computer against humans with bigger boards

Discussion of anything and everything relating to chess playing software and machines.

Moderator: Ras

Uri Blass
Posts: 11222
Joined: Thu Mar 09, 2006 12:37 am
Location: Tel-Aviv Israel

computer against humans with bigger boards

Post by Uri Blass »

I wonder if bigger board can give humans some years of superiority against computers or not.

Bigger board can be for example 29*29 in go or 16*8 in chess(the generalization in chess for more than 8 ranks when you have 8 files is simple).

Of course we need some match with big prize money and one year to prepare both for humans and computers.
Dirt
Posts: 2851
Joined: Wed Mar 08, 2006 10:01 pm
Location: Irvine, CA, USA

Re: computer against humans with bigger boards

Post by Dirt »

I don't think a larger board will help humans relative to computers in chess. I think it will help computers more.

Go is a tougher question. I would guess it would help the human a little, but the boards are already large enough that there is not much to be gained.
Deasil is the right way to go.
User avatar
Evert
Posts: 2929
Joined: Sat Jan 22, 2011 12:42 am
Location: NL

Re: computer against humans with bigger boards

Post by Evert »

Uri Blass wrote:I wonder if bigger board can give humans some years of superiority against computers or not.

Bigger board can be for example 29*29 in go or 16*8 in chess(the generalization in chess for more than 8 ranks when you have 8 files is simple).
It's not so obvious to me how you'd generalise chess to a 16x8 board. I can think of several ways to do it and none of them seem to me more obvious than the other.

I don't think it would help people. If you do not add stronger pieces, the game becomes slower, which means the human has to keep his concentration longer. If you do, the game becomes more tactical, which computers are generally better at.
Uri Blass
Posts: 11222
Joined: Thu Mar 09, 2006 12:37 am
Location: Tel-Aviv Israel

Re: computer against humans with bigger boards

Post by Uri Blass »

Evert wrote:
Uri Blass wrote:I wonder if bigger board can give humans some years of superiority against computers or not.

Bigger board can be for example 29*29 in go or 16*8 in chess(the generalization in chess for more than 8 ranks when you have 8 files is simple).
It's not so obvious to me how you'd generalise chess to a 16x8 board. I can think of several ways to do it and none of them seem to me more obvious than the other.

I don't think it would help people. If you do not add stronger pieces, the game becomes slower, which means the human has to keep his concentration longer. If you do, the game becomes more tactical, which computers are generally better at.
It is obvious for me.
Every side has the same pieces as normal chess(black king at e16 and black queen at d16 (instead of e8 and d8) and black pawns at the 15th rank if we take about 8*16.
pawns can go forward one or 2 squares in the first moves and later exactly one square.

pawns needs to go more moves to promote of course but castling and all the rules are the same.

I think that when the game is more complex it is easier for humans relative to computers.

Bigger branching factor means that the computer can calculate less forward with brute force.
User avatar
Evert
Posts: 2929
Joined: Sat Jan 22, 2011 12:42 am
Location: NL

Re: computer against humans with bigger boards

Post by Evert »

Uri Blass wrote: It is obvious for me.
Every side has the same pieces as normal chess(black king at e16 and black queen at d16 (instead of e8 and d8) and black pawns at the 15th rank if we take about 8*16.
pawns can go forward one or 2 squares in the first moves and later exactly one square.

pawns needs to go more moves to promote of course but castling and all the rules are the same.

I think that when the game is more complex it is easier for humans relative to computers.

Bigger branching factor means that the computer can calculate less forward with brute force.
Oh, you meant 8x16. That is indeed a bit more obvious, but at the same time it makes the problem I mentioned worse because pawns need to advance much more before the armies come into contact.
A game is easier for humans relative to computers if it is more complicated because of strategy (and then only if you cannot train a neural net that does the strategy very well), not if it is more complicated because of tactics. Computers are good with tactics.

Either way, this could be tested easily; SjaakII can play games on an 8x16 (or 16x8) board. Of course it is no where near strong enough to challenge a good human player on 8x8, but it could be tested if it is easier for a human to play on 8x16 compared to 8x8. I rather suspect not though.
User avatar
velmarin
Posts: 1600
Joined: Mon Feb 21, 2011 9:48 am

Re: computer against humans with bigger boards

Post by velmarin »

8 x 16, seems bad for bishops and Queens, diagonale are not symmetrical.

I like the 16 x 16, also can I play at FICGS
http://www.ficgs.com/
Image

WinBoard boasts his big chess version and Daniel shawul has nebiyu engine...
There are few active players but it is very interesting to me...
Dirt
Posts: 2851
Joined: Wed Mar 08, 2006 10:01 pm
Location: Irvine, CA, USA

Re: computer against humans with bigger boards

Post by Dirt »

Uri Blass wrote:Bigger board can be for example 29*29 in go...
Why odd by odd, anyway? Particularly on small boards, even by even works better.
Deasil is the right way to go.