No subtitle necessary.
This is a point that I talk about a lot on my stream and with other highly rated players. It’s such an accepted reality for top players that there is not much to discuss; we all recognize that one-tricking ruins games because it so often ruins our games. I would imagine that the behavior is similarly frustrating to flexible and competitively minded players at any rating. Nonetheless, for many it is a point of contention and this post will regard it as such. I felt that it might be helpful to the discussion to clarify the argument from my perspective. I encourage you to read through and comment below with your thoughts.
‘One-tricks’ is a term that picks out those players in Overwatch who only play one hero. This is an importantly distinct concept from ‘Mains’, which is a term that picks out players who (perhaps significantly) specialize in one hero, but are willing to swap if the game demands it. This article will contend only that the former (one-tricks) are significantly corrosive to the competitive experience in ranked matchmaking.
One-tricking is always a problem because every hero is, to some extent, situational. The Skill Rating system that has been engineered to create competitive matchups is, however, unilateral. By ‘unilateral’, I mean that each and every player has a numerical value that distills their expected game impact and contribution to victory. This number is reflective of both historical W/L and individual statistical performance, but this is of course a topic for another article.
Since the matchmaker seeks to create teams with similar average SR, it crucially relies on the accuracy of its judgement of ‘expected contribution’ to create fair games. If a given player were to consistently under- or over- perform relative to their expected contribution, it will adjust in pursuit of equilibrium. This is of course the way the Skill Rating system is supposed to work.
What about in the inconsistent case?
Every player is to some extent inconsistent; we are not machines and our success rate will of course vary from game to game. However, for the vast majority of players this inconsistency will entail a fluctuation above and below an average. This fluctuation is unavoidable, but I would argue that it is not significant enough to pose a tremendous problem for the matchmaker. For some players, though, the inconsistency of their real contribution is much more significant.
For one-tricks, the impact of map selection on their real contribution is massive. Since every hero is at least better on some maps than on others, one-tricks will see radical differences in their performance on favorable vs. unfavorable map draws. For extremely specialist one-tricks (namely the builders), this RNG can spell victory or defeat before the assemble screen is over. Even the coin-flip to start on Attack or Defend can be significant; Torb/Sym players can do especially well in the first defense and, in many cases, sap at the enemy team’s will to live. Momentum is actually a very important part of the game. An extremely successful first round is much more likely to produce a victory than a calamitous failure followed by a miracle comeback.
The matchmaker, in its current state, is thus unable to accurately predict the contribution to victory from these players. This significantly reduces the likelihood that matches which include one-tricks will be competitive. If you are placed on a team with a one-trick on a map which favors their hero, their impact will likely be much larger than the expectation of the system and vice versa for an unfavorable map. Perhaps one-trick players accept this randomness and can enjoy the game whilst reducing a significant portion of their games to a coin flip, but for the rest of the players who are matched with them it is a deeply joyless experience.
The natural further consequence of this is that players in a match affected by one-tricking have their Skill Rating distorted as ‘underdog’ teams with favorable map selection are unduly rewarded and ‘expected victors’ are unduly punished. These players then take their distorted Skill Rating into the next game, wherein they are slightly less likely to perform at the expectation of the match-maker. Game quality declines marginally, even for games devoid of one trick players.
The more specialist the character, the more significant the distortion. The more significant the distortion, the worse competitive matchmaking gets. Some would contend that players who main ‘off-meta’ characters are unfairly loathed compared to those who are extremely specialized in a character like Tracer, Zenyatta, or Winston. This is where the distinction between specialists and generalists becomes very important. Although even these characters are to some extent map dependent, describing them as ‘generalists’ is accurate because they are only marginally affected by map and side RNG. There is still a distortion effect, but the impact of matchmaking/map RNG on a generalist is far less significant.
Succinctly, this is why I think that one-tricking specialists ruins the competitive experience: when the true randomness of map selection becomes a crucial determinant of victory or defeat, I lose interest. I don’t play Overwatch because I want to watch particle effects while I flip a coin. I thought we figured this out in season 1.
Let me know your thoughts in the comment section below or on twitter (@jake_overwatch).