Future Work

One of the nicest things about computer chess is the fact that the to-do list of interesting ideas seems to grow steadily despite all the improvements made since the last time you checked. For DARKTHOUGHT, in particular, the whole field of parallelization is still untapped. While parallel or distributed probability-based B* [26] looks very attractive, DARKTHOUGHT already contains all localized data structures as necessary for massive alpha-beta parallelization [47].

DARKTHOUGHT on a single CPU will use fractional ply extensions [135] in the foreseeable future. Singular extensions [8], enhanced transposition-table cutoffs [177], and MTD(f) [171] are all worth a trial, too. Automatic learning and tuning of opening books ranks high on the list as well. More mundane tasks like low-level speed tuning and similar programming ``hacks'' are scheduled to take place just before important tournaments such as world championships. Please visit the WWW page of DARKTHOUGHT in order to follow its progress and download further information (e.g. notations of all tournament games played by the program).

Created by Ernst A. Heinz, Thu Dec 16 23:28:11 EST 1999