Best teamplay and most enjoyable matches usually happen on dedicated servers. That’s because each dedicated server is a community of sorts. Likeminded players tend to cluster around certain servers. Players who enjoy 32 players on a map designed for 16 players will go to server A. Players who use microphones to coordinate team efforts (rather than spilling insults, small talk and socializing) will go to server B. Matchmaking makes that impossible. There’s nothing to crystallize a community around.
Playing against people of equal skill can be a bad thing. You don’t get better by playing against average people. Computer players, musicians, and athletes all get better by playing with people better than themselves. If you see someone doing much better than you, you will probably do your best. Yes, sometimes getting **** kicked out of you is a good thing. Unless you just want to relax on a couch…
Even an ego admin is better than no admin. Cheating, foul language, racial slurs, abusing bugs… or perhaps you just want to try an interesting map or server setting. A nice admin can help with that. And if he’s a jerk, find a better server. A jerk admin will attract jerk players anyway. With matchmaking, all you’re going to get is some kind of automated system.
You have a really simplistic view of matchmaking. Match-making isn’t based on just one, but rather a number of variables.
There’s your actual rating, but you also have a (couple of ) derivative that determine how well you’re doing across different intervals. This means that a player that is on a winning streak is quickly put up against exponentially better subsequent players. Likewise a player that keeps on losing is put against players that keep getting worse and worse until he starts winning again.
Because there’s always a varied set of players available online, and because nobody is completely consistent (you can win against higher rated and you can lose against lower rated) this means that the matchmaking system is unable to be completely confident in your rating.
This way your pool of opponents is following a bell curve. Most people are equal to you, a third is considerably lower than you and a third is considerably higher than you. That way everyone gets a varied bunch of opponents to play against.
Even if the matchmaking was strict and you were always playing against equal players (Warcraft III had this) then so what? At least the games are always fair. Your argument that players should be playing against better players to get better is utterly retarded. For every player that plays against a better player, there’s a player that plays against a worst player. It’s a really weird sense of entitlement you got there.
There is no method to measure a player’s skill and there will never be such method that actually works. Given a set of rules to compute the “greatness”, someone will always play by those rules and not adapt to what’s going on. I rather see all these methods removed and let people decide themselves. Anyone happen to play on a laptop just for fun? Will i be matched against super pros like i used to play on a desktop pc?
Good point, murka. And if I want to show the game to a friend, I need to create a new account for him ? Don’t get me wrong, I’m a nerd. But matchmaking strikes me as antisocial.
I like to join the weaker team. This means I’m less likely to win than otherwise, yet it doesn’t have much to do with how I play. If you measure me by measuring my team’s win ratio, you’re going to count me as a crap player.
More people do this, because we enjoy challenge and games where teams are about equal. But there are plenty of people who enjoy seeing high numbers next to their name, so they will join a team full of good players.
I think we’re not on the same page in this discussion. Pub matchmaking happens on an individual level and is determined by the player’s xp. It’s not measuring skill, but rather the well, ‘experience’ of the player. Within pub matches you’re free to switch sides, and yeah, I rather be the underdog as well. I’m not joining the losing team (unless it gets really skewed) but I rather join the side which the map is usually favoured against. \
Then there’s clan ladders. I haven’t tried this in Brink yet (and by the looks of it I don’t think I ever will) but Blizzard has team-matches down to a science. What I wouldn’t give to have a peak at the algorithms they use (I also really want to see Facebook’s matchmaking algorithms, but that’s a different story).
Does automated matchmaking work with that many players? Surely it would require a huge competitive base, one far larger than what Blizzard needs for their system to work because of the number of players involved in an 8v8.
[QUOTE=jazevec;389255]Best teamplay and most enjoyable matches usually happen on dedicated servers. That’s because each dedicated server is a community of sorts. Likeminded players tend to cluster around certain servers. Players who enjoy 32 players on a map designed for 16 players will go to server A. Players who use microphones to coordinate team efforts (rather than spilling insults, small talk and socializing) will go to server B. Matchmaking makes that impossible. There’s nothing to crystallize a community around.[/QUOTE]Matchmaking only makes dedicated servers impossible if it is enforced as the only possible method for joining a game. As long as the player is presented with an equal choice of a server browser of a faster, matchmaking option, a thriving dedicated server community is totally viable.
Playing against people of equal skill can be a bad thing. You don’t get better by playing against average people. Computer players, musicians, and athletes all get better by playing with people better than themselves. If you see someone doing much better than you, you will probably do your best. Yes, sometimes getting **** kicked out of you is a good thing. Unless you just want to relax on a couch…
Your main point is accurate, but this doesn’t describe how a skill-based matchmaking bucket works. A (good) skill-based MM bucket is a range of players above, at and below your own skill rating. It’s very unlikely that a MM implementation will try to very closely match your skill rating because that will generally take too long to find a match, or it will create too many servers, all with not enough players populating them.
Even an ego admin is better than no admin. Cheating, foul language, racial slurs, abusing bugs… or perhaps you just want to try an interesting map or server setting. A nice admin can help with that. And if he’s a jerk, find a better server. A jerk admin will attract jerk players anyway. With matchmaking, all you’re going to get is some kind of automated system.
I disagree about ‘ego admins’: they are never enjoyable. A lot of what you describe can be accomplished without an admin via vote systems (all but the custom settings). Games that run as a persistent service, such as League of Legends, reward players for good behaviour and ban persistent offenders after a rigorous screening by the community at large, which weeds out the ‘jerk players’ much better than any ‘jerk admin’ could.
ELO is actually pretty good. Greatness is measured by wins and losses against players of higher or lower skills. Over time, the system is actually very accurate. However, ELO on its own is not a silver bullet, the design and implementation must work in tandem.
There’s always scope for less casual, non-skill-based matchmaking buckets. Matchmaking queries don’t always have to be based on skill rating!
Matchmaking should be like the AI Director in Left 4 Dead; it knows when you’re off form and when you need a greater challenge. It’s one of the most valuable tools you have in multiplayer games for helping everyone have a fun experience and, as such, I find it hard to come up with a case against it that outweighs the many benefits.
The goals may be the same but it’s hard not to see them as interfering with or diluting each other. Allowing team changing allows for a server’s population to attempt to balance themselves but overrides the automated system and visa versa on map change.
Matchmaking should be like the AI Director in Left 4 Dead; it knows when you’re off form and when you need a greater challenge. It’s one of the most valuable tools you have in multiplayer games for helping everyone have a fun experience and, as such, I find it hard to come up with a case against it that outweighs the many benefits.
Having a matchmaking system like the AI Director seems extremely difficult with two teams because it would be intrusive if applied during a match but far less potent if applied between matches as well as in conflict with local server efforts either going to my previous point with internal server shuffling or by shuffling players to other servers which is even more detrimental to local server stability and balancing.
If your AI Director is, instead of manipulating team composition, altering game variables like the time limit to compete an objective, AI number and skill, respawn time and location or opening and closing map areas, it will alienate more competitive minded players by changing the rules mid game as well as making them opaque.
I can’t stand playing L4D 2 versus because the luck involved in the non-player controlled class selection and weapon location spawn breaks coordinated location dependent tactics and makes the outcome more dependent on what the game gives the players when and less how well the players play. I played a lot of L4D 1 versus… There are ways this can be done with an attempt at subtlety but also ways it can be way overdone.
I disagree about ‘ego admins’: they are never enjoyable. A lot of what you describe can be accomplished without an admin via vote systems (all but the custom settings). Games that run as a persistent service, such as League of Legends, reward players for good behaviour and ban persistent offenders after a rigorous screening by the community at large, which weeds out the ‘jerk players’ much better than any ‘jerk admin’ could.
Internal vote systems and external reputation systems, like server or community administrators and moderators, are as good or bad as the people controlling them. The mob has interesting ideas when it comes to ideas of the “correct way to play” a role or where they perceive game hacking. This is not a magic bullet but, when designed well, I agree that it has the potential of improving the player base as a whole far more consistently than even the best local efforts could.
Indeed, I don’t disagree with any of the systems you’ve mentioned. However, if their benefit outweighs their cost depends on how well they are designed, and there is a significant cost when not done well.
I enjoy having a “go to” place rather than meeting up with random players. Matchmaking pits you against randoms unless you’ve actively added them to your friends list, grouped together as a party and actively joined matchmaking together. Having a hangout means you can all play together regardless of all the nonsense inbetween and you’ve got a place to gather with like minded players.
matchmaking works well for drop-in, drop-out casual play in games where it’s fine to play with randoms. In a game that requires a lot of team coordination, however, you’re almost always better off playing with people you know, ranking system notwithstanding. And dedis give players a “place” to gravitate towards.
All that being said, I don’t believe matchmaking works as well, simply because you can attribute so much fault to how the rating systems work, and I like the idea of dedicated server communities rather than a built-in clan system anyway. I would rather have a dedicated server system that doesn’t promise to separate players based on any sort of rating, but from what I’ve seen that actually drives a lot of players off when they can’t reliably find matches where the average skill level is around their own. You can make the case that you only get better playing against people who are better than you, but it’s simply not true that matchmaking only pits players of similar skill against each other. It simply tries to narrow the skill gap a bit; you will always run into players better or worse than you and there will always be someone to learn from.
On the other hand, I can imagine (and would like) both systems to somehow coexist. But this is probably too complicated and expensive to put into most games, and would probably turn a lot of people off instead.
ELO is actually pretty good. Greatness is measured by wins and losses against players of higher or lower skills. Over time, the system is actually very accurate.
An ELO system is designed with the idea that everyone tries to achieve ‘greatness’ when playing pub game.
Take ETQW for example, the better players out there never really put any effort, because it would kill the server.
As a high level player you need to allow the pub players, to feel like they have the opportunity to win the game. And a lot of the time, that means actually allowing them to win the game, when in reality you can single-handedly ensure they never exit their spawn. Sounds counter-intuitive right? Whats the point in playing if you’re not playing to win? Players use pub games to chill, chat in voip about things that arent even related to the game - they use the game as a social hub and play seriously when they play in competition.
Competition ladders outside of the game, exist to ensure that everyone is on the same page in terms of “greatness”. That everyone is using voip, everyone is playing at their 100% and everyone is playing to a plan.
edit: also allows competitions to set their own rules based on their values not the devs
I think it’s important to understand the following points regarding matchmaking, because there seems to be a lot of confusion regarding it:
[ul]
[li]Matchmaking describes the system for streamlining the process of locating a multiplayer game that best fits the player’s needs and connecting to the server. Anything that happens after the player has connected generally doesn’t have anything to do with the matchmaking system.
[/li][li]Matchmaking is essentially just like a filter query in a server browser. A list of specifications is made for game mode, or map, or team size, or latency threshold, or region, etc. and a list of matches is generated. There are four important differences:
[/li]- Not all aspects of the query are defined by the user, some are appended by the developers to streamline the process for the end user (e.g. latency, region, server has empty slots, try to join a friend’s match if possible)
The user generally doesn’t see the results of the query
Connection is attempted automatically once a match has been found
A matchmaking query generally isn’t looking for an exact fit, instead a range of possible matches is produced and the ‘best fit’ is then selected
[li]Games will frequently employ several different queries side-by-side to cater to the needs of as much of the playerbase as possible. It doesn’t make sense to rely on a one-size-fits-all query (there’s no such thing as a one-size-fits-all query).
[/li][li]A matchmaking query does not necessarily have to involve measuring a player’s skill.
[/li][/ul]
An ELO system is designed with the idea that everyone tries to achieve ‘greatness’ when playing pub game.
Correct, but as I have stated matchmaking does not always have to involve a player skill rating. You can have a matchmaking system that allows for both non-skill-based and a skill-based queries alongside each other. You can have a matchmaking system that allows servers to opt-out of certain types of matchmaking queries (e.g. ‘non-ranked’ servers). A lot is possible. For skill-based matchmaking queries, ELO is a pretty good blueprint.
matchmaking works well for drop-in, drop-out casual play in games where it’s fine to play with randoms. In a game that requires a lot of team coordination, however, you’re almost always better off playing with people you know, ranking system notwithstanding.
Matchmaking works for games where everyone has a specialised and pivotal role. Case in point: League of Legends (5 unique specialised roles per team: tank, ad carry, ap carry, support, jungler). LoL has several pre-defined matchmaking queries available to the player and actually works very well to group randoms with randoms or friends vs. friends (to list how it works would require a separate post).
Having a matchmaking system like the AI Director seems extremely difficult with two teams because it would be intrusive if applied during a match but far less potent if applied between matches as well as in conflict with local server efforts either going to my previous point with internal server shuffling or by shuffling players to other servers which is even more detrimental to local server stability and balancing.
You took my quote too literally. I’m not talking about something that updates as frequently as the AI Director, I was simply making the point that a good skill-based matchmaking system is adaptive to a player’s historical peaks and troughs.
Furthermore, matchmaking is not there to do the job of team balancing. Team balancing can either be left alone (whereby players or admins can look after it), or be managed by a dedicated system as strict or lax as is needed.
Yes. There’s still the confidence the system attributes to one’s ELO (for example if a player starts winning against higher skills or losing against lower skills) and how ‘tight’ the match-making then becomes.
Do you guys have any algorithms for all this? I’m really curious as to what Blizzard is using but they keep it under wraps.
That’s so disingenuous. That is a professional Korean starting a new account.
The only thing that I find an issue with SC’s matchmaking is that the short games have the same weight as long games. So people who want to climb the ladder fast usually go for quick strategies. I think long games, which show more of the player’s skill and have a naturally balancing effect should incrementally weigh more.
Mission-based shooters have a similar issue with non-symmetric maps. Attackers and defenders have different chances of losing. The weight of each game should be modified by the average chance of winning and losing the map on either offense or defense.
