So the lesson learned from these simple graphs is:
Stop letting people transfer to servers with ridiculous WvW queue times, especially for free. Allow guilds to transfer to lower populated servers and keep their guild perks / upgrades and influence.
I’d actually be in favor of ANet blocking server transfers completely, unless you’re transferring to a lesser populated server. 24 hours isn’t nearly enough to stop people from server hopping.
I’ve never seen a MMO allow this many server transfers this often. Completely ruins what little point there is to WvW in the first place.
It shows without a doubt that there is an off-peak capping issue. Dragonbrand has huge spikes in queues, when Tarnished Coast has none.
Hopefully this data will spur ArenaNet to do something about the issue.
I think that (hopefully) everyone realizes that there are issues, but to a degree these issues are to be expected in WvWvW PvP. Many problems arise when developers attempt to add varying rules and structures to “world PvP” and I’m sure that the struggle to find a reasonable middle ground between unregulated PvP (which some prefer) and a system maintaining the possibility of competition (which others prefer) must be difficult.
I think that some great suggestions were made so far in this thread, ideas like an underdog achievement offer rewards for players who stick it out and fight for their server rather than just evacuating. It is surprising to see players complaining about free server transfers, but it serves as a great example of the difficulty of pleasing everyone. I’m not sure that barring people from joining the “obviously going to win servers” is a good idea, it feels pretty knee-jerky to me. The idea that these servers are being flooded in part due to their WvW dominance though definitely deserves consideration, ultimately how much should devs be responsible for regulating player behaviors?
No surprises here. While on Maguuma during peak times we are very competitive and so is TC. Usually we hover around 200 to 240 per team in the matchup. Once off-peak hits it’s a total steamroll with Dragonbrand. The graphs just confirm exactly what I experience.
We have people in Maguuma shouting at others in the map saying “Get your butts over to such and such, they are attacking! WHY ONLY 5 of us defending?!” Why? Because there are probably only 15 of us in the entire map and 5 of those are doing a stupid puzzle.
I agree that it could help a bit to change the matchup priority to “queue overlap” followed by points. Obviously everyone has to be matched up to someone, but I’d personally rather be getting dominated by a server with similar populations logged in than dominated by a server while most of our population is sleeping/at work.
Maybe it could also help if somewhere from server selection you could see a link to these graphs so new players can see where the shortest queues will be for the playtime they’re usually going to be in.? At this point so many people have already found their home that it might be difficult.
Also agree with potentially letting guilds transfer with perks ONLY to servers with less queue times for a certain period if it’s feasible.
You know what ALL those charts have in common? The queues got smaller and smaller, to non existent on most servers. Even on HOD the queues seemed to be about half what they were at the start of the match up midway through it . SBI went from having massive queues a week ago to having none at all. I didnt need a chart to tell me that though. I expect once the ones for the 19th onward come out they will dip down again. I am sure if there are major match up differences there might be a spike but I would bet money that if people are facing the same opponents this week they faced last week the downward spiral will continue. The only server that remained fairly constant was Chrystal Desert. All other saw a major dip as the week progressed.
So one could conclude that people have left the game or they just arent bothering to queue up for WvW, or maybe a little bit of both.
But the obvious is still obvious, looking at server match ups and overlapping the queues and the ties and it doesnt take a rocket scientist to figure it out.
But to me the most telling sign is that WvW ‘popularity’ spiraled downward Since Last Saturday. Like I said will be very interesting to see what the 19,20, and 21 numbers are.
You know what ALL those charts have in common? The queues got smaller and smaller, to non existent on most servers. Even on HOD the queues seemed to be about half what they were at the start of the match up midway through it . SBI went from having massive queues a week ago to having none at all. I didnt need a chart to tell me that though. I expect once the ones for the 19th onward come out they will dip down again. I am sure if there are major match up differences there might be a spike but I would bet money that if people are facing the same opponents this week they faced last week the downward spiral will continue. The only server that remained fairly constant was Chrystal Desert. All other saw a major dip as the week progressed.
So one could conclude that people have left the game or they just arent bothering to queue up for WvW, or maybe a little bit of both.
But the obvious is still obvious, looking at server match ups and overlapping the queues and the ties and it doesnt take a rocket scientist to figure it out.
But to me the most telling sign is that WvW ‘popularity’ spiraled downward Since Last Saturday. Like I said will be very interesting to see what the 19,20, and 21 numbers are.
Or it’s a function of the graph starting on a Friday, and moving to the right into the work week. Seems self-explanatory that there will shorter queues after the weekend.
The more interesting statistic would be those servers who experience less of a drop-off. That would indicate there is an “off-work” capping problem (students, under-employed, unemployed) as well as an off-peak one.
Thrumdi, Captain of The Tarnished Coastguard
—
The ultimate GW2 troll.
You know what ALL those charts have in common? The queues got smaller and smaller, to non existent on most servers. Even on HOD the queues seemed to be about half what they were at the start of the match up midway through it . SBI went from having massive queues a week ago to having none at all. I didnt need a chart to tell me that though. I expect once the ones for the 19th onward come out they will dip down again. I am sure if there are major match up differences there might be a spike but I would bet money that if people are facing the same opponents this week they faced last week the downward spiral will continue. The only server that remained fairly constant was Chrystal Desert. All other saw a major dip as the week progressed.
So one could conclude that people have left the game or they just arent bothering to queue up for WvW, or maybe a little bit of both.
But the obvious is still obvious, looking at server match ups and overlapping the queues and the ties and it doesnt take a rocket scientist to figure it out.
But to me the most telling sign is that WvW ‘popularity’ spiraled downward Since Last Saturday. Like I said will be very interesting to see what the 19,20, and 21 numbers are.
I’d attribute a bit of that to just people being busier during the week and having less incentive to join up when they see their side dominating the score or being dominated, not necessarily that they quit the game. Also Friday and Saturday will always be the busiest times and will see significant queues no matter the matchups.
I was looking at peak hours, the whole ‘work week’ argument doesnt fly. Peak hours happen after people get home from work. So there should be a steady ‘mountain peak’ effect each day. Look at Crystal desert for an example.
Every server had those peaks, but as the week went on the peaks got smaller and smaller and in some cases completely disappeared. Weekend should actually be a plateauing effect where it stays pretty flat since most people arent working and are trying to play. So the mt peaks are a little less dramatic.
dont worry about it bro, give it 4 days and everyone is back to WoW.
I mean my whole guild hate wow and are so fed up with it, but this game is worse than any fail i ever seen (even swtor) that they actually are going back.
well done on the “NO REWARD but pretty gear” system
well done on the "No Tank " Fail
WELL DONE on the “No raids”
and a big well done on WvWvW.
you should ask for a share of blizzard and bio for making their game seem so epic in comparison.
P.S Mods, before you delete this comment like you did with my other 10, THIS IS CONSTRUCTIVE criticism so your developers wake up and fix the game before its too late, but if you keep on deleting these comments, then you will be left with nothing but your fanboys telling you how wonderful this fail game is
Actually that’s not constructive at all. Honestly, can you point out one thing in that post that was constructive? It’s just a bunch of sarcastic “well dones.” What should be done differently? The complete opposite of everything you listed? I for one love that spvp is completely balanced gear. Also really enjoy wvwvw, even though there are clearly nightcapping problems that I’m sure they’re looking into as well.
Basically, you apparently have no idea what constructive criticism even is. Later (but not really because you’ll still be playing this game for quite a while I’m sure and then use it as an example of something that’s better than the next game you play).
I was looking at peak hours, the whole ‘work week’ argument doesnt fly. Peak hours happen after people get home from work. So there should be a steady ‘mountain peak’ effect each day. Look at Crystal desert for an example.
Every server had those peaks, but as the week went on the peaks got smaller and smaller and in some cases completely disappeared. Weekend should actually be a plateauing effect where it stays pretty flat since most people arent working and are trying to play. So the mt peaks are a little less dramatic.
People do other things besides “go to work” and “play GW2”. I’d wager the vast majority of people with GW2 spend their time doing other things in addition to playing GW2 every waking and/or available moment. Like eating, sleeping, seeing to personal hygiene, leaving the house, socializing, dealing with household upkeep, and pursuing other recreational activities. You can’t assume such a simplistic model and then use the absence of a pattern from that model as evidence for anything.
[Edit: Removed response to a moderator deleted post]
These statistics do display a disturbing trend that should be sorted out sooner rather than later. Matchmaking algorithms using a combination of scoring and overall activity should remedy the issue; perhaps with more weighting preference given to activity, rather than score, as the scores from this week are clearly skewed and likely not representative of overall server performance/skill level. If this continues into the two week match-ups, WvW will quickly lose popularity, or so I assume given the current state of the forums and the morale on my own server.
[Edit: Removed response to a moderator deleted post]
These statistics do display a disturbing trend that should be sorted out sooner rather than later. Matchmaking algorithms using a combination of scoring and overall activity should remedy the issue; perhaps with more weighting preference given to activity, rather than score, as the scores from this week are clearly skewed and likely not representative of overall server performance/skill level. If this continues into the two week match-ups, WvW will quickly lose popularity, or so I assume given the current state of the forums and the morale on my own server.
I’d argue that score is a very good indicator (right now) of server participation. We aren’t at the point yet where things are so organized that skill is more important. Every time I login at prime time I see that the 3 sides in the match are roughly even in scoring.
[Edit: Removed response to a moderator deleted post]
These statistics do display a disturbing trend that should be sorted out sooner rather than later. Matchmaking algorithms using a combination of scoring and overall activity should remedy the issue; perhaps with more weighting preference given to activity, rather than score, as the scores from this week are clearly skewed and likely not representative of overall server performance/skill level. If this continues into the two week match-ups, WvW will quickly lose popularity, or so I assume given the current state of the forums and the morale on my own server.
I’d argue that score is a very good indicator (right now) of server participation. We aren’t at the point yet where things are so organized that skill is more important. Every time I login at prime time I see that the 3 sides in the match are roughly even in scoring.
Even on servers where the score is 400k – 150k – 150k? The overall score is the one being used in matchmaking to my knowledge; not the amount coming from controlled points each interval during prime-time. If one server holds the entire map overnight, score is not a good method of matching servers; this is especially the case when metrics like the ones that are provided here are available. One could argue that score is a derived method of gauging participation, which is what I presume you are doing. But why use a derived method when the actual raw data is being collected and is more accurate than a number that compounds every 15 minutes?
Well that explains Fort Aspenwood, They have nearly 24/7 people playing.
…which is interesting, since Dragonbrand do as well. Yet both TC and Maguuma are between those two in the rankings and neither have a strong off-peak presence.
…which is interesting, since Dragonbrand do as well. Yet both TC and Maguuma are between those two in the rankings and neither have a strong off-peak presence.
One might expect it is because the two servers still put up a competent fight? I’d almost think that the rating is partially measured by the relative rates of gains and a few other factors besides mere map domination.
Also have to look at who those servers are playing as well? If a server that has almost 24/7 participation is against servers with similar numbers you would expect everyone’s scores to be closer. Consider 2 groups of wvwvw below (extreme examples)
A: 24/7 participation
B: Hardly any “night defense”
C: Hardly any “night defense”
You would expect server A to dominate and have a ton of points, whereas servers D and E, splitting the potential points a bit more actually end up with a lower score. If this is how the rankings are based right now…purely by score (I don’t know)…you can see how this would be a problem.
Even 400k – 150k – 150k is a good indication of participation. It’s not an absolute indicator, but to me that says one server has roughly 2.5 times the presence as the others during a given day. Just taking a look at the queue graphs one can see this trend on the servers dominating the matchups — the winning ones usually have a queue or near queue at most times of the day whereas the losing ones only have a fraction of a day with a queue.
Of course, queue numbers don’t tell us exactly the number present as you could have 90% participation with no queue but to me the score is the one that really tells the story. You would expect, due to “pubbie zergs”, that players don’t like being slaughtered and will naturally stick together when they see they are being effective. If there are enough players, there should be a strong enough presence to capture something even if they are doing rotating captures.
NA server queues (Red)
EU server queues (Blue)
Merged NA + EU server queues
(everything scaled to VS’s scale since it was the one with the highest number queued)
That’s interesting, so EU servers have higher activity peaks during their prime time than NA servers; but NA servers have more consistent activity all hours of the day than EU servers do. I suppose it makes sense, as Oceanic players probably play on NA servers for less latency.
Edit: Word Choice changed for clarity; removed inaccurate interpretation.
There is a definite trend that over the course of the week less and less people queue for WvW. Now, is that a function of the graphs starting on the weekend and moving out into the week? Or is it a sign of WvW fatigue / discouragement.
What are the chances of getting the battleground population graphs for each battlegroup? Although, I guess for some it’d be pretty boring – a straight line at full capacity.
Could you elaborate at how that 3% rate for those that logged in and queued that never got in was derived? Does this include those that logged in, queued and then logged out due to the inability to get into WvW due the extraordinarily long queue duration? If it it doesn’t, there needs to be some way to account for those in the numbers…perhaps if they exceeded a certain elasped queue time. I think the 3% rate that is cited is not showing the whole picture regarding the state of people attempting to get into WvW.
So the lesson learned from these simple graphs is:
Stop letting people transfer to servers with ridiculous WvW queue times, especially for free. Allow guilds to transfer to lower populated servers and keep their guild perks / upgrades and influence.
I’d actually be in favor of ANet blocking server transfers completely, unless you’re transferring to a lesser populated server. 24 hours isn’t nearly enough to stop people from server hopping.
I’ve never seen a MMO allow this many server transfers this often. Completely ruins what little point there is to WvW in the first place.
This is a very well stated post. The only servers that should be allowable transfer targets are those that are not high/full servers in peak time. The other servers should be off limits for the time being. Server populations won’t distribute if you you allow the crowded, top rated servers to be a transfer destination.
You can see from the That score determines the server’s When You can not give “queues” on all maps, ie fill the WVW on all 4 maps are Behind Those who can, fill the whole WVW. That is the line Reflects the flow of people in WVW, the map above can support Which, When it is full, is because the map is “full”.
There are many people on servers with a lot of scoring in WVW, and few people on other servers That can not close all 4 maps (EB – Eternal Battlergrounds RB – Borderlands BB Red Team – Blue Team GB Borderlands – Borderlands Team Green), have guild’s who are Transferring from servers with a lower score and the queue for the scoring better and a larger queue only worsens the queue, Which was bad and worsening Already the situation hard, low servers with the server’s low score, giving the trend of more and increasingly worse in queues TOP servers and servers with low scores with few playes and competitiveness increasingly worse.
Sorry, google translate is bad, if anyone has a better English and be able to explain what I’d like, I’d appreciate it.
I appreciate the information but something is definitely hosed with the queues. I queued tonight and then 5 minutes later 2 friends did, they got in about 5 minutes later while mine did not pop. I waited… 2 other people that queued AFTER me got in. I relog and requeue and 1 other guy gets in before me and he queued AFTER I re-queued.
I left it in queue for a hour and nothing happened. 5+ guildies got in the same zone I was queued for even tho they queued AFTER I did.
I would try and give you details but seeing as the queue mechanism is game is about as vague as you can possibly make it (you are in queue, yes we promise!) I cannot say what is wrong. All I know is, I was/am bored waiting for the darn thing to pop.
This data is incredible to study. Even without the graphs, TBH had already decided to move back to NSP and forget ET. The queues were way too long, and it was just too hard to get in to accomplish anything meaningful as a guild. Since Tuesday (our switch) we have been able to get 30 or more into a borderlands at once. It’s been so much fun to deal with 15 min queues max during peak NA hours.
Just a side note, if you’re looking to transfer, Devona’s and Kaineng could use the pop. Get away from the queues and go have fun in WvW. Forget the scoreboard, enjoy yourself and have fun with your friends and guildmates. Log out every night feeling like you accomplished something meaningful!
MasterofKhaos
WvW Leader and Co-leader of [TBH]
Co-Founder of Northern Shiverpeaks Council [NSC]
That aside there are a few key items the charts point out:
First off the zone has to be full in order to queue OR if the populations are balanced to the smallest population. — Please clarify how populations are controlled in WvW.
Given this we can answer the following questions:
On a given server do people tend to queue for a specific zone first (EB for example) and then overflow to other zones.
We can see the overall growth of the queues looking at WvW as a whole by adding them together.
We can see how long a specific “prime time” and “off time” cycle is on a server.
We can see the ramp up and ramp down rate
R\S-PLUS time:
Plugging in final scores in regression analysis we can see that servers that queue for EB correlates to higher ranking more then queueing for a borderland.
The wider the peaktime window correlates to higher scores.
*The narrower the offpeak window correlates to higher scores. (Which reinforces the prior)
Queue length itself doesn’t appear to strongly correlate to higher scores
Average queue length in EB as a stronger correlation to higher scores then average borderland queue lengths.
I had to guesstimate as best I could the actual data points on the graph and took two sets of random samples of graphs 6 each. Please post raw data for us in a CSV please!
Without the raw data there is some uncertainty in the accuracy since I didn’t have time to run all the graphs. However I can draw a conclusion with a certain confidence: The activity in EB, based on the graphs, is the ‘best’ indicator of serious WvW activity. This could be the result of high scoring servers locking up borderlands and then heading to EB. It could be the reverse.
“No Mercy for the Panda Kids. We don’t want you. We don’t need you.”
kitten kitten kitten kitten kitten kitten kitten kitten kitten kitten kitten kitten kitten kitten
Because censorship is the most important part of the MMO business.
If off peak flipping an issue? Not really as far as I can tell, it’s more likely then not that high score servers have a wider “peak time” window (thus shorter off-peak window) is impacting scoring then opposing off-peak activity. It sounds like arguing semantics but the math folks understand. The ramp up and ramp down I would assume is a significant indicator also but it is way too late in the evening to try and test that.
One key metric would be the queue size after the reset which is roughly the first 5 datapoints (you can ignore all the rest actually).
The queue counts here have substantially more impact on scores it appears then any of the other datapoints. In short: There is a much stronger correlation between the first 5 queue counts to score then the first 5 queue counts of any other prime time.
In short: Higher scoring servers bring their A game when the matchups first begin.
This last part tells a very important factor: There may be a fundamental problem with turning around a WvW match when there is early domination.
Could we get, for that same time period, a chart of their score at that sampling time?
“No Mercy for the Panda Kids. We don’t want you. We don’t need you.”
*Necrod
First off JQ is obviously the new HoD. lol…
Please post charts like this every week A-Net so people know where to transfer if they don’t want a queue! It will especially help the new players out that are coming into GW2 and really aren’t enjoying their servers WvW. If this is posted, they would know where they are most needed. Just a thought
Commander [ICoa] Hollywood Fiend
Primal Emperor of Imperial Coalition
imperialcoalition.enjin.com
Also, some updates on actual WvW capacity would be welcome
Single-handed Mexican-Vietnamese mercenary. Now fighting for Piken Square, tomorrow, who knows?
Tons of chars. If only one day I manage to guess how to play any of them, I will post it here
I thought this was going to be something new, I was so excited to see that they were actually looking into the major WvW discrepencies and showing us the data.
Even if they provided up to date information on queue size, it means nothing in regards to whether or not a bug really exists in how it is determined. In fact, given what we all have been experiencing, I would say it is quite likely.
On a side note, for months we have been asking to be able to see our spot in a queue so we can determine how long we have to wait instead of staring at the screen for an hour or more rather than getting involved with something else. Displaying that to the user is not a big deal. This fact along with our in game experiences pointing to there being no real order to the queue system, makes me firmly believe that there code is extremely flawed to the point that they can’t even make sense if it.
I would take anything they post on the subject with a grain of salt, because they have proven that they are unable to handle something as simple as this.
If you seriously don’t want to deal with que’s anymore, go find a server that is down over 200k points right now. You won’t have to que…you can be one of the 5-8 people that dare WvW, and can’t even get enough to get a siege breaker commanders going while there are hundreds of enemies standing just out of range of your legendary defenders waiting for something to kill.
I would trade my uncompetitive server with 0 ques for 3-5 hour ques on a competitive server any day of the week.
Also will be good to know what WvW cap is for now as we noticed is smaller like before.
EU servers have no queues or little by all day but heavy during prime time because most players longing at evenings and EU doasnt have big differences between time zones across server +- 2h
If you seriously don’t want to deal with que’s anymore, go find a server that is down over 200k points right now. You won’t have to que…you can be one of the 5-8 people that dare WvW, and can’t even get enough to get a siege breaker commanders going while there are hundreds of enemies standing just out of range of your legendary defenders waiting for something to kill.
I would trade my uncompetitive server with 0 ques for 3-5 hour ques on a competitive server any day of the week.
DB is down 213k right now and we had a queue last night. It’s been on and off queue this week for us and most of the regulars I normally see aren’t even logged on.
These are old and they do not represent the actual data anymore… I doubt such a bandwagon like Kaineng still have 0 queues while “dead” servers like IoJ and SBI still have a lot. I wish they get data from THIS week (or last week, when the free transfers ended and people settlet) and show us after this is over.
