While an ICS does send the opponent's rating, that rating cannot always be trusted. Many times Symbolic has had an opponent with a ~1200 rating magically play 2400 chess, obviously cheating. In those cases, a contempt offset doesn't work so well.

Aside: I've thought about adding a feature to ...

I've done the random game thing myself, long ago. Since all moves for both sides are selected at random with no thinking, thousands of games per second can be played. A hundred million games takes only minutes to generate on a fast machine. What can be simpler or more easy than picking a move at ...

I will not dispute the observation that careful use of an adjustable contempt offset (not "factor") will likely increase a program's score over time.

However, I've never used it in my programs because it violates the spirit of the adage "Play the board, not the man". A contempt offset will taint ...

The answer is to use Monte Carlo testing. Have the program play a hundred million random games (each move picked at random for both sides until a mate or draw). After each move, the retro generator is run and a test is made to verify that the move just made had been retro generated. Simple and easy!

The Half Move Clock (or Half Move Counter) is the number of ply (half moves) since the last pawn move or capture. This is the definition from the FEN specification and it was taken from the FIDE rules in force at the time (ca. 1994). Castling does not affect the Half Move Clock other than to ...

The projected completion date is currently June 20, 2012, some 320 days since the start of the occasionally interrupted run.

The estimated total elapsed time easily beats that of a run I made in 1994 which took 87 days to calculate the KBNKN DTM tablebase on a creaky old 33 MHz 486DX machine ...

Eight of the twenty draft 12 subtotals have been calculated. As with the draft 11 results, at

http://dl.dropbox.com/u/31633927/Perft/Perft13/draft11

these draft 12 records have also been posted on the net:

http://dl.dropbox.com/u/31633927/Perft/Perft13/draft12

The first estimates for total running time were in the 5 to 10 month range; it now looks like 11 to 12 months total.

The program doing the work is my chess program Symbolic modified to use a 120 bit length hash, a 64 bit increase over the standard version. The change was made because trial runs ...

Greetings to all:

Those who frequent TalkChess already know about my little perft(13) project; I'd thought I'd join and post here for the rest of you.

Perft(13) is the number of distinct move paths of length 13 ply from the initial position. Paul Byrne has already posted a number for this; what I ...