Re: [computer-go] future KGS Computer Go Tournaments - two sections?

["Jim O'Flaherty, Jr." <jim_oflaherty_jr@xxxxxxxxx>]

Daniel,

When you wrote "One the other hand no matter how many copies of the same 
program participate, it doesn't really change the odds of wining.", I 
think you are making the assumption that randomness does not play a 
factor in the game outcomes.  I do not mean an disrespect to those who 
have produced computer players...at this point, I would venture a guess 
that randomness is a large influencer in specific games between two 
computer players.  So, a person having more his program participate as 
more than a single entry does in fact elevate one's "odds" in terms of 
"getting lucky".

For example, if you had a tournament of five players all consisting of 
GnuGo 3.6, I would be willing to bet you would not get a uniform result, 
given games were constrained by time (as they are in the tournaments) 
and not by move search.  Even though it would be the same "engine", 
unless the system is highly deterministic and not impacted by slight 
variations in timing constraints, my guess is repeated game sequences 
between the same two players would diverge.  Has anyone tried this 
experiment to see what would happen?

And it only takes one of the two players in a game to be "slightly 
random" for the particular game to be "unrepeatable" for both.  So I 
sense luck is still a pretty significant factor in computer go 
tournament outcomes.  Perhaps someone has done some extensive multiple 
game testing to falsify my assertion?  And does anyone know how to go 
about measuring how much luck is an influencer in a particular 
tournament game and then for the entire tournament?

I do like your idea of charging some nominal fee (perhaps through 
PayPal?) to help regulate submissions, if they are getting out of hand.


Jim O'Flaherty


Compgo123@xxxxxxxxxxxxxxxxx wrote:

> In a message dated 5/10/05 3:21:25 AM Pacific Daylight Time, 
> tesujisoftware@xxxxxxxxxxxxxxxxx writes:
>
>
> > The question of how to deal with derivative work has always been
> > difficult.
>
>
> One solution is to charge $30 for entrance fee. I wish 1 million GnuGo 
> derivatives  would participate.
>
> One the other hand no matter how many copies of the same program 
> participate, it doesn't really change the odds of wining. Unless your 
> program has the exactly the same playing strength as GnuGo. In this 
> case you may as well join the crowd and use a copy of GnuGo to 
> particite, beacsue it makes no difference.
>
> Daniel liu
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> computer-go mailing list
> computer-go@xxxxxxxxxxxxxxxxx
> http://www.computer-go.org/mailman/listinfo/computer-go/
_______________________________________________
computer-go mailing list
computer-go@xxxxxxxxxxxxxxxxx
http://www.computer-go.org/mailman/listinfo/computer-go/