Nov 12 World Restructuring Beta Delay

[Fire Attunement.9835]

Hi all,

On October 21, we released “Inside ArenaNet: World Restructuring Beta 1,” a blog post that detailed the problems we encountered during the September 24 World Restructuring beta attempt and our next steps for the feature. In that post we mentioned that there were ten issues we needed to address before we could run our next beta event. Since that post, we’ve identified another must-fix issue, and unfortunately, we’ll need to postpone this week’s scheduled beta test until it’s resolved.

During the beta, players were being erroneously counted toward map caps twice when entering Edge of the Mists with World Restructuring enabled: once for their assigned World Restructuring team and once for their home shard. Further investigation revealed that this issue was also affecting the player counting for the primary WvW maps (Eternal Battlegrounds and the Borderlands maps).

While the code that manages map capacity is used game wide, player counts for WvW are more complicated than other maps because there is a necessary limit to the number of players per team to help maintain competitive integrity. The counting is handled in two places: (1) when a player connects and joins the WvW map, and (2) it’s updated periodically based on a connected player’s HomeShard. The problem is that players are connecting to the map and initially being counted using their World Restructuring team, but the periodic update is looking at their HomeShard. This causes all sorts of interesting side effects.

Changing the player count code to use a player’s BattleShard (the name we gave the team id that gets assigned to players by the World Restructuring system) instead of the HomeShard will fix this problem. To do so, the Game server needs to give the Connection/Communication layer information about the players’ BattleShard. Data normally goes in one direction between these two layers, but this situation requires data to flow in the opposite direction.

This means we need to make a change to a core service that could have game wide implications if done incorrectly. As we mentioned before, the code used for map capacity management is used across the entire game, and it’s legacy code that we don’t modify frequently. In a case like this, we need to slow down, triple check our work, and thoroughly test the changes in a number of different scenarios. When changes like these are made under time pressure they become more prone to error and increase the risk that something bad will occur. We’d prefer to spend a bit more time with the problem to make sure we are addressing it correctly and not just introducing more issues by rushing.

From our perspective, the September 24 beta attempt is still paying dividends. We’ve been able to identify and address several significant systems issues from that short test. We’re glad we were able to catch this one before running another beta.

Thank you for your time and patience while we work through this. We’ll provide an update on when that next beta test will take place as we’ve validated the changes we mentioned above.

-          The Guild Wars 2 Team

[Ganz.3917]

Thanks for the update!

[Justine.6351]

The one legit time people use confused.

[Sierra Nox.1402]

Thank you for the update, WvW team - KEEP AT IT!!   We'll still be here.

[TheSkeptic.4780]

Alliances when? 

[Naxx.6203]

It's all good devs. Keep your great work, we will wait till its ready.

[Kyban.4031]

Thanks for the update.  Since you're making modifications to the legacy system, is it a close enough system to add PvE map queues while you're at it?

[Tulki.1458]

Sad news but the fact that this update and a lot of others have been giving fairly detailed technical reasons for decisions has been good. It reassures people that things aren't being put off indefinitely but are actually being worked on.

[noncha.2761]

Nice, we wait 4 years already, we can wait 4 years more. Pls send me an email when "alliances" arrive. 

[Jwake.7013]

Thanks for keeping us in the loop.

[Doctor Who.2081]

itt: obsequious userbase.

 

Just another day in the neighborhood of being a WvWer. Maybe we can at least get the rewards in the interim that bots farm in pvp?

Edited November 10, 2021 by Doctor Who.2081

[joneirikb.7506]

Thank you for the feedback, it was an interesting read (I like "interesting side effects")

[eyestrain.3056]

Appreciate the transparency and being double-triple safe is better than sorry. Thanks for the update 👍

[andymcollie.7602]

as a long time WvW player, its better to postpone, dot the i's and cross the t's and make sure its all working ,   no one wants a repeat of the september beta   , thanks for the update

[Tranquillity.8697]

Super cool to see you guys talk so detailed about those problems. Fascinating stuff.

[ElderNewt.5840]

Thank you for letting us know of the delay but also actually why the delay is needed.
I found this post very interesting.

And as I've said before in many places... Take your time. Yes. No one likes delays.. but once communicated properly.. both sides can reach an understanding.

[Excursion.9752]

Will we see a beta before February?
Seeing how this is supposed to roll out with/around EoD I have a very negative outlook on this being ready in time. Seeing how long it has taken to address the last beta's problems. You are not leaving yourself with much time to address any newly found issues after a second beta. God speed to you!

[Charybdis.9042]

Yes, thank you for the detail in the post. Very much appreciated and welcome.

Touching a core service that is shared across the entire game is always risky (speaking from professional experience), so delaying the beta for this reason is the right call for WvW and the game as a whole.

[zealex.9410]

If player caps were reached twitce as fast then didnt that skew the population balance of the first beta?

[NotNoPants.9573]

So many paragraphs to blatantly say you don't have your core developers anymore and no one knows what's going on with the core game engine anymore. 

Just the other day, you said you had a 3-4 week pipeline AFTER things had passed QA testing, "Add a healthy portion of testing, bake for 3-4 weeks in our release pipeline (the average time from the moment something is finished and ok-ed to be released, to actually being in a release)" https://en-forum.guildwars2.com/topic/104393-legendary-trinket-effects-update/ 

Yet, here we are, only being told a few days before? when so many players have been shifting, making plans, trying to ensure they are ready for alliances because you gave a hard date when you thought you found the original problem? from a user perspective, you realize how being lied to is upsetting right? if you knew earlier, say something earlier, if your release pipeline is just a few days, maybe update that trinket post. (how hard is it to just make new unique leggos that follow the pattern of the wvw/pvp balls and just replace the ones on peoples account? months of work? oof)

WvW hasn't had an update in years, balancing is never centered around it, the meta has been the same for years, clearly you don't have the best project managers or developers on it (otherwise root cause investigations would of been properly scoped, and you wouldn't of given out a "we figured out what we did wrong" message last update). Yet, you still have the audacity to call it the core pillar of your game?

 

You're actively losing players to other mmo's that aren't even good. None of the new specs are wvw centric, all are outshined by previous specs for their roles, and all of their animations are borrowed from other classes, some are so obvious it's not funny. It's very clear anet just has devs that can make skins and maps while borrowing what the previous devs left them. 

 

Doesn't put a great view for the xpac, or the timeline for any other wvw update

[GaboDrogalan.8715]

Big sad

 

[bambi.6214]

Pretty cool we'r getting so much detail, thank you. Looking forward for the next try.

[snebzor.4851]

I appreciate the continual updates. Thank you! Keep working hard.

[saerni.2584]

Appreciate the update and the continued work on what is obviously critical code that can't just be casually changed without consequence. 
 

Keep it up and looking forward to finally trying the Alliances feature when it goes to Beta in NA. 

[Woop S.7851]

3 hours ago, Fire Attunement.9835 said:

...While the code that manages map capacity is used game wide, player counts for WvW are more complicated than other maps because there is a necessary limit to the number of players per team to help maintain competitive integrity. The counting is handled in two places: (1) when a player connects and joins the WvW map, and (2) it’s updated periodically based on a connected player’s HomeShard. The problem is that players are connecting to the map and initially being counted using their World Restructuring team, but the periodic update is looking at their HomeShard. This causes all sorts of interesting side effects...

Thanks so much for the hard work and update! To clarify (correct me incase I'm wrong), is the below roughly what happened? 🧐

Variables:
X: Player battle shard (rank)
Y: Home shard (home world server)

Example scenario:
Player1: X = 1500 rank, Y = Blackgate server
Player2: X = 100 rank, Y = TarnishedCoast server

In-game flow/logic:
-> Player1 is friends/a guildie to player2 showing them the ropes to WvW
-> Player1 groups with Player2 (party/squad)
-> Both players join EOTM (clicks button)
-> Server splits the 2 into two different instances where:
    EOTM Instance1: 100 grouped blue players (ranks between 1000 to 1200) vs 100 Red vs 100 Green
    EOTM Instance2: 100 grouped blue players (ranks between 0 to 120) vs 100 Red vs 100 Green

Result -> The 2 players are now split in two different instances of the 'same' EOTM map even though they're in the same party/squad

Analysis:

• In the current PvP, I believe there's a 'timer' system that matches groups only by player battle shard X (regardless of world server)? Maybe a good reference? Of course the only drawback are the long waits, but you do end up with even matches with people of the same level on both sides
• In GW1 Alliance Battles, this same scenario existed (waiting for a proper/even match) where groups are matched based on player shard X with '3 groups of 4' on each side (12 v 12)
• To use an analogy in this scenario, for WvW, the goal is to group match 12 v 12 v 12 of similar skill?
• Since EOTM is already built this way, to mitigate the risk of touching legacy code when GW2 1st launched, maybe a re-analysis of how EOTM worked might help?
• The hard/long way would be to solve above scenario where a match are even while skill levels are close, the issue might be sparked from a veteran player grouping up with a new player, there'd need to be some algorithm or 'condition' to calculate that in the system...
• Hope this helps! 😁

Edited November 11, 2021 by Woop S.7851