Author: Peter McKenzie
Date: 16:56:15 12/02/02
Go up one level in this thread
On December 02, 2002 at 13:55:37, Uri Blass wrote:
>On December 02, 2002 at 13:25:28, jefkaan wrote:
>
>>On December 02, 2002 at 08:27:09, Georg v. Zimmermann wrote:
>>
>>>no way. strong humans can even win with handicap (!) on 9x9
>>
>>ok, might be, my statement (that 9*9 is easy for a computer
>>program) was based on having read some info about a program
>>Gotools, which can solve 'life&death' problems within a certain
>>area (with sufficient calculation time) using an alfa-beta
>>algorithm; this life/death engine also is part of the program
>>SmartGo, and as a result i would expect that it would be very
>>strong in 9*9 Go, although apparently such an area (9*9) still
>>is too large for the Gotools a/b engine to solve it exactly.
>>
>>But it would be reasonable to imho expect that within a few years
>>such problems could almost be 'solved' for smaller area's (maybe 6x6 ?),
>>provided the programming work would continue in this field; you
>>can try it out by yourself by downloading it (9x9 version) from
>>www.smartgo.com; the helpfile (tactical analysis) says it can solve
>>enclosed life/death problems "as well as most master('dan')players
>>and often better"; it also "can solve complex ladder problems
>>accurately and quickly". Apparently these features are not
>>sufficient yet to get topresults at a 9*9 board; so..,
>>maybe after 10 years or so ?
>>:)
>>best regards
>>jefk
>
>I believe that it is only a question of money.
>
>I believe that computers on the hardware of today can beat the world champion in
>go if people write the right program.
A very big 'If'!
>
>They only need the right motivation to write the right program.
There is plenty of motivation. The market for Go programs is large, there are
many millions of serious Go players. Of course many are relatively poor
financially, but in Japan alone there are several million potential customers.
The problem is that current programs are so bad that they can't give a serious
player an interesting game.
>
>If someone gives a prize of 100M$ for that target and also spend a lot of money
>about advertisment so at least 50% of the world are going to know about the
>prize then I guess that we will find in less than 5 years a program that beats
>the world go champion.
I know of no factual basis for such an opinion.
>
>I read that in the past there was a big prize for a program that get master
>strength in go(today it is too late to get the prize)
>
>I guess that the problem with that prize is that not enough people heard about
>the prize and the people who heard about the prize are a small minority(I heard
>about the prize only here when it was too late).
No that wasn't the problem. Many many people heard about the prize.
Many smart people were, and are, working on Go programs.
It is a VERY HARD problem.
>
>It is also possible that the prize was not big enough to encourage a big team of
>programmers and go players and mathematicians to work together about the
>problem.
>
>It is possible to decide in case that there is a big prize that a big team of
>people work about the problem and give their ideas and later in case of success
>there is going to be a jury that decides how much money to give to every
>participant based on the discussions.
>
>There should be one programmer in the team that has the last word for the final
>program.
>
>Uri
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.