Affiliations: Computer-Science Department, Carnegie-Mellon University, USA
Note: [1] This research was sjXlnsored by the Defense Advanced Research Projects Agency, ARPA Order No. 4976, monitored by the Air Force Avionics Laboratory Under Contract F33615-84-K-1520.
Note: [*] This work is the product of the whole Hitech group which includes Carl Ebeling, Gordon Goetsch, Murray Campbell, Larry Siomer and Andrew Palay.
Abstract: This paper is intended to fill a gap in the documentation of Hitech, the highest-rated chess computer in the world. Thus far, publications have appeared on the move-generator design (Ebeling and Palay, 1984), the design and construction of the Hitech machine and software (Ebeling, 1986), and the artificial-intelligence implications of the design (Berliner and Ebeling, 1986). Here we fill in some of the details of less general interest, but of interest to those that build chess programs and follow their development. Most of the information is chess-specific. What is of general interest is the developmental history of Hitech in which we intend to show how a complex system evolves with unexpected turns and twists in the road. In projects such as this, it is not possible to determine the full effect of any change, nor can one anticipate the vehicle’s needs over more than one change at a time. At this writing, we feel that the pattern-recognizer structure described herein is principally responsible for Hitech’s success. We have not begun fully to exploit this structure, so the potential for large further improvement remains.
DOI: 10.3233/ICG-1987-10302
Journal: ICGA Journal, vol. 10, no. 3, pp. 111-117, 1987
