[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [computer-go] I know we disagree,but I choose to do nothing about it.




From: David G Doshay <ddoshay@xxxxxxxxxxxxxxxxx>
Reply-To: computer-go <computer-go@xxxxxxxxxxxxxxxxx>
To: computer-go <computer-go@xxxxxxxxxxxxxxxxx>
Subject: Re: [computer-go] I know we disagree,but I choose to do nothing about it.
Date: Fri, 22 Jul 2005 21:48:33 -0700

On 22, Jul 2005, at 8:58 PM, drd@xxxxxxxxxxxxxxxxx wrote:

ddoshay@xxxxxxxxxxxxxxxxx wrote:
And because any bot that chooses not to score correctly the first time
can force this protocol upon the opponent, it is not much different
than a change in the rules of Go. Or maybe I don't understand either.
I would say you don't understand.

If 2 players sit down to play a game for fun, and they have an honest
disagreement over dead stones,  what do they do?   They play it out!
I was trying to avoid going back to the arguments we had earlier, but
here we are. When two people disagree over the status of stones at the
end of a game they discuss which stones are disputed. Then the players
determine the life or death of those specific stones. This is what I had
suggested earlier. My objection is to a protocol whereby if there is any
disagreement then each side is "strongly encouraged" to play until
*all* dead stones are removed, because after two more passes *every*
stone on the board will be considered alive.
I understand your objection here and agree that in an ideal world the protocol would be a little more sophistocated and 'human like'. For example, there was a suggestion for a protocol which lets the programs have multiple attempts at agreeing which stones are dead - effectively allowing multiple iterations of the currently suggested protocol.

The problem with a more sophistocated protocol is that I suspect even fewer programmers will implement it. I would much prefer the scenario where the majority of programs can play fully automated games using a simpler protocol than the scenario where only a few programs can play fully automated games using a fancier protocol. I would also say that the currently proposed protocol is more 'human like' than not being able to finish a game at all!

I think that the currently proposed protocol adequately (for now) solves the problem of allowing programs to play (and finish/score) fully automated games in a nice simply way. I also see it as a stepping stone towards the 'ultimate' protocol which could be introduced at a later date and would get greater traction once programmers have experience with the simpler approach.

cheers,
Peter

<snip snip>

_________________________________________________________________
Shop ‘til you drop at XtraMSN Shopping http://shopping.xtramsn.co.nz/home/

_______________________________________________
computer-go mailing list
computer-go@xxxxxxxxxxxxxxxxx
http://www.computer-go.org/mailman/listinfo/computer-go/