Corrected Test Positions

As offered by Hyatt and Newborn in 1997, we obtained their 347 original test positions from <ftp://ftp.cis.uab.edu/pub/hyatt/plytest/positions.gz>. To our utter suprise we discovered that the original test positions still contained 4 doubles and 4 other errors in their FEN specifications although they were already used in Hyatt and Newborn's experiment. We cross-checked our discovery by inspecting Hyatt and Newborn's publicly available result file from <ftp://ftp.cis.uab.edu/pub/hyatt/plytest/results.gz>. There, the suspected errors and doubles were indeed visible in the output of CRAFTY (wrong castling rights in position #34, Pawn on 8th rank in position #90, and two black Kings in position #255). We removed the doubles and corrected the errors as described below. Overall, we were left with 343 corrected test positions to conduct our own experiment with.^1.9

• Doubles occurred in the original positions #1 and #16, #44 and #47, #115 and #279, and #240 and #247 which are pair-wise identical.

• The FEN specification of position #34 mis-spelled the castling rights as Kqkq instead of the correct KQkq.

• The FEN specification of position #90 placed a non-existent Pawn on the 8th rank by reading pr4k1/ instead of the correct 1r4k1/.

• The FEN specification of position #255 placed a second black King on the 1st rank by reading /1r3k2 instead of the correct /1r3K2.

• The move information associated with position #61 suggested to play the impossible Rxf7 instead of the correct Ng5.