World Restructuring Status Update

[Arya Whitefire.8423]

9 hours ago, Cyninja.2954 said:

 

A system that adjusts for playtime fluctuations can only do so with correct play times. Activities which alter the play times after players have been sorted undermine this.

 

The problem is the system not the players.  You can't make people play when they don't want to.  Far better to give up on the illusion of a week long matchup, and get some shorter duration auto balancing matchups going so the game is never in a state of 80 vs 30 vs 5 .....

 

[Cyninja.2954]

11 hours ago, Chaba.5410 said:

Please do go on and explain what "correct play times" are.  You're almost there...

A player has 2 accounts for simplicities sake. Account A, his main, and account B his alt. Player A plays account A for 20 hours per week on average for the last 8 weeks, he plays his alt accounts 5 hours on average over the same time period.

The matchmaking now assigns said player to a shard, valuing his account A with around 20 hours of play time per week, his alt account with 5.

The player now switches to his alt account after the new shard creation. He now plays 20 hours on his alt account per week and 5 hours on his main.

Total play time for the player has not changed at all, but the 2 shards his accounts are assigned to see very different active play times.

No algorithm can adapt to such behavior. The "play times" used for creating the shards in not "correct" any longer.

Edited July 7 by Cyninja.2954

[Chief.7143]

All i can say is, with so much time to plan this, and so many problems already, I believe this is GG for WvW. Not being dramatic but they have shown me two things with this release:

1. They don't have any solutions to the issues in WvW, so they outsourced it to the players which well.. will never go well.

2. They have no good data points to make matchups and are more concerned with Friday night Que's than everyday WvW Qol (addeda  tier and made the game dead outside of prime times). Friday reset which = skill lag all night anyway and is usually not a great experience cause the engine cant handle 3 teams fighting.

All in all, they are probably understaffed for WvW and the staff they have is concerned with what I would consider the wrong things.

Sure people will still play but honestly it will just slowly die now. Our link was placed in t1(t2 now), and has 3 active tags. 1 doesnt even use discord, 1 tags 3-4 days a week for about ~2 hours at a time and the other tags ALOT and is great but carries far too much of the burden. On the flip side, an unnamed server (cause I think that is not allowed) which seems to be a collection of people from an unnamed server started in I think t5, and has round the clock populations just griefing with cheese builds and clouds.

[Chaba.5410]

43 minutes ago, Cyninja.2954 said:

A player has 2 accounts for multiplicities sake. Account A, his main, and account B his alt. Player A plays account A for 20 hours per week on average for the last 8 weeks, he plays his alt accounts 5 hours on average over the same time period.

The matchmaking now assigns said player to a shard, valuing his account A with around 20 hours of play time per week, his alt account with 5.

The player now switches to his alt account after the new shard creation. He now plays 20 hours on his alt account per week and 5 hours on his main.

Total play time for the player has not changed at all, but the 2 shards his accounts are assigned to see very different active play times.

No algorithm can adapt to such behavior. The "play times" used for creating the shards in not "correct" any longer.

Why does a player's total playtime prioritize above their account playtime?

Do you want all 25 hours credited to the main account when that account may only get played for 5 hours?  How is that any different from the type of imbalance you think is being caused by playing alts?
 

Additionally, those hours are rolling averages smoothed out over weeks.  That's how the algorithm adapts.  It's been adapting that way ever since population calculation for servers was changed.

Edited July 7 by Chaba.5410

[Cyninja.2954]

43 minutes ago, Chaba.5410 said:

Why does a player's total playtime prioritize above their account playtime?

Because account play time is absolutely irrelevant when it comes to current/recent play time in WvW. If a player played thousands of hours of WvW in the past but is inactive currently, he is of little value to a newly created shard.

Unless you mean account play time in WvW in recent months, which I am addressing later.

Quote

Do you want all 25 hours credited to the main account when that account may only get played for 5 hours?  How is that any different from the type of imbalance you think is being caused by playing alts?

What are you talking about? For all we know shards are created with looking at total play time over the last week/weeks. Of course the main account gets the full 25 hours attributed when it is placed in a shard and if the player now plays less, that shard will be missing play hours. You are questioning me not understanding the system?

I think you are under the false assumption that play time is being averaged out over extended periods of time, which I highly doubt is happening because the behavior in the previous as well as current algorithm matchmaking does not support this assumption.

Quote

Additionally, those hours are rolling averages smoothed out over weeks.  That's how the algorithm adapts.  It's been adapting that way ever since population calculation for servers was changed.

No, this is not how it has worked, at least not in the past. If it did players would not have been able to artificially reduce their server size shortly before resets (aka play less 1 week before a relink to open up the server). For all we know, the most recent week (at best match-up aka relink) is ever been accounted for and that would already be far to short when it comes to assigning a value to an account.

Now technically IF the WR system does average out the play time across multiple months, the effect would be mitigated somewhat (even if not entirely removed) but we have no basis to believe this to be the case yet.

Notice that accounting for play time in the last match-up does NOTHING to alleviate this issue. There would need to be averages of data for multiple match-ups to significantly impact this calculation, likely not less than 1 year. THAT most certainly is NOT the case currently.

Even then the issue is just being managed and reduced, it does not go away. Averaging out play time for an account only makes the outliers in performance for that account be less extreme, it does not negate or remove the active undermining of the matchmaking system.

So your best case scenario is essentially: we are managing players manipulating the matchmaker in their personal favor to mitigate the negative effect on other players.

Edited July 7 by Cyninja.2954

[One more for the road.8950]

16 minutes ago, Cyninja.2954 said:

Because account play time is absolutely irrelevant when it comes to current/recent play time in WvW. If a player played thousands of hours of WvW in the past but is inactive currently, he is of little value to a newly created shard.

Unless you mean account play time in WvW in recent months, which I am addressing later.

What are you talking about? For all we know shards are created with looking at total play time over the last week/weeks. Of course the main account gets the full 25 hours attributed when it is placed in a shard and if the player now plays less, that shard will be missing play hours. You are questioning me not understanding the system?

I think you are under the false assumption that play time is being averaged out over extended periods of time, which I highly doubt is happening because the behavior in the previous as well as current algorithm matchmaking does not support this assumption.

No, this is not how it has worked, at least not in the past. If it did players would not have been able to artificially reduce their server size shortly before resets (aka play less 1 week before a relink to open up the server). For all we know, the most recent week (at best match-up aka relink) is ever been accounted for and that would already be far to short when it comes to assigning a value to an account.

Now technically IF the WR system does average out the play time across multiple months, the effect would be mitigated somewhat (even if not entirely removed) but we have no basis to believe this to be the case yet.

Notice that accounting for play time in the last match-up does NOTHING to alleviate this issue. There would need to be averages of data for multiple match-ups to significantly impact this calculation, likely not less than 1 year. THAT most certainly is NOT the case currently.

Even then the issue is just being managed and reduced, it does not go away. Averaging out play time for an account only makes the outliers in performance for that account be less extreme, it does not negate or remove the active undermining of the matchmaking system.

So your best case scenario is essentially: we are managing players manipulating the matchmaker in their personal favor to mitigate the negative effect on other players.

You do realise it doesn't count all your WvW time on the account? I am not sure what you are trying to get to here.

Activity is measured over something like the last two months, iirc. If you stay out of WvW long enough you don't count. And coming back, it's just starting counting hours from then. Earlier activity has nothing to do with it. People have indeed been able to artificially reduce their servers activity before, which is why they extended the length of measure to what I seem to remember is a couple of months.

Why would an account be credited for playing 25 hours if it has only played 5?

Any other added measures into the algorithm for WR, I don't know, but I am sure they have to experiment a bit with what they are able to do to get it right.

Edited July 7 by One more for the road.8950

[One more for the road.8950]

Let's say I have two accounts that I play on average 5 hours each a week.

Do you mean that one account should be credited for 10 hours and the other none?

That would pretty unfair to the team getting the account that only plays half the amount it's counted for, and a nice bonus to the team that gets the account that counts for 0 hours.

[Cyninja.2954]

22 minutes ago, One more for the road.8950 said:

You do realise it doesn't count all your WvW time on the account? I am not sure what you are trying to get to here.

Activity is measured over something like the last two months, iirc. If you stay out of WvW long enough you don't count. And coming back, it's just starting counting hours from then. Earlier activity has nothing to do with it. People have indeed been able to artificially reduce their servers activity before, which is why they extended the length of measure to what I seem to remember is a couple of months.

Yes that's what I was arguing. Glad you agree and if an account is avergaed out over 4-8 weeks, the impact of switching accounts is greater on the shard than if it was averaged out over a longer period of time, which is what Chaba is arguing.

Quote

Why would an account be credited for playing 25 hours if it has only played 5?

Did you read what I wrote?

The shard is created with a play time up to that point. Reducing play time from 25 to 5 hours does not affect the previous shard creation. That account is now missing X amount of hours (in this case 20). For the next shard creation it will obviously be credited with 5 hours, which becomes an issue if the account becomes more active again (where it might go from 5 to 25 again, thus over-performing with 20 hours now).

Now if the play time was averaged out over say 8 weeks (2 shard creations in the current system) and the account was active during 1, and inactive during the other, it would end up with 15 hours credited ((25+5)/2 = 15 ) which while reducing the extremes does not remove them. During an active week that account now plays 25 hours and during an inactive week it plays 5, even though credited with 15 in either case (thus trowing off the algorithm by 10 hours in either direction at different times).

So even with averaging out play time you never negate the impact fully. There is almost no instance where playing alt accounts is beneficial to the match making algorithm or WvW as a whole. There are individual fringe cases where having alt accounts can increase activity on an individual level (in case that player would have other wise logged) and thus benefit the mode, but even this comes at a cost of that player potentially investing time into a losing side (because it promotes fair-weather play behavior).

Which is the hypocrisy I was pointing at in my original reply.

Edited July 7 by Cyninja.2954

[Boki.8156]

it is fascinating to me how the lads from WSR got an Active "server" or im wrong.

 

Edited July 7 by Boki.8156

[Chaba.5410]

40 minutes ago, Cyninja.2954 said:

For all we know shards are created with looking at total play time over the last week/weeks.

You never read anything about the old population algorithm, did you?  A rolling average smoothed out over several weeks is not total playtime.  It is that algorithm which formed the basis of player/account evaluation.

Edited July 7 by Chaba.5410

[One more for the road.8950]

10 minutes ago, Cyninja.2954 said:

Yes that's what I was arguing. Glad you agree and if an account is avergaed out over 4-8 weeks, the impact of switching accounts is greater on the shard than if it was averaged out over a longer period of time, which is what Chaba is arguing.

Did you read what I wrote?

The shard is created with a play time up to that point. Reducing play time from 25 to 5 hours does not affect the previous shard creation. That account is now missing X amount of hours (in this case 20). For the next shard creation it will obviously be credited with 5 hours, which becomes an issue if the account becomes more active again (where it might go from 5 to 25 again, thus over-performing with 20 hours now).

Now if the play time was averaged out over say 8 weeks (2 shard creations in the current system) and the account was active during 1, and inactive during the other, it would end up with 15 hours credited ((25+5)/2 = 15 ) which while reducing the extremes does not remove them. During an active week that account now plays 25 hours and during an inactive week it plays 5, even though credited with 15 in either case (thus trowing off the algorithm by 10 hours in either direction at different times).

So even with averaging out play time you never negate the impact fully. There is almost no instance where playing alt accounts is beneficial to the match making algorithm or WvW as a whole. There are individual fringe cases where having alt accounts can increase activity on an individual level (in case that player would have other wise logged) and thus benefit the mode, but even this comes at a cost of that player potentially investing time into a losing side (because it promotes fair-weather play behavior).

Which is the hypocrisy I was pointing at in my original reply.

I have no idea where you are getting these 25 hours from. I think you are making this just extremely complicated for some reason and therefore it's hard to understand what you mean.

Let's do tea spoons. A player has two accounts. They are both in a guild. Let's say they raid on different days, and this way the player gets four good raids in a week, and about five hours in each guild - as in each account. The algorithm counts that account A does about five hours a week and account B about five hours a week. Who owns the account has nothing to say. If the player plays 7 hours one week and 3 hours the next week on an account has nothing to say, it's averaging at about five hours a week. But you want to add this number of 25 in here somewhere that I must have missed where comes from?

[Cyninja.2954]

2 minutes ago, Chaba.5410 said:

You never read anything about the old population algorithm, did you?  A rolling average smoothed out over several weeks is not total playtime.  It is that algorithm which formed the basis of player/account evaluation.

and as I explained, even in that case this mitigates the issue, it does not fix it.

[One more for the road.8950]

14 minutes ago, Boki.8156 said:

it is fascinating to me how the lads from WSR got an Active server or im wrong.

 

The lads from WSR are spread over about three teams and still seems to mostly dominate their MU's which says something about how skewed and easy to cheat the old server system was.

Edited July 7 by One more for the road.8950

[Chaba.5410]

Just now, Cyninja.2954 said:

and as I explained, even in that case this mitigates the issue, it does not fix it.

I've been saying the system MITIGATES.  You're almost there, just not quite.

[Chaba.5410]

45 minutes ago, Cyninja.2954 said:

If it did players would not have been able to artificially reduce their server size shortly before resets (aka play less 1 week before a relink to open up the server)

Say what?  So all those threads about Gandara tanking for weeks on end and not opening up is what again?  You seem to confuse servers that are just on the edge of population thresholds with the algorithm's calculations.

Edited July 7 by Chaba.5410

[Cyninja.2954]

Just now, One more for the road.8950 said:

I have no idea where you are getting these 25 hours from. I think you are making this just extremely complicated for some reason and therefore it's hard to understand what you mean.

I used example numbers....

Quote

Let's do tea spoons. A player has two accounts. They are both in a guild. Let's say they raid on different days, and this way the player gets four good raids in a week, and about five hours in each guild - as in each account. The algorithm counts that account A does about five hours a week and account B about five hours a week. Who owns the account has nothing to say. If the player plays 7 hours one week and 3 hours the next week on an account has nothing to say, it's averaging at about five hours a week. But you want to add this number of 25 in here somewhere that I must have missed where comes from?

Okay let's use your example but actually change the values so their are not identical because sure, it a player plays each account the same time, the issue does go away.

The player plays on 1 account 25 hours per week (5 of which he raids with his guild) and 5 on the other. Now shards are created and his previously 25 hour per week account is no fun to play. He will still stick around for the guild raids of 5 hours, but now spends his 20 additional hours on the other shard with his other guild.

FYI: this calculation can be done with you 7 and 3 hour example just as well.

1 minute ago, Chaba.5410 said:

Say what?  So all those threads about Gandara tanking for weeks on end and not opening up is what again?  You seem to confuse servers that are just on the edge of population thresholds with the algorithm's calculations.

A completely over stacked server..... really? This is your example? Yes, some server never opened for a while because they got over-stacked so far that no amount of reducing play time brought them remotely in the sphere of other server sizes.

[Chaba.5410]

3 minutes ago, Cyninja.2954 said:

The player plays on 1 account 25 hours per week (5 of which he raids with his guild) and 5 on the other. Now shards are created and his previously 25 hour per week account is no fun to play. He will still stick around for the guild raids of 5 hours, but now spends his 20 additional hours on the other shard with his other guild.

This is still bad analogy.  You need to know how many hours average per week each account is being played smoothed out over a period of 30+ days.

Then the teams all get reshuffled.

Edited July 7 by Chaba.5410

[Cyninja.2954]

5 minutes ago, Chaba.5410 said:

I've been saying the system MITIGATES.  You're almost there, just not quite.

Okay, so what is your issue with me pointing out the hypocrisy exactly? Mitigating an issue is not removing or solving it. 

Quote

mitigate

/ˈmɪtɪɡeɪt/

verb

verb: mitigate; 3rd person present: mitigates; past tense: mitigated; past participle: mitigated; gerund or present participle: mitigating

1. make (something bad) less severe, serious, or painful.

 

Making something less severe means it is still a negative factor.

[One more for the road.8950]

2 minutes ago, Cyninja.2954 said:

I used example numbers....

Okay let's use your example but actually change the values so their are not identical because sure, it a player plays each account the same time, the issue does go away.

The player plays on 1 account 25 hours per week (5 of which he raids with his guild) and 5 on the other. Now shards are created and his previously 25 hour per week account is no fun to play. He will still stick around for the guild raids of 5 hours, but now spends his 20 additional hours on the other shard with his other guild.

FYI: this calculation can be done with you 7 and 3 hour example just as well.

A completely over stacked server..... really? This is your example? Yes, some server never opened for a while because they got over-stacked so far that no amount of reducing play time brought them remotely in the sphere of other server sizes.

Why do you keep saying changing hours from one week to another? It doesn't matter, that's not how it works. It's averaging. So the player would have the average over the last 8 weeks or something iirc.

[Chaba.5410]

4 minutes ago, Cyninja.2954 said:

A completely over stacked server..... really? This is your example? Yes, some server never opened for a while because they got over-stacked so far that no amount of reducing play time brought them remotely in the sphere of other server sizes.

Again, bad analogy.  Gandara did open up finally after the 30+ days.

[Chaba.5410]

1 minute ago, Cyninja.2954 said:

Okay, so what is your issue with me pointing out the hypocrisy exactly? Mitigating an issue is not removing or solving it. 

It's not a problem that needs solving, just mitigating.  And it doesn't need solving because it's background noise.  It is no different from a player who logs out and plays later.

[Arya Whitefire.8423]

I'm like 70-90% convinced, that if you don't play for 8 weeks, then you stop counting as an active wvw player.  If it's not 8 weeks, then it's probably 10-12 weeks.

[Cyninja.2954]

1 minute ago, One more for the road.8950 said:

Why do you keep saying changing hours from one week to another? It doesn't matter, that's not how it works. It's averaging. So the player would have the average over the last 8 weeks or something iirc.

You need to step back, read, try to understand what is being written, then get back here.

I am talking about 25 hours attributed to an account. Be it average or not, that's what the algorithm, assigns. Then altering that play time, be it average or not, leads to mistakes in the shard creation. All you did was give individual examples which I could also average and you did so with 2 accounts which have the same play time. Yes, playing each account the same amount of time will not lead to issues, but that's not the reality in WvW because players will be driven to their account with better content for them.

[Chaba.5410]

2 minutes ago, Arya Whitefire.8423 said:

I'm like 70-90% convinced, that if you don't play for 8 weeks, then you stop counting as an active wvw player.  If it's not 8 weeks, then it's probably 10-12 weeks.

I think it's at least 30+ days cuz Grouch one time posted for a past WR beta asking players this question, if they played WvW within the past 30 days or not.

Edited July 7 by Chaba.5410

[Cyninja.2954]

1 minute ago, Chaba.5410 said:

It's not a problem that needs solving, just mitigating.  And it doesn't need solving because it's background noise.  It is no different from a player who logs out and plays later.

and I never said anything else. I said it's a minor issue which needs potentially addressing depending on where the development of mitigating factor is at.

That still does not remove the hypocrisy of someone affecting the mode in this way coming around to complain about balance.

So I don't get why you felt the need to disagree on this issue.