Many competitive online multiplayer games, such as Ubisoft’s Rainbow Six and Sony’s SOCOM II: U.S. Navy SEALs, use a ranking system based on ELO. Designed for chess by Arpad Elo and implemented by the U.S. Chess Federation in 1960, the system has some shortcomings, but with some tweaks is a reasonable and reliable method for rating competitors in games in which one player faces off against another.
As such, it has been adopted for the ranking of many competitive games like Go. However, with no provisions for multiple-player games, those wanting to apply it in this realm have had to bolt on additional logic. SOCOM II, one of the most popular online games of all time, needs to factor in variables: relative team strengths, relative team sizes, multiple-round matches, and individual player achievements such as rescuing hostages or defusing bombs.
By extending the ELO system to work for a team-based game, these tweaks have resulted in exploits enabling gamers to falsely improve their rankings. For example, it’s possible—with some willing participants—to create games with only one higher-ranked player facing off against a team of lower-ranked players. By his killing all but one of the opposing team and then submitting to being killed by the sole remaining opponent, the high-ranked player ups his ranking for being outnumbered and killing so many despite losing, whereas the opposing team gains points for killing a high-ranking opponent and winning.
Situations such as these inevitably result in criticism for the game in question. With TrueSkill, Microsoft is attempting to put in place a single reliable ranking system designed to work for any type of online multiplayer games.