[Computer-go] Practical significance?

Mark Boon tesujisoftware at gmail.com
Mon Nov 26 16:18:21 PST 2012

On Mon, Nov 26, 2012 at 1:09 PM, Kahn Jonas <jonas.kahn at math.u-psud.fr> wrote:
> That's not too far from the winrates between Cho Hunhyun and Seo
> Bongsoo.
> The latter is far from the glory of the former, but when they were by
> far the two strongest players in Korea, Seo has won his share of titles.

To be honest, I don't have solid statistics to back it up. Having said
that, I'd be surprised if the real winning percentage of Cho Hunhyun
vs. Seo Bongsoo would be 62:38. How often did they play?

I remember in the old days the Go World magazine would publish the
winning rates of the top players in Japan. If I remember correctly,
the handful of players dividing the bulk of the titles over a decade
would still only have a winning rate of about 55% over the whole year.
Of course this is partly caused by top players playing among top
players. But they also still play a lot against lower-tier players in
for example the qualifying rounds of tournaments.

The other thing you have to take into account is this: if a 'weaker'
player played lots of 5-game matches against a top player, he might
occasionally win one set. But the qualifying systems of these
tournaments are so extensive that the weaker player very rarely makes
it through. So the weaker player doesn't get a lot of shots to try.
Occasionally it happens that someone wins a title as a one-time off.
Most titles are divided among a small number of top players while the
majority of 9p players never win any. But I'm hazarding the guess that
if you'd pit a top player against a 'weaker' 9p player who never won
anything major, the winning percentage of the top player would be less
than 62%. But it's still only a guess, I admit. If someone has
statistics at hand I'd be interested to see.


More information about the Computer-go mailing list