On Matchmaking

If you have followed any communities around competitive first-person shooters this past year, you have probably seen a flood of complaints about matchmaking. Late last year, there was a surge in skill-based matchmaking, or SBMM, in games like Call of Duty, Apex Legends, Destiny 2, and even Fortnite. But what is skill-based matchmaking, why does it exist, what are the alternatives, and why did it become so prevalent if there was seemingly such vocal opposition to it from players?

Let’s start in the beginning by looking at the problem that has given rise to this influx in skill-based matchmaking. Imagine a brand new player who picks up a game with a competitive game-mode. This inexperienced player excitedly hops into their first match and goes against high-skill players and gets stomped. It’s just one match though so the novice player gives it another go... ...only to soundly lose again. And again. And again. This player who is just starting out keeps matching and losing against better players and actually improving or even having fun in this game starts to look like an insurmountable hurdle, so they quit playing altogether. Look, no one goes undefeated and losing is part of gaming. But it is something different to lose in a game where you never stood a chance. In the simplest terms, the main issue that skill-based match-making is intended to solve is players matching against others so outside their skill range that the game is decided before it even begins. Game developers want players to have fun because players who are having fun are more likely to keep playing a game. They are also aware that it is not fun to get trounced over and over. My theory is that the rise in skill-based matchmaking in competitive games lately, especially first-person shooters, is because game developers have looked at the data and seen that skill-based matchmaking is the best way to keep players having fun and keep them from moving onto another game. So how does skill-based matchmaking work? I cannot speak for every game, but a lot of games maintain a background rating for every player in player-versus-player game-modes commonly referred to as a player’s match-making rating, or MMR. This is different than a player’s rating in a ranked playlist as it is a hidden value which is often affected by a lot more than just winning or losing. Players typically do not know what all goes into it, and it is likely that every game developer has their own formula. For instance, in a first-person shooter, a player’s match-making rating might factor in wins, losses, accuracy, head-shot accuracy, kills and deaths, efficiency, and many other metrics. While we might not know everything that goes into a game’s match-making rating formula, at the end of the day, those factors produce a score for every player that goes up as a player performs better and goes down if their performance declines. The game’s matchmaking then attempts to build matches so that players with roughly equal scores (or scores within a given range) are placed in the same match. In its ideal form, skill-based matchmaking creates two teams of roughly equally skilled players, and if a player is playing at their best, they should win half their games and lose the other half.

One of the purported benefits of skill-based matchmaking is that it creates an environment that is more conducive to player improvement. A lot of people claim that the best way for a player to improve is to play opponents who are slightly better. Conversely, some suggest that better is better, and skill-based matchmaking makes it harder to improve because the skill brackets only expose players to others who are making the same mistakes and do not punish those mistakes in the same way as higher skill players. The reality is, there is probably some truth to both these statements. The competitive ranked ladder for Pokémon Sword and Shield is broken into eleven ranks and players who reach the highest rank are given a numeric ranking. Every month, the rankings reset and players fall back two ranks and have to climb back to max rank. As a result, every player in the top rank, whether they were a top-one-hundred player or someone ranked over one-hundred thousand, is put in the same pool. For a brief time at the start of each month, there is a massive disparity in skill levels as players work their way back to max rank. But, on the other hand, I have also seen matches that have pitted players at complete opposite ends of the skill spectrum that, while not the most competitive matches, introduced the lower skilled player to aspects of competitive matches that they would likely have not been exposed to for some time if they had been playing others at their own skill level. Having said all that, there are obviously some significant criticisms of skill-based matchmaking. One of the most common is that developers sacrifice connection quality to ensure players are at roughly the same skill. After all, what’s the point of an even match if you cannot hit your target? In first person shooters, lag can prevent you from hitting your shot or allow your opponent to close the gap and kill you before you can even see or react. It’s arguably worse in fighting games where even a few frames of lag can drastically change the outcome of a fight. Skill-based matchmaking also makes it difficult for players to experiment or try ‘fun’ load-outs or strategies. If a player works to master and become successful with a specific character or weapon, their matchmaking rating rises, which would then place them against higher skilled opponents. But because of their success and the tougher opponents, it becomes difficult to try to learn how to use a new weapon or character. This issue is especially problematic given how buffs, nerfs, and balance patches can have a major impact on what is and is not successful in competitive matches.

Another problem tied to skill-based matchmaking is the effort and incentives to put in effort at different skill levels and how it affects match-making ratings. This problem is not inherent due to skill-based matchmaking, but when it is present, skill-based matchmaking shines a light on it. Before that though, we need to talk about sweating. A lot of gamers use the term sweaty to refer to high-skill players. However, the term did not originally relate to a player’s skill, but their effort. Calling a player sweaty was another term for a try-hard. Skill and skill ceilings are relative to each player, and if skill-based matchmaking is working correctly, a player should be winning around half their matches only if they are playing at their best, or in other words, sweating. Just because a lower skilled player cannot perform at the same level as a higher skilled player does not mean the lower skilled player cannot put in the same relative effort. I recently listened to two competitive gamers discuss the nerves that come with competition. A player who has reached the highest level knows what it is like to compete to win a championship or title. On the other hand, a player who has only ever made it to the second day or the second round of a tournament has that feeling when they reach and try to overcome their own competitive ceiling. These two agreed that players can only relate to the highest level they have reached, but that does not mean the player who has never reached the finals cannot relate to the nerves a player feels competing in the finals. The player who has not made it as far just feels those pressures much earlier than the player who knows what it is like to move past that point.
When Destiny 2, Year 3 started, Bungie switched most player-versus-player activities to skill-based matchmaking. This change was met with vastly different reaction from players at different ends of the skill spectrum. Lower skilled players loved it because they had a more casual experience and, as intended, they did not have to worry about getting crushed by higher skilled opponents. Conversely, high-skill players lambasted skill-based matchmaking because they were constantly thrown in matches that forced them to try hard and use their best load-outs. When skill-based matchmaking is working correctly, it prevents lower skilled players from getting stomped, but it should also mean that all players, regardless of their skill level, should have to try their hardest if they want to win. In other words, players who had a casual experience had that because their match-making rating was not accurately assessing their actual skill level, and was putting them in matches against lower skilled opponents. In the case of Destiny 2, the reason this happened was because the incentives to put in effort, and therefore have a more accurate match-making rating, were greater for players at the higher ends of the skill spectrum. During Year 2, the game’s ranked playlist offered a powerful weapon for players who could hit the max rank in the ranked playlist (which also changed to skill-based matchmaking in Year 3, but that is a different discussion). This meant that players had to either perform at a high skill level to acquire the weapon or work to improve to reach that level. On the other hand, players who avoided the ranked playlist or only played to acquire the lower tier rewards never reached their true skill ceiling. This issue only became more pronounced when Bungie reintroduced Trials of Osiris, an endgame player-versus-player activity designed for high-skill players to prove themselves which, again, only provided an incentive for high-skill players to put in effort.

So if players are not matched based on skill, what are the alternative forms of competitive matchmaking? Apart from some specific cases, there are two major alternatives, connection-based matchmaking (CBMM) and rank-based matchmaking (RBMM). As the name implies, connection-based matchmaking places the emphasis on connection quality. This has the benefit of reducing issues like lag, but risks creating scenarios like the aforementioned situation where new players or lower skilled players might struggle to find their footing. Connection-based matchmaking also creates a more casual experience as participants are more or less drawn from a random pool of players with varying skill levels. If you want to try in every match, you can. If you want to relax and play a few matches while enjoying a beer, you can do that as well. If you want to try some crazy load-out or strategy, you can do that too without feeling like you are playing at a disadvantage.
Rank-based matchmaking is similar to skill-based matchmaking, but is usually reserved for specifically denoted ranked playlists. Unlike match-making ratings, in ranked playlists, a player typically gains point for a win and loses points for a loss, regardless of their performance. As a player wins more matches, their rank goes up and they gradually face tougher opponents until they reach a point where they are winning and losing roughly the same number of games and their rank stabilizes. At that point, the player needs to improve their own performance to keep climbing the ranks. One of the big advantages over skill-based matchmaking is that rank-based matchmaking provides a visual representation of a player’s skill. Players know where they fall on the skill spectrum and can see if they are improving. When I started playing Overwatch, I originally placed in gold tier. Some weeks my rank went up, some weeks it went down, but I was able to gradually improve and work my way to mid-diamond over the course of several seasons. I have since stopped playing Overwatch and prefer not to imagine where I would place if I completed my placements. Of course, that visual information only works if the game actually uses rank-based matchmaking. As I mentioned earlier, Destiny 2 switched its ranked playlist to skill-based matchmaking in Year 3, which effectively made the ranks meaningless. However, Bungie did not confirm this change for over six months, so the ranked playlist served to misinform many players of where they fell on the skill spectrum. When Trials of Osiris returned, there were a lot of players who struggled to win even a single game and many of these players could not understand why Trials was so difficult when they had been able to hit max rank in the ranked playlist.

While there are similarities, there are some key differences between rank-based matchmaking and skill-based matchmaking. For starters, rank-based playlists typically have stakes attached to winning and losing. And especially in team-based games, the emphasis on winning can be a frustrating experience. Since players can only increase their rank by winning, a player’s rank will go down if they lose, even if they were the best player on their team or even the entire lobby. Anyone who has spent time in a ranked playlist in a team-based competitive game has probably heard of the term, Elo Hell. This term refers to the situation where a higher skilled player finds themselves repeatedly losing, and therefore dropping in rank, because of low-skill teammates. It is worth noting, some games have developed systems to prevent players from falling too far below their actual skill level (intentionally or not) and there are many who argue that Elo Hell does not actually exist.
Unlike skill-based matchmaking and connection-based matchmaking, because of those stakes, rank-based matchmaking typically cannot exist on its own. If skill-based matchmaking is seen as unforgiving to experimentation, rank-based matchmaking punishes experimentation. Still, games have found ways to work around that criticism. For instance, Super Smash Bros. Ultimate gives players a different rating for each character. That way, if a player unlocks the Elite Smash tier with their best character, they do not have to worry about matching against others at that skill level if they decide to try out a new character. Around a year ago, Overwatch adopted a similar system, splitting its ranked playlist into three categories. Now, a player earns a different rank based on whether they play as a tank, damage, or support hero. While it is not as specific as the one in Super Smash Bros. Ultimate, it allows players to experiment with different roles without worrying about whether they are playing their best role.

There is one important caveat to this discussion that I have been avoiding, and that is the source of complaints against skill-based matchmaking. Love them or hate them, there is no doubt that streamers and content creators have become a core part of the gaming experience in recent years. When it comes to competitive games, these players want to create content that features amazing victories, crazy combos, massive kill streaks, and in the same vein, their audiences want to be entertained by that content. It is more entertaining to watch someone get a tactical nuke in the latest Call of Duty rather than see them finish a match with a barely positive kill/death ratio or to see them get a string of Victory Royales in Fortnite instead of dying before finding a gun. However, while it sometimes feels like content creators and their followers have a vocal influence on games (whether that sentiment is accurate or not), it is important to realize that they are, at best, a vocal minority. Having said that, one of the arguments in favor of skill-based matchmaking is to protect average players from content creators and the highest skilled players. But the same criticism that content creators are a small minority goes both ways. Content creators and players in the top one-percent are a small portion of any game’s player-base. When games use connection-based matchmaking, lobbies are going to be filled with a wide range of players. Sometime yes, there will be a top-tier player in the lobby. However, if a player is repeatedly getting crushed, it is more likely that they are a below average player getting crushed by average players rather than a case of them going against top-tier players over and over.

Skill-based matchmaking is not perfect, but it is also not as flawed as the loud voices that clamor for its demise suggest. For a long time, I supported skill-based matchmaking because I believed it gave players the best opportunity to improve. However, as I have talked and listened to more players, I’m not as convinced that is always the case. I also do not think every player is always looking to improve, and do not think there is anything wrong with wanting a casual experience from time to time. Part of what motivated this piece was the removal of skill-based matchmaking in many, but not all, of Destiny 2’s competitive playlists and the resulting discussions. Personally, in part because of this change and my own experiences, I believe a diversity of matchmaking styles is better in the long term. But, having multiple playlists also means dividing your player-base and I am thankful that I am not responsible for balancing that diversity with adequate playlists and rewards and other incentives. Given the different experiences that different matchmaking styles offer to players at different points on the skill spectrum, I doubt there will ever be a real consensus. That said, I think we can all agree, a ranked playlist should use rank-based matchmaking.