:

Hasn’t FA slipped to very high since the system was changed?

Only extremely briefly last week. I wonder if that was a result of golem week averages.

Swamurabi.7890:

A more problematic result … is that there will also be a slow reaction of a server receiving players and could be VERY HIGH for longer than it should.

The highest VERY HIGH server … could get a spike of coordinated transfers before it becomes full.

Sounds like us players are figuring out possible ways to game the new system. Anet better stay ahead of us. :O

Chaba.5410:

morrolan.9608:

Hasn’t FA slipped to very high since the system was changed?

Only extremely briefly last week. I wonder if that was a result of golem week averages.

Good point – hadn’t thought about that…..

Phoebe Ascension.8437:

Then comes desolation (third spot). If we purely count the Queues (that are next to non existant, low commander prescence, low raid prescence, then it doesn’t match the ‘full server status’ we own at the moment. But I admit we are close to it. My fear is old vets who rarely/barely/sometimes play are messing the system up, making it think of desolation ‘to high’.

And yet, Desolation has 50% more points in its matchup against FSP and Gandara and has pretty much already won (currently ticking 365).

Do you think it would be more balanced if Deso got more players? Balance is far more complicated than numbers or even player hours. Single players – commanders – can turn a complete mess into PPT. Try balancing them. Impossible. Just ONE well known commander taking a break for 2 days can be the loss of a matchup in the current high tier WvW climate.

If you describe the true Deso however… its obvious its “full”, because all you got left after pointing out low commander and raid presence is coverage.

Fsp sometimes on PURPOSE looses objectives, just to hang in the ‘nice fights’, or prolong them. FSP is the opposite of a PPT server. As a result, with similar (or more people) as Desolation, they perform worse. I also admit, that Desolation got an american guild (I think, not sure, but our night presence certainly improved). It’s nothing crazy, but this small amount of extra people, make a gazzilion difference to ppt. I’d say about 70% of people are PPT based on Deso and 30% Raid/GvG/fight oriented. FSP is 90% Fight based and 10% PPT.

Like FSP wins most fights against SFR. But they loose the matchup. Why? See above: more number, more coverage, more ppt oriented, etc. Doesn’t make FSP a smaller server with less Raid-minded people. They have the lead amount of EU in that, simple.

As for why Deso is so strong this week: We were since beginning in top 3 (sometimes top 5). But we won very rarely. Now that SFR is slacking, what do you think Desolatians prefer? To slack another weak. Nope to finally grab the victory. Once SFR (and other servers) stop slacking, Deso will be again on part (perhaps slightly in front of ) fsp again. The ‘blowout’ will decrease a lot. So again, this score says only half of the story.

Conclusion: Glicko says nothing about the number of ‘raid’-oriented people, just the PPT focus. It makes ‘very good servers’ look bad, cause they don’t focus on ppt, but win fights all the time.

Most Guilds on desolation are on a break or even disbanded, that’s why there’s little to no queues atm. I do not think T1-3 EU is even that active anymore.

Should they open up transfers right now?
I don’t think so, unless Deso had less players around PPTing (PvDooring @night), you might not even see it dropping in Population status, because there is probably players taking up slots right now; that are inactive.

Thankfully the restriction is there to stop bandwagoners.

I’m curious the relation between the VERY HIGH to FULL threshold and how that equates to players per map.

It seems that FA is getting closer to VERY HIGH and yet is not even close to getting queues on any map, usually only having players on two maps.

WvW NA will eventually resemble a lava lamp

There is a block of four overstacked servers that even a FULL server has no chance of competing with. Servers which are #5 and #6 are going to have morale problems as the #4 server, whichever of JQ/TC/BG/YB it is, plays the PPT game and pushes everyone off the map. Either the players from the losing servers will transfer, or sit out until their glicko rating drops, which could take weeks.

This will continue to unbalance T2 and T3 for some time because the only way the block of four lose players is by attrition, which could take years, maybe longer than it takes to get the yakslapper title.

In order to prevent even more overstacking by players returning with the release of HoT, I’d suggest to Anet that any returning player who has been inactive in WvW for more than 6 months be forced to choose a new server if their old server is FULL.

Swamurabi.7890:

WvW NA will eventually resemble a lava lamp

Since the new algorithm has been described as smoothing out “time artifacts” such as timezones, my guess is that EU prime times are going to end up the place to be for the large scale WvW battles that long time players are used to compared with NA prime times. What I mean is that NA prime on NA servers now has a higher competition with OCX and SEA primes for “WvW activity” than EU prime does on EU servers. Did what I describe make sense?

Chaba.5410:

Swamurabi.7890:

WvW NA will eventually resemble a lava lamp

Since the new algorithm has been described as smoothing out “time artifacts” such as timezones, my guess is that EU prime times are going to end up the place to be for the large scale WvW battles that long time players are used to compared with NA prime times. What I mean is that NA prime on NA servers now has a higher competition with OCX and SEA primes for “WvW activity” than EU prime does on EU servers. Did what I describe make sense?

You’ll have to find where it was said that the algorithm will smooth out timezones, because what I read from Samual Loretan was

The goal of this change is to make the population spread out better, which will in turn make the ratings get closer, which will result in both more interesting matchups, and more variability as well. This will take some time to happen.

I don’t equate spreading out the population with balancing the population in each timezone. I think the WvW players are reading more into what was said.

From here: https://www.reddit.com/r/Guildwars2/comments/3eh64z/changes_coming_to_transfer_cost_calculations/

"
[–]anet_samuel 39 points 1 month ago

The new method isn’t based on players currently playing or logged in. It looks at the active WvW population over a period of time.

permalinksaveparentreportgive goldreply

[–]Lksaarlocal wvw backliner 5 points 1 month ago

Is thatperiod of time timezone agnostic (a whole day?) or even matchup agnostic (more than a week?)

permalinksaveparentreportgive goldreply

[–]anet_samuel 44 points 1 month ago

It looks at a period long enough to smooth out rapidly changing artifacts like timezones or matchups.
"

I’m interpreting that to mean the range on the rolling average is longer than a single match, maybe even two matches.

To me that sounds like an effect of the algorithm would smooth out a single server’s timezones.

I’m reading that they’re using a moving average that’s at least greater than 7 days (matchup), maybe a 15 or 30 day moving average.

This would mean that fluctuations in server population over a timezone will not have much affect on the population level of the server. It would also smooth out spikes in server population.

What it doesn’t mean is that if you are on a server that needs SEA players that the next player that transfers to your server will be from SEA prime.

Swamurabi.7890:

It would also smooth out spikes in server population.

Yes, exactly.

EU servers don’t have non-EU Prime population spikes like NA servers have non-NA Prime population spikes. They have some off-hours, but not like NA servers have. The NA population on NA servers getting smoothed out with OCX/SEA population would lead to lower scale WvW during NA prime than what EU servers would experience (and what higher tier NA servers are used to.)

Is there a specific time (like once a week) when the server population update takes place? And when a server hits “very high” how long can it take before it bounces back to full? Sorry if it’s a stupid question :P

And I wonder if hardcapping servers is the right way to go. Surely you get possibly movement from full or very high servers to lesser populated servers. But that won’t solve the unbalanced problem at all I think.

If I take myself as an example, right now I am on a full server (desolation) and i’m active during euro primetime which is also the busiest time on deso. I am waiting until I can transfer to one of the NA T1 servers (got friends on NA and t1 is the only place with euro presense in WvW). But obviously I can’t, nor will I pick a server where there is no presence during EU timezone (which is everything pretty much below t1). So right now, I am staying on a full server as a dead weight in a full timezone while I can move away to another tier 1 server where the EU timezone is not full.

In the meanwhile, movement from guilds active during a servers primetime, for example euro guilds on EU servers and NA guilds on NA servers don’t really change much to the balance in primetime, nor will WvW guilds move to a dead WvW server.

I can imagine that the current situation keeps a lot of people from making a move in the first place instead of moving to lower tiers. Which results in very little changes to population right now and possible huge overstacking servers when players come back for HoT.

edit: If anet really wanted to do something about balancing servers (and unstacking the full ones), they should start out by lowering the costs to transfer and even lower it to 0 gems for the medium sized ones and below (if that exists). Very high should stay at 1800 gems or maybe make that even higher to prevent servers to get full too quickly again.

Chaba.5410:

Swamurabi.7890:

It would also smooth out spikes in server population.

Yes, exactly.

EU servers don’t have non-EU Prime population spikes like NA servers have non-NA Prime population spikes. They have some off-hours, but not like NA servers have. The NA population on NA servers getting smoothed out with OCX/SEA population would lead to lower scale WvW during NA prime than what EU servers would experience (and what higher tier NA servers are used to.)

Sure they do.

Some servers, at least in gold league and some in silver, have day crews that start hours before prime. And a few more have night crews. You can get prime-time blobs of 35+ during those times, depends on the server.

Jayne.9251:

Chaba.5410:

Swamurabi.7890:

It would also smooth out spikes in server population.

Yes, exactly.

EU servers don’t have non-EU Prime population spikes like NA servers have non-NA Prime population spikes. They have some off-hours, but not like NA servers have. The NA population on NA servers getting smoothed out with OCX/SEA population would lead to lower scale WvW during NA prime than what EU servers would experience (and what higher tier NA servers are used to.)

Sure they do.

Some servers, at least in gold league and some in silver, have day crews that start hours before prime. And a few more have night crews. You can get prime-time blobs of 35+ during those times.

The non-EU prime spikes in EU aren’t anywhere near what is seen in NA.

Dunno man, I keep seeing more and more high ppt ticking in off-hours as more and more NA players venture over to EU. It used to be only one or two servers had “day” or “night” crews .. now it seems like at least 10 have them. The good news is at least they’re not all congested on one or two servers and appear to have followed the “norm” for EU and spread out.