22/05/2017

The Curse of Queues

The other day I was playing my Cathar agent on the Progenitor - yes, the one I created two years ago and who still hasn't finished her class story... but I will get there eventually. While doing the rounds on Belsavis, I decided to queue for a midbie PvP match. As time continued to pass without a pop, I started to get restless. I'm spoiled by how quickly everything pops on the Red Eclipse and I'd heard the rumours about other servers supposedly being quite dead, but had it really come to this already? I double-checked my achievement panel: Yes, I had done PvP on the Progenitor before, but quite some time ago. Had the server really lost so much of its population since then?


A warzone did eventually pop, though only after what felt like a pretty long time to me... it must have been between half an hour and an hour. And it really got me thinking about how strongly that queue time had affected me, even though I hadn't even wanted to PvP that badly. The whole point of a queueing system is to make finding groups super accessible so that it happens fast and frequently. If there aren't enough people around who can be bothered to do as much as press a button, surely that's a sign of "dead game"? Or at least of a dead server?

But then I also asked myself: Is that really fair? Or have automatic queueing systems for group content made us expect too much? The first time that thought occurred to me was actually after reading a blog post about Wildstar (not sure where anymore), though my train of thought didn't so much focus on the players at the time as on the devs. The game was supposed to launch with 40v40-sized "warplots", alongside a plethora of other queueable PvE and PvP content. I remember wondering just how big of a playerbase the devs were expecting, to think that at any time there would be enough people online for each type of content to keep all those queues popping, if warplots alone already required 80 players per match.

It seems to me that a game with automatic queueing systems requires a considerably larger player base to be perceived as healthy than a game where such systems don't exist. When I played on the private WoW server Kronos, which rarely had more than a couple of hundred players online at any given time, I was kind of surprised by how little it took to make the game feel alive. There were always people around me, and I got into groups by simply talking to those same people I met out in the world. In comparison, fully automated queueing systems are largely detached from the rest of the virtual world and give little feedback. SWTOR's in particular provides you with no information about how long you've been in the queue, how long the average queue time for that content is, or how close you are to having assembled a full group. There are just two modes: Either you press the button and stuff happens... or it doesn't. if it doesn't... dead game!

Eventually I decided to do some simple maths to find out just how many players are required to keep all the different queues rolling at a healthy pace. Obviously this requires me to make a lot of assumptions and guesses, but I think the logic behind them is mostly sound. Here's how it goes:

In total, there are 11 different queues for group content in SWTOR, and each requires a certain amount of people per match/run:

- Galactic Starfighter (16)
- Lowbie PvP (16)
- Midbie PvP (16)
- Max-level unranked PvP (16)
- Ranked solo PvP (8)
- Ranked group PvP (8)
- Veteran flashpoints (4)
- Master mode flashpoints (4)
- Story mode uprisings (4)
- Veteran uprisings (4)
- Story mode operations (8)

For simplicity's sake, we'll be ignoring the fact that those people need to be balanced across certain levels, roles and factions for the content to function and are just assuming that the player base is perfectly spread out. So we need a minimum of 104 players to make all this stuff happen... at some point.

But we don't just want the queue to pop once a day. We want it to pop, say, every fifteen minutes. Sooner would obviously be better, but I think most players would consider fifteen minutes a reasonable wait, and it's easy to work with since most of the content can be split into fifteen-minute chunks.

GSF and PvP matches do take a little less than fifteen minutes on average, so even if we only have the minimum amount of people queueing, as long as they keep queueing, a match will indeed pop every fifteen minutes. So far, so good. Same for uprisings.

Flashpoints take a little longer, about half an hour if it's a smooth run going at an average pace. So if you only have one group running, the queue would only pop every thirty minutes. To achieve the desired fifteen-minute pop, we need to have eight people interested in running each type of flashpoint at any given time.

For story mode operations, this is magnified even more, as they take about an hour each, so you would need four groups running at the same time, resulting in a total of 32 players doing this content at the same time.

That brings us up to 136 players required - they can probably keep things running for an hour, but they can't be online 24/7. To keep those queues popping around the clock, we'd probably need something closer to 3,264 players (136x24).

However, no MMO has a player base that hardcore, where everyone does group content all the time. Devs have gone on record not long ago to say that almost every activity in an MMO is only interesting to a minority, and when Syp from Bio Break ran a poll on the subject to see what his readers were interested in (you could tick as many as you wanted), even the most popular option barely got more than 10% of votes. So again, let's be generous and assume that about 10% of players are actually interested in running this stuff at any given time. Suddenly you need more than 32,000 players per server, and that's with all those generous assumptions like people being spread perfectly across the levels, never having a shortage of tanks etc. In practice the required numbers to keep things running smoothly are probably a lot higher still - for example if only 5% of players are actually interested in group content, the server would need more than 64,000 players on it for those queues to keep popping, and if you want them to go faster than every fifteen minutes, you'd need to add yet another multiplier.

What is my point here? Probably that while automated queues are in high demand from players these days, they are also dangerous in terms of PR because they require a pretty big population to run smoothly, with the requirements only becoming larger and larger the more different types of content you add. In an environment where people are actually willing to go out and build groups themselves, you don't have the same kind of pressure because it is accepted that other people won't be available to group with instantly and at all times of day. You'll still get people complaining about that of course, but at least they can't claim that your game is dead just because they couldn't get a group. (Didn't try hard enough, I guess!)

I don't expect automated queues to go anywhere, but maybe it's time to reconsider our expectations of what kind of experience they are supposed to deliver. WoW's dungeon finder was introduced at a time when that game could draw from a pool of about three million players per region to fill groups, and that has clearly set expectations. But we can't expect smaller MMOs to always deliver the same kinds of features and experiences when their smaller populations can't necessarily support them.

6 comments :

  1. It also of course doesn't help that a slower queue might also have additional unforeseen effects which make the experience less likely to be enjoyable for all parties.

    A player might go AFK while bored rigid waiting for a particularly slow PvP or FP queue, miss the pop, and then the group is either forced to wait a minute for it to go nowhere or to start the match a man down. Both groups do have systems to fill the gap, although of course it's technically less tedious to fill a WarZone gap than it is a Flashpoint gap, since the WarZone isn't locked from starting simply because the sixteenth player hasn't turned up.

    (Although whilst a Flashpoint can be undermanned, obviously a WarZone shorts out if not enough people are present; this should hopefully never be the cause of a single person, though.)

    This is also not accounting for the amount of time it takes for said player to realise that they did miss the pop and thus try to account for it, at which point the entire process may begin again. If the slow-queue problems occur enough times, the player might shrug off the queuing as too much of a hassle, and all-but retires never having seen the system at its 'best'.

    I appreciate them trying to make the Group Finder have a different sound effect to catch people's attention more, but of course this doesn't help if people are tabbed-out and thus unable to hear the game sounds if they don't have that particular setting turned on.



    Also, your point in the penultimate paragraph reminds me somewhat abstractly about a player I saw in a GSF match who kept asking how to "hack"; he never gave any context as to what he was trying to hack (it was a Domination, so I'm guessing he was asking how to control a satellite?), and because nobody responded satisfactorily (which, considering that everyone was busy fighting at the two guarded satellites, is no big surprise) he quickly denounced the entire game - not just GSF - as "dying".

    ReplyDelete
    Replies
    1. Fun fact: The Voidstar where I took the screenshot featured in this post popped very quickly but was aborted after 30 seconds because not enough players joined.

      Delete
  2. Worüber Du alles nachdenkst?! Faszinierend.

    ReplyDelete
    Replies
    1. Das Leben eines Bloggers... :P

      Delete
  3. Amen.

    What people frequently fail to realize is that what the most vocal players want isn't necessarily best for the game.

    The last time devs tailored a game strictly to the loudest players in the room, we got Wildstar. (Or, in regular pencil and paper RPGs, D&D 4e.) And that didn't exactly go over so well with the greater MMO community. (Neither did 4e, which ended up splitting the D&D community in half, between Pathfinder and 4e.)

    Sometimes, it's better to trust the devs and their data rather than the common complaints in the forums.

    ReplyDelete
  4. Interesting, that touches on something I experienced last weekend. When GSF was up for the CXP bonus, I decided to try and queue for that. Almost two hours later, while doing planetary quests from the CXP screen, I still didn't get a pop.

    I wouldn't go so far as saying the game is dead, because the activity on the fleet the moment Tyth became the active GF Operation was frantic. Groups were forming all the time, usually filling all roles within minutes. It's rather that some aspects of the game, combined with time of day, are not popular enough to support a queue system. GSF unfortunately is one of them. I hope that the updates that are coming up (as mentioned by Keith) will help raise the popularity of that part of the game, because having it included in the CXP bonus rotation seems to not really have done anything for it.

    ReplyDelete