Author: Willie Wood
Date: 21:15:37 12/13/97
Go up one level in this thread
> >Thanks for information. An EARLY beta-version of Alex is downloadable: >http://www.math.jyu.fi/~huikari/alex.exe > >I'll update it often. Interface won't change much anymore. Except I'll >make 'setup position'-feature. I'd like to read comments. > >Remember: EARLY beta-version. (With enormous horizon effects.) > > Jari Hello Jari, Congratulations on your program, called Alex (though I hear that name has been used before). In response to your call for comments, I downloaded it and ran it on my P100. Anyone who has gone to the trouble of writing a working chess program, with a usable user interface, deserves much credit and accolades. I assume you intend to work further on it, and that this is a place of beginning, otherwise these comments will be of no use to you. User Interface -- The board display and pieces are very good. I found no bugs or problems, and it appeared to operate in the way it was designed. However, it was initially a mystery how to operate, make moves, etc. I figured it out, but I do think that you need to provide much better help for the first-time user. You must add mouse support. There isn't any way someone will actually use your game with the arrow-key and enter-key move method. The levels do not add strength as the level number increases. For example, Level 1 is weaker than 2, but level 4 is stronger than 5. This is confusing. Also, the difference between the 30 sec level and 3 minutes is too much. Game Play -- I started with level 5, and it played a very weak game. I switched to level 2 (30 seconds/move), and found the same. I then went to level 6, and it improved quite a bit. However, as you say, it displays a very obvious horizon effect, so much so that it gave up pieces quickly and easily. It also did not seem to understand pawn structure and king safety, allowing doubled isolated pawns and breaking up the castled-side pawn structure without any justification. You seem to have eval or scoring problems. When my program was showing +400, yours showed -0.01! You were two pawns down at the time, in a terrible position. I suppose there are two ways to attack horizon problems. One is to search more deeply, and the other is to use check and capture extensions. I suggest you look closely at the basic methods of move-ordering and cutoffs, which will help you search deeper. Investigate the whole idea of check and capture quiesence. Then think about killer moves, history hueristic, transposition tables and null-move. You're a brave guy to post your program in such an early stage of development. Mine has been under development for six months, and while it plays ok, it needs another few months before I'll even let it play on ICC, let alone posting for others to tear apart. Good luck, and thanks for the opportunity to look at it. WW
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.