The Institute - September 2021 - 64

TECH HISTORY
How IBM's
Deep Blue Beat
Chess Champion
Kasparov
200 million possible chess positions per second
was hot stuff in 1997
BY JOANNA GOODRICH
CHESS IS MAKING a comeback thanks
to The Queen's Gambit, a popular
Netflix miniseries about a prodigy's
journey to becoming the world's
greatest player. But Beth Harmon-
the fictional prodigy portrayed by
Anya Taylor-Joy-never faces a
supercomputer the way real-life world
champion Garry Kasparov did.
IBM's Deep Blue made history
in 1997 when it became the first
machine to beat a reigning world
chess champion. A research team
led by IEEE Senior Member Murray
Campbell and Feng-hsiung Hsu developed
the supercomputer.
Kasparov accused the IBM team of
cheating its way to victory. In reality,
though, scientists had been interested
in programming a computer to play
chess since the late 1940s, according
to an article on IBM's blog about Deep
Blue. It took years for engineers and
computer scientists to perfect the artificial
intelligence program that would
one day beat a world champion.
Five decades in the making
Deep Blue's story began in 1985,
when Hsu, then a Carnegie Mellon
graduate student, started working on
his dissertation project: ChipTest, a
chess-playing machine. Hsu worked
with Campbell, who was a research
associate at the university, and graduate
student Thomas Anatharaman, an
IEEE member, to develop ChipTest.
Hsu and Campbell later joined IBM
Research in Yorktown Heights, N.Y., in
1989. The duo continued developing
a chess-playing machine but this time
with other computer scientists working
on the Deep Blue project.
The final version of the machine
consisted of two 2-meter-tall towers,
more than 500 processors, and 216
accelerator chips designed for computer
chess, according to a paper Campbell
and Hsu wrote about Deep Blue for the
Artificial Intelligence journal.
The machine's software would
Deep Blue, which consisted
of two 2-meter-tall
towers and more than
500 processors, in IBM's
headquarters in Armonk, N.Y.
calculate the basic moves it could
make in response to its opponent
before the accelerator chips carried
out more complex calculations such as
assessing possible outcomes of various
moves and determining the best one.
The computer would decide which
YVONNE HEMSEY/GETTY IMAGES

https://www.netflix.com/title/80234304 https://www.kasparov.com/ https://www.ibm.com/us-en/ https://www.ibm.com/ibm/history/exhibits/vintage/vintage_4506VV1001.html https://researcher.watson.ibm.com/researcher/view.php?person=us-mcam https://researcher.watson.ibm.com/researcher/view.php?person=us-mcam https://www.chessprogramming.org/Feng-hsiung_Hsu https://www.ibm.com/ibm/history/ibm100/us/en/icons/deepblue/ https://www.cmu.edu/ https://www.linkedin.com/in/thomas-anantharaman-15a1997 https://www.research.ibm.com/ https://www.research.ibm.com/ https://www.sciencedirect.com/science/article/pii/S0004370201001291 https://www.sciencedirect.com/journal/artificial-intelligence/vol/134/issue/1

The Institute - September 2021

Table of Contents for the Digital Edition of The Institute - September 2021