Sumerian: Is your Gothic Vortex even still subject for improvements (beside of providing more looking-up information)?.

The evaluation function for Gothic Vortex is very primitive as well. It has a few innovations. Some of these have been around for a while, but here is what I think is unique:

1. Sliding scale weight for the "Supermajors" (Queen, Chancellor, Archbishop).

The values for these pieces are not "constant", they change during the phase of the game. The Archbishop gets weaker as the game progresses, the Chancellor gets stronger, then weaker, and the Queen slowly gets stronger.

2. Pawn Races - The program can play ANY king and pawn ending properly, even with just a 3-ply search. This makes it strong in the endgame.

3. Castling awareness - Castling is not just a "straight bonus", it evaluates whether or not the opponent is castling into danger.

I have changed the evaluation function since the 2004 Computer World Championship, but this was mostly bug fixes.

For exmaple, in some of the wins that Chessmaster emailed to me, there were obvious bugs regarding king safety that had the program think it was ahead as much as +400 when really it was -200.

I have long since fixed this bug (4 months ago) and Chessmaster1000 has not sent me any wins since that time.

I have introduced something else that might have disasterous effects, but it only an experimental version for now. I am trying to teach Vortex how to play like me. It is not easy to do! I am using some of my games on BK as an example, but no matter how long I let Vortex think, it just does not want to make sacrifices like I do.

When I figure out how to code this, I will release the new version. For now, the latest and greatest version is at:

http://www.GothicChess.org/vortex.zip