Life In 19x19

Below is a quick analysis. It looks at the rate of games won by resignation in about 43,000 even games played since 2005 in my data base. It is divided between komi (as a proxy for rule sets) where a komi of 6.5 indicates a Japanese, Korean, or Taiwanese game while a komi of 3.75 or 7.5 indicates a Chinese game. The Chinese games have a resignation rate of 79% while the 6.5 point komi games have a resignation rate of 68%.

The whole thing started because I suddenly wondered whether or not the high rate of resignations that I think I see in Chinese professional Go might be due to the extra care needed in filling in of all liberties and the counting process thereafter. The numbers seem to indicate a difference between play in China and elsewhere but of course the numbers alone will not show whether this has anything to do with rules versus culture, for example.

When I finished that part of the analysis I realized there turned out to be a few hundred games unaccounted for. These turned out to be mainly komi 8 games. These seem to be Ing-rule games. I have never played in an Ing-rule game so I have no clue how games finish or scoring is done. However, since the resignation rate was 69%, I cheated and put those games with the 6.5 komi games to protect the story that I am telling.

Any thoughts on what the difference might really reflect?

If you want to tease out the effect of ruleset rather than nationality/culture of the player, how about looking at resign rates of Koreans vs Chinese in the Chinese league and Koreans vs Chinese in Korean international tournaments (LG, Samsung); though I suppose it might be difficult to query such information and there are not so many games to get good stats...

If there is enough data, it would also be interesting to plot out the fraction of non-resigned games with each numerical result (adjusted appropriately for ruleset), which might give a clue as to what part of the distribution those resigned games are coming from.

Quite a while back I made the suggestion that Chinese players were more apt to resign (apparently fact - seemed to be borne out by figures) and that this was because of the tedium of Chinese scoring (pure speculation, though after having heard that some Chinese pros use Japanese counting).

This was all based simply on an impression from inputting games into the GoGoD database and a little pencil-and-paper analysis.

As this was some time ago, when Mickey Mouse tournaments were much rarer, I had no reason to assume time limits were a factor, although I did observe that Chinese games were then generally a little faster than in Japanese and even Korean events.

FWIW I now have a slightly different impression (stress, impression only but based on a lot of games) that the great resigners of go are now Koreans (whoa, Kirby!), and above all in fast games. I posit two reasons for this. One is that fast games are more mistake ridden. The other is that there is a subtle pressure from tv producers to skip the counting stage if possible, though when I have seen games on tv in Korea, the counting is done with remarkable speed, so I'm not a great adherent of that theory.

There may be cultural factors such as young players resigning prematurely against respected elders, but that particular doesn't strike me as significant.