/cdn.vox-cdn.com/uploads/chorus_image/image/49134789/2771883-halobeta.0.0.jpg)
Each and every time you are matched in an online game, there's a lot of mathematics going on under the hood. Game companies are investing more time and effort into effective match-making, because it keeps players from straying to rival attractions.
That was the view of Josh Menke, speaking at Game Developers Confrence last week on "Skill, Matchmaking, and Ranking Systems Design." Menke has been working on skill systems for more than a decade, including work on World of Warcraft, Starcraft II, Diablo III, Hearthstone, Call of Duty: Black Ops II and Halo 5. He is currently a senior systems designer at Activision.
He pointed out that balanced matchmaking helps new players, who are often wary of being trashed by superior players when they venture into competitive arenas. But veteran players must also be given satisfying challenges, especially as they are often vocal boosters — and critics — of a game.
"A good skill system finds players' skills really fast and predicts outcomes correctly."
Most games have some sort of skill system that tells players and game-makers the level at which the player is ranked. These might include a kill/death ratio or a numbered ranking based on hours played or on achievements unlocked.
But sometimes these are not enough to predict various skill levels. Bad players can have a decent kill/death ratio if they only ever play against other bad players.
So games companies find other means to match players appropriately, balancing various data points to create a bigger picture. Performance ratings based on the strength of opponents is always best, he said, adding that every system requires constant testing. The outcome should be a classic bell curve that portrays the skill levels of the entire community.
"A good skill system finds players' skills really fast and predicts outcomes correctly," he said. "It should give the right probabilities."
Menke added that the ideal outcome is for "planned experiences of varying intensity" so that all players get a good mix of games that are easy, evenly matched and hard to win. "That keeps the most amount of players in your game having fun," he explained.
He spoke about how new weapons are sometimes introduced to games, and are then taken up and dominated by a small number of players. The developers sometimes react by seeking balance and by nerfing powerful weapons. But Menke said that could be a mistake. Those players ought to be matched together, though he warned that fragmenting the audience between too many modes and specialities risks increasing wait times to games.
One area he worked on was the problem of mixing groups of friends with groups of random people. The friends generally had an advantage over the strangers. He said that every team should create its own skill level so they can be matched correctly. A mediocre team of pals might get a better game against a bunch of strangers with individually high skill levels.
The most important thing is finding the right balance between getting people up and playing, and making sure they get a satisfying challenge. "People don't like to wait," he said.