[2021] Cluster Queue

Hey there,

in the past weeks, there has been a lot of discussion about the smart cluster queue feature introduced with Queen.

With this post, we'd like to give an overview of the situation and share with you the steps that we consider taking.


What is the goal of the cluster queue?

There are hard tech limits to how many players can fight in a single zone. These tech limits are not subject to hardware, meaning that no matter how many servers we use, they can't be exceeded without causing massive issues. We have been able to extend this limit substantially since the game's release, but that won't change the fact that such a limit will always exist.

Now, prior to the cluster queue, it was possible to block off a zone by placing a large number of players there and reaching the zone player cap. This stopped other players from entering entirely.

The goal of the cluster queue is to prevent zone capping.



How does the cluster queue balance?

In simple terms: if you bring exactly twice as many players to a fight than your opponent, you can expect to get twice as many players into the zone than your opponent - not more, not less.

The reason why the cluster queue logic is set up like this is to make sure that you cannot "cheat" the system by splitting up your group into multiple alliances. The cluster queue does not really care about alliance split ups.

To be a bit more specific: say 4 alliances A, B, C and D try to enter a zone, and their respective player counts are A=400, B=300, C=200, D=100 for a total of 1000. Assume that the cluster accepts 500 players. The count inside of the cluster would then be: A = 200, B = 150, C = 100, D = 50. Now imagine A wants to "game the system" and splits up into A1 and A2 - what would happen? Nothing really, as the new player count inside of the cluster would be A1 = 100, A2 = 100, B = 150, C 100, D = 50 - the split up A still having 200 players in total as before.

What this however also means is that if an alliance A with 900 players attacks an alliance B with 100 players, and the zone cap was again 500, inside of the zone, A would get 450 players and B would get 50. The cluster queue evens out fights in such a way that the ratio of players in the zone reflects the ratio of players in the queue - not more, not less, as anything else would be subject to heavy abuse.

Note that the above explanation assumes that all players have the same item power - the cluster queue also takes item power of individual players into account to prevent groups from stuffing the queue with a lot of under-geared players at low risk.



What about the "cluster queue abuse" that is often brought up?

Currently, the cluster queue gets updated every 90 seconds. Every 90 seconds, it checks the players in the queue and in the zone, determines their ratios, and then kicks players out and lets others in accordingly.

There is a strategy that seeks to exploit this: the attacker A camps all entrances to the zone that they try to attack. When they zone in, the cluster queue gets activated. The defender D, already inside the zone, then has some players ported out by the cluster queue. So far, so good. Now here is the key of the strategy: If the attacker can prevent the the players of the defender to re-queue for the zone (as they control all the entrances), it means that when the cluster queue updates the next time, the total number of players of the defender that they have in the queue and zone has been reduced while the number of players of the attacker has stayed the same. Hence, additional players from the defender will be removed by the queue.

Right now, the counter-play to that strategy for the defender is to make sure that they control at least one of the entrances to their zone. If they do, they have safe port out spot for players affected by the cluster queue. These players can then immediately re-queue after being kicked out. Then, the next time the cluster queue updates, there will be no further kick outs.

If, however, the attacker outnumbers the defender by such a margin that they can easily control all 4 entrances, then it is extremely hard for the defender to prevail. Having said that, one could argue that if you are so heavily outnumbered, you would also not have prevailed if there was no zone cap at all. However, this does not mean that we consider the current situation to be fine. We are closely looking into it and have a set of measures lined up to improve the situation.



Measures we intend to take

Note that as always when it comes to balancing highly complex game mechanics, the below list is subject to changes and adjustments.

• With the next patch, the disarray curve will be flattened to reduce the relative power of split up alliances. However, at the same time, players inside in the cluster queue will now count towards the disarray debuff inside of the zone. Hence, if you heavily stuff the queue with a lot of players to as many people inside of the zone out, you'll be penalized for that through a stronger disarray
  
• A potential weakness here is that the attacker - then subject to a stronger disarray - could immediately zone back out and then solely focus on hunting down the defenders players who got ported out.
  
• To counter this, shortly after the patch (was not ready in time for the patch as we needed to rework some queue logic), we intend to do one of the following:
  
• Option a) We will introduce a feature that prevents players who entered a zone through the cluster queue from leaving the zone for a certain time (say, 60-90 seconds). With this in place, once an attacker stuffs the queue to kick players outs, his own players that get ported in will be subject to a high disarray debuff and won't be able to immediately run back out, hence, giving the defender a very proper counter-attack option. We might combine this with a longer zone-in bubble to prevent the defender from effectively camping the entrance on the other side.
  
• Option b) An alternative to the above is to allow attackers who enter a zone via the CQ to leave the zone as normal, but if they do and then re-queue, their "queue weight" for that zone would be set to 0 for that zone. We might combine this with a longer zone-in bubble to prevent the defender from effectively camping the entrance on the other side.
  
• Both a) and b) would allow the defender to launch a counter-attack option against one of the entrance once the attacker has zoned in, in particular if the attacker decides to split his force to camp all 4 entrances, as the defender can then choose to counter-attack only against one of the opponents forces.
  
• If the attacker concentrates his force at 1 entrance, however, he can pretty much zone in as normal, having his forces in one place starting a normal engage after the original zone in.
  
• We are also looking into ways to give alliances more options in deciding in which order their players get removed by the cluster queue - for example, to allow them from keeping their party structures largely intact or preventing their healers from getting ported out, etc.
  
• We are also looking into giving players who are ported out of the zone via the CQ a longer lasting and better protection bubble, possibly one that is similar to the one you can get in portal zones (i.e. invisibility, invulnerability and silence). This should make hunting down players who got ported out much harder.
  

We think there is a very good chance that the above changes will largely deal with the above mentioned "cluster queue abuse" strategy. We'll keep evaluating the situation and make further changes and adjustments if required.

We are looking forward to your feedback.