One interesting, at least to me, bit of data is how Deep Blue beat Kasparov.
There have been a number of different chess programs out there, with various
degrees of complexity.  Deep Blue, in many ways, had among the simplest
algorithms.

The first part was rather simple; it incorporated virtually all of the
published opening knowledge in the field of chess. As a "B" chess player, I
have a 500 page books giving openings to, say, the 14th or 15th move.
Monthly magazines cover the latest innovations.  It was straightforward to
convert them into a data base.

Virtually all good chess programs had such a library.  Some of the programs
then developed a complex algorithm for weighing the value of every position.
(Often times, the calculated value can be given as a number for the player
who is playing against a computer.)  Deep Blue did not have a complex
algorithm.  Rather, it relied on speed: its ability to calculate up to
200,000,000 positions every second.  (Since that was close to 5 years ago,
if it were to operate today it should be able to calculate about 2 billion
positions per second.

With such horsepower, it was able to examine more plys ahead than any human.
Even though it's analysis was based on a simple algorithm: the ability to
pick the move that gave the best postion after, say, 10 moves on either
side, was usually enough.

The point is that, the best way for a computer to compete against a human is
brute force power.  Even in a situation where it seems straightforward to
develop algorithms to define the human thought process: brute force beats
the best algorithms.

Dan M.

Reply via email to