Comments by GregoryStrong
I would like to second David's comments, particularly regarding the pawn promotion. You have created a new piece combining all moves which can only be attained by pawn promotion; I see little reason to offer under-promotion. You already have to have an Amazon piece in the set, so you shouldn't need the option to promote to weaker pieces for that reason. Also, the more promotion options you allow, the slower computer programs which play the game become. The more promotion options there are, the more legal moves there are, and the larger the search tree becomes.
I do think this game looks interesting, though. I like the starting array, especially the symmetry. I'll post a Game Courier invitation shortly, and give it a try...
Thank you, Robert. Actually, there is something that you (and others) can do to help. If you beat it (as you did in Grand Chess), and if it's obvious how you did it, that information could be helpful to me. It should be possible to tweak the evaluation function to fix strategical weaknesses, provided we can identify those weaknesses. A save-game file could be helpful to me, too. Of course, I realise that often a win is achieved by staying slightly ahead throughout the game, and not by exploiting some weakness, but if you (or anyone else) does identify some specific weakness, I would like to know about it! In any event, if anyone beats ChessV in a game in which it is able to calculate to a reasonable depth (depending on time and speed of your computer), I would appreciate it if you could e-mail the save-game file to me!
Swapper question ... Can a swapper perform 'mutual destruction' with a friendly piece? The rules aren't clear on this. Thanks!
Ohhh... Thank you for pointing this out! I will have to update ChessV to correct this. Actually, I'd like to add support for the Duniho variant, too (now that I'm aware of its existance.) Is this OK Fergus? and what are the additional rules? Thanks! Greg LATER EDIT: I had posted this before Fergus had posted his additional comment. Now I'm really confused ... :)
Hmmm... I do like the idea of the Enhanced Castling presented here, but it might be a bit too powerful. How about a half-enhanced castling, in which the King can go as far as desired, but the Rook must then always be placed just on the other side? That should be a little easier to program, and I think I like the rule better, too... I don't like giving the King a free choice of any square, *and* letting the Rook pick the open file, all on the same move.
Thank you for your feedback, and I sincerely do appreciate it. May I ask which game you were playing? ChessV is better at some games than others. It would also help me if you could tell me which operating system you use. I seem to have problems with Windows 98/ME that I don't have with 2000/XP. Also, save-game files of any game in which you win are helpful. Thanks again! If you are having problems with bugs, please wait for a couple of new versions. I am about to stop adding new games, and concentrating on fixing all bugs. When I feel it is a stable as possible, I will release version 1.0
Sorry for the delay, Roberto, I've been distracted by a couple of things ... I will have version 0.7, Ultima-enabled, posted either late this evening, or tomorrow. Of course, I'll make a post here when it's up. Thanks for your continued interest!
I haven't given a game a 'poor' rating yet, but I really can't give this game anything else. The first thing I think when I look at this is 'Isn't there ANYTHING about the game of Chess that was ok as-is?' He changed the number of files, the number of ranks; changed the move of the Rook, the Bishop, the Pawns (no enpassant) ... He re-arranged the pawns! He doubled the number of Queens!!! And then there's the barrier pawn, which might make center-play more interesting, but boy is it nothing like a 'normal' Chess piece. And no resigning?!? I won't even comment on that one. On the up-side, yes, he did add symmetry, but I just can't see giving it a 'good' rating. It just looks like an extreme over-reach that wasn't all that well thought-out. Of course, I must admit that I haven't played it (yet) ... It is possible that my opinion would improve.
Since you all provided so much input into evaluation, I thought you might be interested in the various terms I used in the Ultima evaluation function for ChessV... I used George Duke's piece values of Pawn = 1000, Withdrawer = 3100, Coordinator = 2900, Chameleon = 4300, Long Leaper = 5300, and Immobilizer = 8200. All immobilized pieces are penalized -25% of their value. The Withdrawer gets a small bonus proportional to the value of the most valuable adjacent enemy piece (provided there is at least 1 square in the opposite direction for it to move into, although it need not be vacant presently.) The Coordinator gets a small bonus proportional to the number of enemy pieces on the same rank or file as the friendly King. The Chameleon gets a couple small bonuses: for standing adjacent to an enemy Withdrawer (if there is at least 1 square in the opposite direction to move into), and when the enemy coordinator is on the same rank or file as the friendly King. The Immobilizer gets no bonuses, instead immobilized pieces are penalized. The Long Leaper also has no bonuses, but only because I have no good answers here. Roberto correctly points out that the Long Leaper is more valuable if the enemy pieces are not clustered, and not on the edge, but I cannot think of a way to determine that without spending far too much CPU time. I will continue to think about it. Also, in the opening, pieces are given a bonus for the first move (development), a small penalty for moving twice, and a large penalty for moving the same piece three or more times. These adjustments are slowly scaled down as the game progresses into the middle-game.
The bugs you metion would a dramatic effect on play skill, even with a good evaluation function. I'll have to take care of those, and post an update ... Thanks for the test-report!
25 comments displayed
Permalink to the exact comments currently displayed.