I’ve read that some people view this as an issue. Players at the top probably never encounter this, so it could be a neglected but serious problem for typical players (the most popular streamers are experts, and so their live and YouTube viewers would not see this happening). I think I read that it makes Quickplay almost unusable?
If the only options are to allow this, or to ban these players, then the better choice is to allow it. But this is not the only option: as the title says, make these players play against each other. Maybe they view it as a legitimate thing to do for some reason: it’s not like they’re breaking the law. But if it’s fine for them to do, then it’s also fine for their opponent to do. (And sometimes, people lose/quit because their game crashes.)
One might think that if there was a solution to this, it would already be popular in games and well-known. But a lot of games have poor system design. A recent video reports that retail World of Warcraft will soon introduce a system that allows the community to punish leavers. And WoW has problems like this resulting from the convenience of easy access to gameplay along with the virtual anonymity from large player pools, leading to terrible behavior by players, for at least 14 years.
So the solution is simple to describe. For matchmaking: 2D distance instead of 1D with just rating. Second dimension is ‘how often players lose matches early.’
As with basic, 1D matchmaking based only rating, the system would attempt to match players within a low distance. Players who wait longer in the queue would be more forgiving of the distance to their opponent, whether that distance is from rating or from the tendency to quit games early.
Maybe discussion could reveal how to detect quitters. For this purpose, it might be a nice long-term goal to allow for the continuation of a ranked game after a client crashes, using game restore: maybe other players could be notified of the crash (which also looks like a disconnect) and have the option of waiting for the crashed player to restart their client and load a restore file.
But generally, a loss or resign within a short period of time would count as quitting, with a gradual fade-out so players don’t just wait a certain time and then quit immediately after. Like, defeat in less than two minutes is 100% a quit, then linearly fade out to 10 minutes so quitting at six minutes in counts as 50% of a quit. Can look at data for people with huge drops in Elo to see if they wait at all before quitting (they probably don’t).
For team games, refine it a bit: maybe, if a player leaves or deletes their units before 10 minutes, allow the remaining players to vote to end the game without penalty, either for being a ‘quitter’ or in rating.
And of course, don’t just use a lifetime average; have recency bias, to allow players to redeem themselves.
Should people who frequently quit games have to play against other quitters?
- Yes
- No
But there’s more. Reportedly, console players with cross-play disabled can only play against other console players with cross-play disabled, which can cause huge differences in rating between opponents. This is most likely just due to a bit of laziness in designing the matchmaking system. It’s easier to place people in separate pools, than to have people be unable to play against certain opponents in the same pool. This same logic might underlie restricting the map bans to less than half of the available maps.
Probably almost no console players disable cross-play, but their case and queue times should still be looked at.
But, there’s more (that concerns more than a tiny sliver of players). I think that, in general, the settings for ranked games are restricted because of concerns about fragmenting the ladder. Suppose that players could choose play with all victory conditions enabled in ranked games, instead of just Conquest: if this was a separate ladder, just like how Empire Wars, Death Match, and Return of Rome are separate ladders and many players have 0 games played in any of them despite hundreds of games of RM and team RM, then there’s a big chance that a lot of players would never use it. If you jump into it, then due to your default rating there’s a big chance of fighting players who either much weaker than you or much stronger. Even if many players used it, then if a top player wants to play against other top players, they have to make sure they choose the right settings. Choosing the wrong ladder could mean a doubling or more of queue time, which might already be 5~10 min.
So OPTIONAL settings make more sense, instead of a separate ladder.
The only time this doesn’t work is when the gameplay is completely different. When TheViper was winning every standard RM tournament, he was also winning many Deathmatch tournaments: the gameplay is not too different. But Return of Rome has different units I think, and the separate ladder might make sense due to knowledge requirements. Team games might be useful to separate from solo games simply to convey what a player enjoys more, even though 2v2 team games might be just as different from 3v3 and 4v4 as they are from 1v1, and ratings for all team sizes are combined.
So: this would require UI changes, to select which game options a player prefers. For example: new players might want to play on slow speed. Players might want to play with Turbo Mode on, or Sudden Death (protect starting TC), or to not have fixed team positions or Explored maps on Black Forest, or to play with Wonder and Relic victories enabled, or to use a population limit other than 200. They can already do this with lobbies, but the matchmaking system might be better at ensuring balanced teams or be quicker for finding opponents. If extra options don’t hurt the gameplay experience of players, why not do it?
For matchmaking, it would be hard to find 8 players who all choose the exact same settings. Calculating a ‘distance’, using extra dimensions (on top of rating and ‘tendency to quit’), to find similarity between chosen settings even if it isn’t perfect. Prefer the default settings in case of disagreement.
Secondly, a feature that would have more benefit if extra options can be chosen: the option to decline a ranked match, and in general have some control over whether to play an ideal game against a similar opponent even if it takes a long time to make the match, or to play a less ideal game. The Quickplay or whatever option might be related to this?
What I imagine is, when a match is made, the player is notified of any special settings for the match (which may or may not be what the player selected, if it’s a team game), as well as the relative strength of the opponent(s), and which map it’s to be played on.
They can then either join a match or flee the battle, but they can only flee twice in a row, with no option to flee on their third match. Other players are notified if a player chooses to flee (like a pokemon at low health). With team games, if more than half of players flee (in total), the match is abandoned; otherwise the system finds replacements. The opponent or opposing team’s relative strength is just described in general terms; it shouldn’t give away an opponent’s identity, even though the highest-rating players already have a website that does this. (Teams are easier to match in general, but with unusual settings the teams might be more unequal, especially if some players queue as a team.)
Minor consideration is preventing rating collusion: people don’t pay for rating in AoE2 the way they do in World of Warcraft, but maybe prevent someone from being matched against the same opponents too often? (Treat the ‘distance’ calculation against that opponent as having a lower bound, allowing other opponents to be found even if extra settings chosen aren’t the same, while not penalizing Hera with extra queue time for farming rating from a 2500-rated player.)
Allow selecting special settings when queueing for Ranked games that might be used if most players chose that setting?
- Yes
- No
Give players the option of fleeing a matched game based on information provided about it like the map or opponent strength, with a limit on consecutive fleeing?
- Yes
- No
