Early computer chess in Pittsburgh

Computer chess engines are routinely used nowadays for work that even the strongest human players wouldn't be trusted with.  You can go to chess news websites like chess24 and get analysis of ongoing GM tournament games by engines.  People use engines to try to “cheat” in online chess, and other people use engines to try to catch the cheaters.  But 50 years ago, strong computer chess players were just a dream.  I was reminded of this by Tim Just's article in Chess Life Online about the rules regarding computer participation in USCF tournaments.  As computer chess programs progressed from primitive academic experiments to GM-strength commercial software, some of the most interesting research and development was in Pittsburgh.

The first computer strong enough to get a rating was MacHack, developed at MIT in the late 1960's.  It reached about Class C.  Not long after that, Hans Berliner, who had won the World Correspondence Chess Championship in 1968, got interested in computer chess, and left a job at IBM to come to Carnegie-Mellon and do computer chess research.  When I started my graduate studies at CMU in 1973, Hans was still a student, but he soon completed his Ph.D. and was hired as a professor.

Hans at the time was still occasionally active in over-the-board chess, playing in at least one tournament and in the Pittsburgh Chess League.  His presence galvanized the local players:  I remember getting a copy of an article he had written about his opening repertoire, from my friends, who were eager to study it.  (The article was the forerunner of a book he published in 1999, called The System.)  However, he retired from over-the-board play in 1975.

For several years Dr. Berliner conducted various small research projects, including a strong backgammon player called BKG.  BKG famously won a match from the world champion in 1979, although Berliner gave some credit to lucky die rolls for that victory.

The culmination of his computer chess research was the Hitech project.  Several students worked on this with Hans, including Carl Ebeling, who designed some hardware.  Hitech was the first computer chess player to achieve a USCF Senior Master rating, a level that it achieved when it won the 1988 Pennsylvania state championship.  Here is the score of Hitech's dramatic last-round game against IM Ed Formanek, which Hitech won with the Black pieces.  I've taken the notes from this article in AI Magazine, by Berliner himself.

In the course of the Hitech project, Dr. Berliner asked a graduate student, Feng-Hsiung Hsu, for some advice about aspects of the hardware design.  Hsu was inspired to start his own computer chess project, which soon came to rival Hitech in playing strength and impact.  Hsu focused on VLSI design, trying to get the move generator onto one chip (instead of 64 chips, as in Hitech).  This enabled the move generator to be a thousand times faster, and speed translated into strength.

The Fredkin Foundation, of Cambridge, Mass., had established a substantial prize for the first computer chess program that could beat the world champion, and in 1988, with the foundation's backing, Dr. Berliner organized a tournament in Pittsburgh to enable leading chess programs to get experience against strong human players.  Participating computer systems included Berliner's Hitech, Hsu's Chiptest and Deep Thought, and BP, by Robert Cullum.  Humans playing included IM Alex Ivanov, SM Ron Burnett, Pittsburgh's young star Vivek Rao whose USCF rating was near 2500, six players rated above 2300, and 9 players rated above 2200.  I still have the tournament book, with all the game scores.  Ivanov won the event, winning his game with Deep Thought, which finished tied for second with Mark Eidemiller and Klaus Pohl.  Chiptest finished a half point behind those three, in a tie with myself, Jerry Wheeler, and David Kolarik.  Hitech, another half point behind, had a mediocre score.  The tournament was directed by Tom Magar.

Here is Ivanov's game with Deep Thought, without notes. 

After Hsu finished his Ph.D. in 1989, he and two other members of the Deep Thought team, Murray Campbell and Thomas Anantharaman, were hired by IBM Research to work on a successor.  This successor, named Deep Blue, ultimately won a match with Garry Kasparov in 1997.  You can read more about Hsu's computer chess work in his entertaining book, Behind Deep Blue (Princeton University Press, 2002).

Scroll to Top