Quote:
Originally Posted by Biesterfield
I'm glad you linked this. I had kinda written 538 off and haven't been checking it out at all.
I interviewed Murray Campbell (head of programming team for Deep Blue, featuring prominently in the article) not too long after that match for a paper I wrote in HS. He was surprisingly easy to get on the phone for some completely random HS student, lol.
I also talked with Joel Benjamin, Grandmaster who was consulting for Deep Blue's programming team, at a chess tournament a couple years after that.
One thing that struck me the most was how much they geared Deep Blue to specifically play Kasparov. I mean, it makes sense since that was their opponent but there's a very good chance no other top 10 GMs would have had as much trouble. Deep Blue likely had a lot of blind spots but they didn't think Kasparov would realize how to exploit them in a short match. He was prepared for Deep Blue from a year before but unlike a human opponent, Deep Blue's style and ability of play changed drastically. Kasparov's approach fundamentally hadn't plus he had almost nothing to prepare with except his brief match the year before. Deep Blue had Kasparov's entire career of games.
Interestingly to me, Campbell seemed to be of the opinion Kasparov beat himself while Benjamin kind of thought "he" had out-tricked/beaten Kasparov himself.
Neither seemed to want to take much responsibility for Deep Blue not offering a rematch and blamed it on IBM (may be true). That was the most frustrating thing at the time to me. Kasparov never got a shot at revenge and a few years later computers had more clearly surpassed humans at chess. I think psychologically he would have been able to recover and have been much better (but still not all that well) prepared in a 3rd match.
Oh, also it's very common ime for a non-serious chess player to make tons of errors when writing about chess. It always seems weird since many can be so easy to fix if you just have a serious chess player proof the damn thing but, meh, I'm not surprised.