M
y cousin is a much better Chess player than I am, and I'm a much better Go player than he is. At a lovely little family gathering today, we got to pick each other's brains about the rankings used between Chess and Go, and how one might go about comparing them.
Chess uses a ranking system known as Elo, developed by the Hungarian-American mathematician Arpad Elo. In this system, players trade "Elo points" with one another when they play against one another. If you win against a stronger player, you get more elo points from them; if you win against a weaker player, you get fewer points. The system is designed such that on average, a player will win 2/3 of their games against a player with 100 fewer points than them. Elo ranges from about 200 to the human peak of 2882, achievedby Magnus Carlsen. Modern AI chessbots can reach Elos in excess of 3500.
Traditionally, Go's ranking has made use of the handicap system that is used in Go (and impossible in Chess), in which the numeric difference in one's ranks is simply the number of "free moves" the weaker player can take in the beginning in order to make the game an even match. Thus, a 5 kyu can take 4 handicap stones against a 1 kyu, and the match should be just about even. In Go, kyu rankings typically start at about 25 kyu for the weakest players, go up to 1 kyu, then switch to the "dan" ranks, in which 1 dan succeeds 1 kyu, and proceeds upward to 7 or 8 dan as the maximum for "amateur" players. Professionals have their own ranking system, from 1 dan professional up to 9 dan professional (however, for professionals each rank difference represents perhaps a quarter of a handicap stone, rather than a full stone).
Comparing Elo directly to traditional rankings is hard, but luckily, Elo can be used for any competitive game, and there are Elo ranks for Go as well. However, getting clarity when comparing numbers is still difficult. For one thing, Elo ranks are relative -- the difference between a 500 ranked player and a 600 ranked player is the same as the difference between a 2300 and a 2400. If you just inflated every rank by 10,000 points, the system would still work the same, making it difficult to compare across games.
This is compounded by the fact that in Chess, ties are possible, and in fact increasingly likely the stronger the players are. This means that outright winning a game against a roughly equal opponent is more difficult in Chess than in Go, meaning that Elo ratings tend to get "squashed", especially at the higher levels, since the Elo system is attempting to predict wins, and a tie only counts as "half a win".
Both Chess and Go essentially start players at 100 Elo, and start counting from up from there. The strongest Chess player ever recorded, as mentioned, is Magnus Carlsen at 2882. The strongest Go player is currently Shin Jinseo, with an Elo of 3873. What exactly this 1000 point difference in "best of all time" means is very much up for debate. Is the entirety of it covered by the fact that Go cannot produce draws, and has longer games (making it harder to weaker players to score upsets)? Or is some of it explained by Go being a "deeper" game, in which humans are able to progress further than in chess without running into insurmountable walls?
Certainly, these are difficult questions to answer, and coming up with an unbiased reply is naturally difficult (I'm rather predisposed to Go as a game, for example). However, when explaining my rank to my cousin, the best I can do is probably the simplest one: I'm in roughly the 95th percentile of Go players on the server in question, and on lichess, the 95th percentile means about 2075 Elo. Guess that's what we're going with.