It is the reverse of what rarley happens on here.
A human helped a computer. (so it is claimed).
In a Nutshell. A Big Nutshell.
In Game 2 of the 1997 re-match Deep Blue came up with a 'human type move'
DB took a while to find this move rejecting a more obvious computer type move.
(this is the famous 'deep think' - the grassy knoll.)
Gary was convinced the computer would chose a pawn stealing move
subjecting itself to a vicious attack and a human stepped in to thwart the
attack first and then go pawn stealing. A Karpov type move.
Analysis does give a vicious counter attack for Black if White steals the pawn
but no forced win. The type of position a GM would reject as White but
a computer...which knows not the meaning of the word fear...would accept.
After this 'human move' appears a dejected Gary does
not play well and....neither does Deep Blue!!
Gary resigns only to discover the next day DB had slipped up and
there was a draw to be had. (it's a human type error from DB).
The IBM teams explanation as to why it allowed a draw in a won position.
"The King went to the wrong square."
It would appear that DB locked up itself deciding on the pawn steal and facing
the attack so the IBM team allowed 'someone else to take over.'
Gary asks for the print out of the DB analysis IBM refuse.
Why would IBM cheat?
IBM were desperate for some good publicity. The previous year their
computer sytem turn the 1996 Olympics into a complete shambles.
They were the laughing stock of the planet as IBM jokes circled the globe.
IBM had recruited quite a few strong players for the re-match. Some of
whom were kept a secret from Gary.
King reports in his book on the match meeting Nick de Firmian and John
Fedorpwicz in the Hotel lobby just before the match.
"You have not seen us! We are only supposed to appear at the end of the match."
End of conspiracy theory.
IBM did employ and hold pre-match training sessions with various GM's
before the match. Nothing wrong in that.
And it was kept secret in as much as Gary's prep would have been kept secret.
The big think in Game 2 was most likely the computer hitting it's horizon
effect. It obviously was playing for the pawn steal when suddenly the
real dangers of the counter attack kicked in.
It eventually played the 'human move' though it saw and had to analyse
the possible drawing line which Gary missed altogether.
The forced draw, last time I looked was still under debate.
The 'wrong square' may well indeed be the best square.
There were too many people in the back room for a stunt to be pulled.
Someone would have by now have let it slip.
Also (a personal reflection) what GM would do such a thing to GK?
Can you imagine the lawsuit that would follow not to mention the complete
and total embarressment to IBM if it ever got out.
IBM did release DB's full log of game two's analysis but not during the match.