Chess Forum / General Topics / August 2006

 Unfortunately, programming chess is not my specialty.
What I do know is that the first step in a brute-force type
program is to increase the speed and depth of search.

 A human player can understand concepts which exceed
the limits of any ordinary chess program, and so I would
advise having your team of programmers focus on the one
great strength of computers: tactics.  If you can improve
the speed and depth of search and see *all* short-range
tactics, you will overtake most players on this alone.  As
one famous Grandmaster put it: chess is 99% tactics!

 If you can reach the point where even Taylor Kingston,
for example, can no longer easily outplay your program
tactically, the games will lengthen and humans will have a
greater opportunity to blunder, as we are prone to do.
Exploiting these blunders normally requires no great
technique, just tactical accuracy.  In fact, most humans
will resign after hanging a piece or two.

 In order to improve tactics, the most effective thing
seems to be check-and-capture extensions, where each
time there is a capture or a check, the program steps
out of its normal depth, and looks further, to see what
happens as a result.  As I said before, there are special
techniques used by expert chess programmers to do
this sort of thing in an efficient manner, but I don't know
enough to tell you exactly how it is done.

---------

 Having replayed many games, IMO Taylor Kingston is not
using a computer to decide his moves, so when you write
that somebody like "Bob" or "Pfuscher" is a better player,
you are most likely comparing apples to oranges.  Except
for the one game which Bob lost, I have yet to see a single
move from either of these two players which could not have
been computer generated.  With Pfuscher this could be a
fluke, but with Bob it means he is Bob-by Fischer, with
nothing better to do than play at GetClub!   :)

 -- help bot