A machine will be the best chess player in the world, sometime before the end of 2000 (even if subsequently eclipsed by human players). Evidence could be: Defeating the Human Champion in a match; Winning a Tournament that included the strongest players; or A Decisive Record against a representative sample of the best hundred players. Games should be played under tournament conditions, with between two and ten minutes allowed per move for each player, on the average. (The Fischer clock, and similar schemes, are ok.) During adjournments, human players may receive advice and consult books and other computers; but particular games must be played by single persons. The machine may use any (or all) of the world's computing resources. After an individual game is commenced, no program changes are allowed, no advice may be given, or program parameters tuned, or chess libraries added, even during an adjournment. Hardware may be added, removed, or repaired, at any time during a game. [Any added storage may only contain chess information that was present at the start of the game. A crashed program or machine may be restarted, and told the moves of the game so far, the time remaining on the clocks, etc.] The machine may be apprised of available hardware resources at any time, and it may query the operator about the time on the chess clock. The machine may know information about its opponents, such as name, ability, weaknesses, and past games. The machine may prepare, or be prepared, for particular opponents in advance of a game. In a tournament, the machine may request to be kept apprised of the status of other games in the tournament, including players, positions, moves, and clock status, but not of estimates of winning chances; the machine may know the expected pairings and playing schedule, and the prize structure. If a game is begun while another is in progress, the games may use different programs, so long as each game uses one consistent program.
I will judge based on the wording of the claim unless it is found to be ambiguous. Such ambiguities will be resolved based on my perception of the author's intent.
Garry Kasparov is (as of April 1997) the current Human Champion. See http://www.chess.ibm.com/ for information on an upcoming competition between Garry Kasparov and Deep Blue that could decide the outcome of this claim; http://www.chesschampions.com/theater.html for online coverage of the May 3-11 1997 match; http://www.sciam.com/explorations/042197chess/ for Scientific American's coverage of the event.