[ExI] sports blammisphy
spike66 at att.net
Sat Feb 5 21:13:26 UTC 2011
> On Feb 5, 2011 1:31 PM, "spike" <spike66 at att.net> wrote:
> Recently the French chess federation has accused three of its own
> players of cheating in last September's Chess Olympiad:
> In any case, I thought of a way to look at the games after the fact,
> using just the game scores, and figuring out a way to determine if the
> players had somehow consulted a computer with some tricky I/O device...
So here's the idea. We have now about 50 or so commercially available chess
engines capable of playing with the big boys. If we had a big enough pool
of volunteers, we could distribute one or more of these engines to each
volunteer. The volunteer enters the game scores for any number of players.
The computer plays each position and derives its own list of choices for
moves, along with it's own estimated evaluation of each move. We get the
software running on a plethora of different computer hardware. If any
player matches exactly and consistently with any of the software's first
choices, well then it is simple, ya got him. No player will match exactly
the way a computer would play. Computers will not match exactly each other.
There have been entire games where the human player chose one of the top
five choices. At grandmaster level, you might well see a human legitimately
choosing the computer's top choice eight or ten times in a row. But fifteen
in a row would make me highly suspicious, and twenty would be a slam dunk.
So I claim there would be a statistical signature of a player using a chess
engine with a tricky hidden I/O device of some sort.
That being said, I have thought of an even trickier trick which would allow
a human to use chess software and sneaky I/O devices, which I will post next
More information about the extropy-chat