can chess program see win for white after e4 h6 d4 a5?

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

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
User avatar
Brunetti
Posts: 112
Joined: Tue Dec 08, 2009 12:37 pm
Location: Milan, Italy
Contact:

Re: can chess program see win for white after e4 h6 d4 a5?

Post by Brunetti » Fri Apr 02, 2010 10:29 am

muxecoid wrote:Against an engine with a good general purpose book is it better to play h6, a5 instead of having no opening book at all?
No, it's better no book. The engine will play reasonable positional moves, like ...d5, ...Nf6 and so. No the best ones, but far better than ...h6 and ...a5.

Alex

Milos
Posts: 3389
Joined: Wed Nov 25, 2009 12:47 am

Re: can chess program see win for white after e4 h6 d4 a5?

Post by Milos » Fri Apr 02, 2010 11:20 am

lkaufman wrote:Perhaps you are unfamiliar with the MC feature of Rybka 3; there should be no duplicates, and the variety should be quite good. If you are simply running games relying on the nondeterministic nature of MP to get some variety, your sample will be vastly less varied than mine. Most likely, you are getting many games that are identical for several moves, and probably White has already made one or more poor moves in those moves. For example, Rybka loves to play Nc3 and Nf3 very early, in front of the c2 and f2 pawns, but in this case this is probably poor strategy. The MC method insures that a wide variety of opening schemes will be tried.
I'm aware of Rybka MC, moreover I have a pretty good idea how it's implemented. In theory it should not impact the engine strength for more than 5-10elo, but IMO it has a certain big drawback. It varies certain evaluation factors following uniform probability distribution. So when there are a few close moves in score not necessarily the best one will be chosen. And that does not happens only on high depths but at any node. Instead far more realistic approach would be what MP and OS do, to very the number of nodes evaluated. When you use fixed ply you remove OS variety component. The conclude I'm not a big fan of automatic MC methods implemented in chess engines and I prefer natural variety.

playjunior
Posts: 338
Joined: Thu Jun 21, 2007 10:53 pm

Re: can chess program see win for white after e4 h6 d4 a5?

Post by playjunior » Fri Apr 02, 2010 11:56 am

Milos wrote:
lkaufman wrote:Perhaps you are unfamiliar with the MC feature of Rybka 3; there should be no duplicates, and the variety should be quite good. If you are simply running games relying on the nondeterministic nature of MP to get some variety, your sample will be vastly less varied than mine. Most likely, you are getting many games that are identical for several moves, and probably White has already made one or more poor moves in those moves. For example, Rybka loves to play Nc3 and Nf3 very early, in front of the c2 and f2 pawns, but in this case this is probably poor strategy. The MC method insures that a wide variety of opening schemes will be tried.
I'm aware of Rybka MC, moreover I have a pretty good idea how it's implemented. In theory it should not impact the engine strength for more than 5-10elo, but IMO it has a certain big drawback. It varies certain evaluation factors following uniform probability distribution. So when there are a few close moves in score not necessarily the best one will be chosen. And that does not happens only on high depths but at any node. Instead far more realistic approach would be what MP and OS do, to very the number of nodes evaluated. When you use fixed ply you remove OS variety component. The conclude I'm not a big fan of automatic MC methods implemented in chess engines and I prefer natural variety.
But does that change the expected score for a certain position? Because even if we assume the implementation is flawed it should be both way, i.e. should not give any side an advantage. On the contrary, your tiny sample size (with normal games) may result in significant distortion of the expected score.

Larry's testing method may yield higher variance but probably also a better expected value estimate. First, because he can have much larger sample size, second, because his method is less sensitive to obvious evaluation mistakes (like playing both Nc3 and Nf3).

lkaufman
Posts: 3837
Joined: Sun Jan 10, 2010 5:15 am
Location: Maryland USA
Contact:

Re: can chess program see win for white after e4 h6 d4 a5?

Post by lkaufman » Fri Apr 02, 2010 1:52 pm

I don't have data on this, but I would expect that having no book vs. a general book would lead to full equality when the no book side has white, so we can estimate the value of the book at around 35 Elo. If this is correct, playing Black it might be reasonable to open 1...a6 for example which would "almost" kill the book, since after one or two normal book moves the engine would probably leave book as there should be minimal book on this line. It would certainly be bad to play two poor moves rather than one.

Post Reply