💡📝M Winther wrote on Sun, Feb 25, 2007 07:16 PM UTC:
Of course, time will tell how a variant is evaluated. But you are wrong
about Zillions. It can be tweaked to play openings good, and to evaluate
the pieces correctly. I nearly always tweak the pieces, and give them new
values. That's why I try different alternatives and playtest them, and
also evaluate them intuitively. If the piece is too weak I often give it
new capabilities, like I did with the Ladon, the Castalia, etc. I also
encourage castling, and discourage early queen moves. All this is very
easy to do, and the effect is marvelous.
For instance, I altered the Korean Chess and Chinese Chess code in the
Zillions standard versions. All I did, more or less, was to tweak the
pieces so that their values became more correct. The result was that my
tweaked versions won one match each against the standard Zillions
versions. Both matches ended 6-0. So the effect is an immediate increase
in playing strength. Please have a look in my zrf's and copy the tweaking code.
It's a pity that people don't tweak their programs. Perhaps it's not always
necessary, but in most cases it is, and suddenly the program is
interesting to play against and the playtests are valuable.
Zillions can play chess very well, if pawn moves are encouraged in the
opening, etc. I playtested my Saitek Travel Champion which is evaluated to
Elo 2080 by USCF. This is a very proper evaluation, I think. This computer
plays a very nice game of chess. I ran it against my Blindfold Chess
(which contains some tweaking) at 10s per move in two games, and 30s per
move in one game. The Zillions computer was a 1.6 GHz. The two first games
ended 2-0 to Zillions and the last was drawn, although Zillions had a pawn
up in the endgame. In the first game Zillions had a pawn up in the
endgame, too, but it should have been a draw. The Travel Champion made a
silly move however.
In the second game the Travel Champion was run over in the opening.
Openings were well-known. The first was a Keres defence. The other a
Caro-Kann, Panov variation, and the third a closed Sicilian. Zillions
played better in the opening in all games, so the tweaking is effective.
While the Saitek computer could put up some resistance, I would judge
Zillions to be, perhaps, Elo 2150-2200, on an 1.6 GHz computer, because it
wasn't that superior. Of course, more playtesting is needed. But a good
guess would be at least Elo 2250 on a standard computer of today, that is,
around 3.2 GHz.
People seem to underestimate Zillions's chess playing capacity greatly.
It's a good program that plays an interesting game, at least when
tweaked. Of course, against humans it would fare even better because we
aren't used to playing these strange chess variants, with their strangely
moving bifurcation pieces, etc. However, in standard chess, strong human
chessplayers would know how to overcome it. /Mats
For instance, I altered the Korean Chess and Chinese Chess code in the Zillions standard versions. All I did, more or less, was to tweak the pieces so that their values became more correct. The result was that my tweaked versions won one match each against the standard Zillions versions. Both matches ended 6-0. So the effect is an immediate increase in playing strength. Please have a look in my zrf's and copy the tweaking code. It's a pity that people don't tweak their programs. Perhaps it's not always necessary, but in most cases it is, and suddenly the program is interesting to play against and the playtests are valuable.
Zillions can play chess very well, if pawn moves are encouraged in the opening, etc. I playtested my Saitek Travel Champion which is evaluated to Elo 2080 by USCF. This is a very proper evaluation, I think. This computer plays a very nice game of chess. I ran it against my Blindfold Chess (which contains some tweaking) at 10s per move in two games, and 30s per move in one game. The Zillions computer was a 1.6 GHz. The two first games ended 2-0 to Zillions and the last was drawn, although Zillions had a pawn up in the endgame. In the first game Zillions had a pawn up in the endgame, too, but it should have been a draw. The Travel Champion made a silly move however.
In the second game the Travel Champion was run over in the opening. Openings were well-known. The first was a Keres defence. The other a Caro-Kann, Panov variation, and the third a closed Sicilian. Zillions played better in the opening in all games, so the tweaking is effective. While the Saitek computer could put up some resistance, I would judge Zillions to be, perhaps, Elo 2150-2200, on an 1.6 GHz computer, because it wasn't that superior. Of course, more playtesting is needed. But a good guess would be at least Elo 2250 on a standard computer of today, that is, around 3.2 GHz.
People seem to underestimate Zillions's chess playing capacity greatly. It's a good program that plays an interesting game, at least when tweaked. Of course, against humans it would fare even better because we aren't used to playing these strange chess variants, with their strangely moving bifurcation pieces, etc. However, in standard chess, strong human chessplayers would know how to overcome it. /Mats