Deep Blue rehash and a possible Deep Go

Discussion of chess software programming and technical issues.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
User avatar
jshriver
Posts: 1105
Joined: Wed Mar 08, 2006 8:41 pm
Location: Morgantown, WV, USA
Contact:

Deep Blue rehash and a possible Deep Go

Post by jshriver » Sun Oct 07, 2007 10:21 pm

Article posted today I thought was interesting, and you all might too.

http://www.spectrum.ieee.org/oct07/5552/

Hope I posted this in the right section, mods feel free to move if it's not.

-Josh

PK-4

Re: Deep Blue rehash and a possible Deep Go

Post by PK-4 » Mon Oct 08, 2007 4:47 am

Very interesting article indeed. The idea of packing a deep blue into a single chip was a little tough on imagination.

P.K.Mukhopadhyay

Mark
Posts: 216
Joined: Thu Mar 09, 2006 8:54 pm

Re: Deep Blue rehash and a possible Deep Go

Post by Mark » Thu Oct 11, 2007 1:52 am

jshriver wrote:Article posted today I thought was interesting, and you all might too.

http://www.spectrum.ieee.org/oct07/5552/

Hope I posted this in the right section, mods feel free to move if it's not.

-Josh
Thanks for the link, Josh. Interesting article!

Mark

User avatar
Bill Rogers
Posts: 3562
Joined: Thu Mar 09, 2006 2:54 am
Location: San Jose, California

Re: Deep Blue rehash and a possible Deep Go

Post by Bill Rogers » Thu Oct 11, 2007 4:19 am

The really funny funny thing is that GO initally looks so simple to code.
I had seen some early attempts of GO code and was supprised what little it took. Unlike chess where there are 6 different pieces that all move in different directions with different rules attached, all GO pieces are the same and they don't move at all. But the amazing thing is as simple as it looks the advancements in GO have been going so slow only go to show just how complicated the game really is.
Bil

nczempin

Re: Deep Blue rehash and a possible Deep Go

Post by nczempin » Thu Oct 11, 2007 8:15 am

Bill Rogers wrote:The really funny funny thing is that GO initally looks so simple to code.
I had seen some early attempts of GO code and was supprised what little it took. Unlike chess where there are 6 different pieces that all move in different directions with different rules attached, all GO pieces are the same and they don't move at all. But the amazing thing is as simple as it looks the advancements in GO have been going so slow only go to show just how complicated the game really is.
Bil
This observation is true even in the human realm: Take any given person who does not know either game, and see which is easier to teach.

But trying to get your head wrapped around Go is much harder (IMHO, having played both) because chess is a much more tactical game, and thus it is easier to determine exactly at which point you lost or won the game.

And yes, it would take much more code to get a chess engine working a basic alpha-beta than it would to support the same in Go (although I have only actually tried one of the two).

yoshiharu
Posts: 56
Joined: Sat Nov 11, 2006 10:14 pm

Re: Deep Blue rehash and a possible Deep Go

Post by yoshiharu » Thu Oct 11, 2007 11:42 am

jshriver wrote:Article posted today I thought was interesting, and you all might too.

http://www.spectrum.ieee.org/oct07/5552/
Has anybody any reference of what he calls the "method of analogy" algorithm?

Cheers, Mauro

bob
Posts: 20916
Joined: Mon Feb 27, 2006 6:30 pm
Location: Birmingham, AL

Re: Deep Blue rehash and a possible Deep Go

Post by bob » Thu Oct 11, 2007 11:58 am

It came from Botvinnik's book "Computers, chess and long-range planning". published by Springer-Verlag in 1970. Dates _way_ back. Most of it appears to be extremely difficult to implement, particularly since his program "Pioneer" never worked at all.

Tony

Re: Deep Blue rehash and a possible Deep Go

Post by Tony » Thu Oct 11, 2007 12:05 pm

yoshiharu wrote:
jshriver wrote:Article posted today I thought was interesting, and you all might too.

http://www.spectrum.ieee.org/oct07/5552/
Has anybody any reference of what he calls the "method of analogy" algorithm?

Cheers, Mauro
It was implemented in Caissa, a Russian chesprogram (1974 ?) They have published some stuff about it.

Tony

User avatar
hgm
Posts: 24740
Joined: Fri Mar 10, 2006 9:06 am
Location: Amsterdam
Full name: H G Muller
Contact:

Re: Deep Blue rehash and a possible Deep Go

Post by hgm » Sat Oct 13, 2007 11:39 am

Bill Rogers wrote:The really funny funny thing is that GO initally looks so simple to code.
I had seen some early attempts of GO code and was supprised what little it took. Unlike chess where there are 6 different pieces that all move in different directions with different rules attached, all GO pieces are the same and they don't move at all. But the amazing thing is as simple as it looks the advancements in GO have been going so slow only go to show just how complicated the game really is.
Bil
The rules for capture in Go are not that simple. You would have to do a boundary scan on each chain that borders a newly placed stone.

I strongl doubt that a Go program could be written that is as small as micro-Max...

User avatar
Zach Wegner
Posts: 1922
Joined: Wed Mar 08, 2006 11:51 pm
Location: Earth
Contact:

Re: Deep Blue rehash and a possible Deep Go

Post by Zach Wegner » Sat Oct 13, 2007 5:32 pm

hgm wrote:I strongl doubt that a Go program could be written that is as small as micro-Max...
I do not. I have a fully functional UCT program that is about 500 lines total, that I plan to compress down a lot. Right now it has many unnecessary lines and long variable names.

Post Reply