Author: Bruce Moreland
Date: 13:18:37 09/30/99
Go up one level in this thread
On September 30, 1999 at 11:35:07, Torstein Hall wrote: >Just a probably stupid question to satisfy my curiosity! >Do you have to extend after the null move? Lets say the oponent can take a >Bishop with a Queen, but that would loose the Queen next move etc. etc.? > >If you not extend, would that not keep reasonable moves down "on the list" and >keep you from seeing a lot of good moves? I don't understand what you are asking but I'll try to answer it anyway. You enter a node, and you're all ready to search all of the moves here to, let's say, 6 plies. The case we are interested in is the one where you are massively winning, so you will not actually search all of the moves. You'll search the first one, and the resulting position will be awesome, so you'll quit. That's just normal alpha-beta search. But that's still one move searched to 6 plies, just to tell you that you are winning. What you do with the null move is assume it's the opponent's turn to move and search to some reduced depth, probably most commonly 4 plies in this case. In the case I'm talking about, maybe the opponent can find a salvation if they get to move here, in which case you ignore the result of the null move, and search the rest of the moves normally. But perhaps they are losing, and getting to make a move doesn't help. In this case you assume that if you made a move you'd be even better, so you quit. To try to answer your question, this search is shallower, but it's still a fairly strong search. Letting the opponent move first is usually an advantage, so the result from this search, if it lets you quit (fail high), is usually pretty accurate. You get burned in two cases: 1) You are in zugzwang, meaning your position will disintegrate if you move, but it is fine if you don't. In this case, you could search 50 plies instead of 4, and you'll still think your position was dandy. So in this case the null move is terrible. You get burned because passing a move isn't legal in chess. If it was legal, this problem wouldn't happen. 2) There is a threat against you, which is a little too deep to see with the reduced depth search. Your opponent can rip you in 5 plies, but the 4-ply search didn't see this. If you'd done a normal 6-ply search, you get to make a move, but since the threat is unstoppable perhaps this move doesn't do any good. Now the opponent gets 5 plies in response, and that's enough to see that they can kill you. So in this case the null move overlooked an unstoppable threat, which has nothing to do with zugzwang. Unfortunately a common case involves unstoppable mates, so when this case comes up it is often a disaster. Both of these situations come up, but in the typical case they don't outweigh the advantage of being able to go 10x faster in dumb positions, of which there are a lot. bruce
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.