Re: computer-go: GoeMate is now 1st in 13x13-CGoT

[stefanmertin@xxxxxxxxxxx (Stefan Mertin)]

> > Please have a look at the games and tell me your opinion:
> > Does a tournament like this show the real playing strength of the 
> > programs or is there still too much of luck and chance? 
>
> Playing more games certainly reduces the element of luck. If you look 
> at how many different winners there have been in recent computer go
> tournaments in the past couple of years you could either conclude
> that there are five or six programs incredibly close in strength
> or that the contests have too few games to really find the strongest.
> I personally would conclude the latter.

Yes I agree - 
thatīs why in my actual tournament I want to play hundreds (!) of games 
and thatīs why I reflected my next tournaments to be "open"
i.e. I will try to play as many games (sets of n games!) as possible 
and the result of each pairing will be 
a percentage (not a number) of won games. 

> > Is it possible that medium-board-games on 13x13 may give us already
> > a good rating list or would a 19x19-result look very different?
>
> It could look very different, mainly because of the way programs 
> handle joseki. Many 19x19 josekis will be too big for 13x13 boards
> so how will programs cope with this? I guess many will just abandon
> their joseki systems altogether putting them at a great disadvantage.
> Some may have automated systems for attempting to use only the 
> josekis that fit. Another thing to consider is that, because 13x13 
> is not a popular boardsize, many programmers won't have tested 
> their 13x13 opening systems very carefully.
>
> Mick.

Yes and no -

For sure a program with a good 13x13-joseki-database and a carefully 
tested opening system will have a big advantage in medium-boardsize Go
compared to all the programs that are only optimized for 19x19.

So I am a little bit hopeful that my tournament will contribute in
making the 13x13 boardsize a little more popular...   :-)

But the more games I see the programs play,
the more I have the impression that only very few games
(I would say perhaps 10%) are decided in the opening -
very much games are decided in the endgame (40-50% ?)
and also very much games by errors or tactical abilities in middlegame...

Thatīs why I asked: I am only a +/- 8 kyu player myself,
but I would say the Go programs still are making so much 
bad moves, errors, or even blunders that 
neither opening nor boardsize are so much important for win or loss!

This might be different for stronger programs -
I already did play some first games with your Go4++
and it seems to be at least one of the strongest programs (!):
it really wins most of the games, where it gets an advantage,
so perhaps opening abilities become much more decisive.

Regards,
Stefan