Author: Robert Hyatt
Date: 08:16:39 01/06/05
Go up one level in this thread
On January 06, 2005 at 05:53:24, David B Weller wrote: >Hi Eric, > >Thanks. > >BTW, that should say GES_134, but I never changed the name until after the >tourney > >Anyway, it seems I am forever finding logical bugs in my code - which sometimes >makes significant improvements. > >If I had it to do over, I would try and dicipline myself to use much debugging >code and tools and be sure the first time through, that everything always did >what I thought it did. > >Mobility. > >As I watched GES lose against Fruit[over and over], it occured to me that Fabien >used mobility [even said it was a marked improvement, iirc] so I started looking >for that. >Low and behold, what Fruit was doing to Ges was literally robbing it of all >'motion'. - It makes sense, if your pieces have no moves, then they cant have a >best move iether! >So I added a relatively cheap and easy mobility term [which btw hasnt been >tweaked at all] and POW! GES_134 > >Prof. Hyatt once commented that it wasnt clear to him, whether mobility was a >cause or effect - and I guess I kinda just took his word for it, but I think it >turns out to be one of the things that makes Fruit so strong. Remember that I _also_ said that there are lots of ways to do mobility without doing mobility. A rook on an open file is a simple example. If the rook is moved to an open file, it obviously has increased mobility. Knights on good outpost squares are the same, as opposed to being on the edge of the board or in the corner. The only real mobility I do is for bishops. The first versions of Crafty used to have both attacks_to and attacks_from incrementally updated, which gave me mobility (and move generation) for free. However the program got better as I eliminated raw mobility in favor of things that were more specific, like the open file for rooks, above... In other cases, mobility is actually bad. IE for your king. :) > >I am beginning to think of it this way: A chess program with mobility term, is >like a Rubics cube with well 'warn' parts. The 'wear' doesnt help solve the >puzzle, but makes it easier for the user to turn the sides. >OTOH, an engine that neglects mobility, is like a new cube which is always >giving you trouble to turn. The engine seems to lack the ability to freely >'rearrange' its pieces to fit the ever changing circumstances. > >Fluidity was a term that came to my mind. > >-David > >http://home.comcast.net/~opraus
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.