Samuel's signature tables
In November 1967 Arthur Samuel published the sequel to his 1959 paper on the checkers program. In place of a linear polynomial of position features he describes a hierarchy of «signature tables» and an improved way of learning from master games. The correlation of the evaluation with the master's move rose to 0.48 after about 175,000 moves for the tables, while for the polynomial it levelled off at 0.26 after about 50,000.
Why it matters
The record gives a numerical answer to what changed in eight years: not the strength of play but the quality of the learning procedure. The author writes that the program still cannot outplay checker masters, though it plays much better. This is an editorial assessment.
What was read: the full text of the paper (17 pages), not only the abstract. The abstract: «a new signature table technique» and «an improved book learning procedure» which the author thinks much superior to the linear polynomial method described earlier. The configuration of fig. 4 and table 2 used 27 parameters (nine groups of three); the polynomial coefficients of table 1, with which the games against Hellman were played, come from an analysis of approximately 150,000 book moves (the caption) or «roughly 150,000 board situations» and 31 functions (the main text); a sample of about 250,000 positions of master play was key punched and transcribed to magnetic tape. For the polynomial the measure of a parameter's quality is C = (L − H)/(L + H), where H and L are the accumulated counts of moves for which the parameter value is larger and smaller than the value at the book move; for the signature tables C = (A − D)/(A + D), where A and D are totals accumulated for the signatures of the book move and of the non-book moves. By the author C has the dimensions of a correlation coefficient. For the polynomial the coefficient «appears to stabilize at a figure of 0.26 after about 50,000 moves», and for the signature tables it «continues to rise and finally after perhaps 175,000 moves reaches a limit of 0.48». The author writes: «While still unable to outplay checker masters, the program's playing ability has been greatly improved». Footnote 2 says that in a 1965 match the world champion Hellman won all four games played by mail and drew one hurried cross-board game, and that the Pacific Coast champion Hanson has twice beaten current versions of the program. What the record does not claim: any figure of playing strength (a rating); that the signature tables learn without human moves (the method rests on book moves). There is no second source: the full text was read in the University of Virginia copy of the same paper.