RandomVisitor: A talented American "Postal" chess player gives his opinion on computers in correspondence chess, circa 1986-1988
From the NM Alex Dunn correspondence chess column 'The Check is in the Mail' in the <April 1986 Chess Life and Review>
Dr. Robert Reynolds, a master in both postal and over the board play, is a psychologist at Fordham University. In the following letter, he offers a personal view and a challenge to computer programmers: "The question of how strong chess computers might be in correspondence chess may be important to both the development of computers and to the understanding of human chess skill. Thus far, the emphasis has been upon developing main-frame computers which calculate as quickly as possible. This strategy has lifted a few computers as high as 2200 in over the board play. In the case of postal chess, one might suppose that a match between Belle, HI TECH or Cray Blitz and a human would favor the analytical power of the computer. However, we need not worry much about it. Adriaan de Groot showed some 20 years ago that the higher levels of human chess are distinguished by quick sight of the board and analysis of fewer total moves. Search heuristics and general positional evaluation become more and more important for humans at just that level of playing strength where computers emphasize long-winded algorithmic calculation. Until there is a new generation of computers capable of parallel-processing, current computers will rarely be able to calculate far enough ahead (even if given three days) to reach a decisive outcome. I agree with one of my postal opponents, U.S. co-champion Ken Plesset, that 'Chess is 90 percent tactics, but the [other] 10 percent is more important.'
"My prediction is that present-day computers would be no better in a relative sense at correspondence chess. In postal terms, this would be a rating between 1650 and 1750 [JLJ - the postal chess rating system was different at this time]. But if anyone still believes that computers are a threat - or even a possible aid at the higher levels - <I will wager $500 on a match of four games, played under the same rules for human postal correspondence chess>. My opponent may consult books, but he cannot have access to or seek adjustment from auxiliary programs of either the mechanical or human variety. Whatever the result we would certainly learn from such a match." - <Sincerely, Robert I. Reynolds.>
From the <September 1988>, NM Alex Dunne postal chess column 'The Check is in the Mail: The Creative Edge' in <Chess Life and Review> magazine:
"Robert Reynolds is a name the postal world should keep in mind. We will probably be hearing much more about this gentleman from New York City. Robert recently won the 6USCCC with a powerful 13 1/2 - 1/2 score...
"Reynolds further states, 'Over two years ago I made a challenge in this column that I'd play any of the top ranked computers in a correspondence match under standard rules. I have not received an acceptance and probably never will. The challenge is a way of highlighting the relative weakness of computers. One might think that an average of three days per move would favor the computer with its ability to rapidly process large numbers of moves. Not so. Even computers dedicated to chess do not gain much in depth with an extra three days of analysis. The tree of possibilities grows so fast that the gain would amount to only five or six half moves. At the end of its analysis, the computer must stop and evaluate. Unless a decisive position has been reached, nothing much will have been gained.'