USC
About this Article
Written by: Lawrence Aung
Written on: May 4th, 2010
Tags: computer science, entertainment
Thumbnail by: ThinkQuest
About the Author
Lawrence was a junior at USC studying to be an Electrical Engineer. He is also a former chess enthusiast.
Stay Connected

Volume XI Issue III > Deep Blue: The History and Engineering behind Computer Chess
Computer chess software available today is a staple of modern computing distractions. Few may recognize the rich history behind the development of that technology. In a revolutionary chess tournament in 1997, the chess world champion was defeated by an IBM supercomputer called Deep Blue, shocking the media and the general public. To the artificial intelligence community, however, this was a long time coming. After almost fifty years of developing adequate computing technology and formulating sufficient chess playing strategies, computer scientists were finally able to solve the “chess problem” and defeat the best human player. The specific search algorithm used was revolutionary for its time, and Deep Blue’s victory served as a defining moment in the history of the field of artificial intelligence. However, some doubts have been raised regarding computer chess’s relevance to the field of artificial intelligence. Strong arguments have been made for both sides by prominent experts in the fields, and the fate of future computer chess studies hangs in the balance.

Introduction

After 50 years of research and development in conjunction with artificial intelligence, the study of computer chess culminated during two matches between Deep Blue, a chess supercomputer funded by IBM, and the Chess World Champion Garry Kasparov. The 1996 and 1997 matches were media sensations, heavily promoted as a battle of wits between man and machine. When the machine finally prevailed in 1997, the artificial intelligence field and scientific community marveled at their own triumph. However, recent experts have disagreed with the predictions of their predecessors, which suggested that such an event would herald in an era of intelligent machines, pointing to several issues of comparing chess to intellect. While continued research on artificial intelligence is necessary before it becomes an everyday reality, Deep Blue’s victory over Garry Kasparov in 1997 still remains an impressive engineering feat and a foundation for current achievements in the field.

History: From Turing’s Paper Machine to Deep Blue

Computer chess has been associated with the field of artificial intelligence ever since the emergence of computer science in the mid-20th century. Many of the biggest names in both fields, such as Alan Turing, a mathematician whom many credit with founding modern computer science; John McCarthy, who helped coin the term “artificial intelligence”; and Claude Shannon, a pioneer in Boolean logic, supported the use of chess as proper starting point for the development of intellect [1]. Undeterred by the lack of technology and establishment of a precedence at the time, Turing sought to support his theory and effectively wrote an algorithm for the first chess program. Unfortunately, the program, which was put on paper and implemented by Turing himself, was perhaps better visualized than implemented, as a colleague soon easily defeated the program. Despite this unceremonious introduction, chess programs would soon flourish, gaining the recognition as a standard for research for many computer scientists [2].
James/Flickr
Figure 1: A computer from the Computer History Museum that is similar to Deep Blue.
Many early programmers devoted a particularly large amount of time and effort to chess because they perceived the game as a unique challenge that would provide a breakthrough for artificial intelligence. These researchers saw in the structure of chess a simplified model for large-scale problem solving. Players in chess only have the primary task of checkmating the opponent’s king. In addition, several other constraints inherent to the game, such as the relatively small board and limited amount of pieces with distinct movement patterns, helped computer scientists define specific parameters within their programs. In terms of human comparison, the already existing hierarchy of rated chess players throughout the world gave engineers a scale with which to easily and accurately measure the success of their machine.
After several decades of research yielded little progress, it is surprising that Deep Blue, which would arguably become the ultimate champion of computer chess, began with a chip that was built in six months by a Carnegie Mellon graduate student and an inexperienced chess player. In 1985, while doing secondary research for another professor, Feng-Hsiung Hsu discovered a method that could reduce the work done by the chess machine’s move generator by simplifying the program’s silicon chip design from a maximum of sixty-four distinct chips used by contemporary machines down to just one. While looking over the design of sixty-four chips, he discovered that the extensive amount of wiring could easily be reduced, decreasing the delay of the design and saving time and precious space on the chip [3]. After further studying the design sketches, Hsu discovered that another large portion of the chip was redundant and could therefore be excised completely. Because he was only removing useless components, Hsu’s new, reduced design did not lose the speed expected with his reduction in size. In fact, Hsu’s proposal actually suggested a theoretical gain of speed by a factor of twenty due to the reduced complexity of his new chip. The single chip design that Hsu developed would evolve over time to form the backbone of Deep Blue’s hardware (see Fig. 1).