[Computer-go] CrazyStone in the 5-dan footsteps of Zen

Jeff Nowakowski jeff at dilacero.org
Sat Jan 7 13:32:14 PST 2012


On 01/07/2012 03:07 PM, Jouni Valkonen wrote:
> Jeff, it does not matter what ten humans does with the games, because
> they are insignificant compared to bulk mass of games that gobots are
> playing. therefore CS's it has no relevance for the rating system.

Maybe, but you may be underestimating how much damage a few dedicated
players can do. This one player accounted for about 8% of the games
against CrazyStone since November, and because he lost so many games
against solid bots, he affects the rank even more than a regular player.

> Also 6 months is very very short time, and when old games are
> starting to expire, KGS rating will follow. See e.g. CS rating graph
> for reference.

I think waiting months for old games to expire when you make an
improvement is not so good. If I was running bots on KGS I'd have 7
rotating accounts, and only play an account for a month so that it
wouldn't get heavy.

> KGS rating system works brilliantly and it is almost impossible to
> exploit,

It's a lot easier than you think, and I've reported a few people where
something is clearly being abused, whether it is sandbagging or cheating
to get a higher rank.

> 6d for MC-gobot seems a bit optimistic, I would bet €5 that it won't
> happen in 2012. 6d's are ridiculously strong. I calculated that in
> January, out of 56 games against 5d's CS won 48%. This is indeed
> impressive, although there is long way to beat 6d's in even game.

Zen19d is already winning 64% out of 59 games that are even and 56% out
of 27 against higher rank since November:

http://kgs.gosquares.net/index.rhtml.en?id=Zen19d&id2=&r=1&ty=2011&tm=11

I would be very surprised if it isn't 6d in 2012.




More information about the Computer-go mailing list