Author: Daniel Clausen
Date: 05:17:12 10/08/03
Go up one level in this thread
On October 08, 2003 at 07:37:03, Dan Andersson wrote: >One can also argue about how knowledge based the approach was. As I remember it >I was less than impressed of the game specific insights while reading it. Surely you can argue about that. And for different persons will be more/less impressed when reading it. Before I read the paper, I didn't know a lot of positional knowledge about the game (apart from building up threats on the right rows) and was impressed how many positions turned into trivial positions once you know a little bit more about the game. :) Most people think there's nothing positional about it but just pure luck/calculation, which in essense even is true. :) I think there are two big reasons, why there's this knowledge about ConnectFour which allows you to solve the game, but there's no such knowledge yet about chess. (1) Zugzwang plays a central point in ConnectFour. A lot of positions change their game-theoretical value (+1, 0, -1) when side-to-move is flipped. I'm not sure how high the percentage in a more or less equal ending in chess is, that the result depends on zugzwang. I would be interesting to calculate that for known endings. I assume it's very high when it comes to pawn-endgames, but also higher than what most people think when it's not a pawn-endgame. (2) Each move in ConnectFour goes towards the sure end of the game. You can't do loops (with move reperition) like in chess. I guess that (together with non-scientific rules like 50-move rules) makes it much harder to find general patterns for chess. (and with these pattern I don't mean the patterns your average chess engines uses for their evaluation, but patterns which really help you to _solve_ the game) > I had more use of Breukers Memory versus Search Ph. D. thesis. The name rings a bell but I'm not sure I've read it. Is it publically available on the net? Sargon
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.