At several points in my life I became enamored with the game of chess. I loved the game. Finding challenging opponents was sometimes difficult but always rewarding. I went so far as to study books on chess. I followed the Fischer-Spassky world championship match as if I were observing a great historical event. A good friend took me to see the Second Piatigorsky Cup, a tournament including all of the best players at that time. I realize that for many of you this sounds about as exciting as watching grass grow, but I assure you that being there, in the midst of those games, was as thrilling for me as watching my beloved Green Bay Packers. I joined several chess clubs and even played in a few amateur tournaments.
In Junior High, I joined our school chess club. We challenged another school to a match. We were supposed to rank our players and then match our best against their best, our second best against their second best and so on until one of us ran out of players. Neither of our two best players, however, wanted to play the opposing school’s best. He was well known and intimidating. I was new to the club and so asked if I would play as our best player. I naively agreed and proceeded to lose. I lost in a most humiliating way, making an embarrassing blunder. I didn’t see a mate in one. The next time our club met, I was ridiculed by one of our best players. “How could you miss a mate in one?”, he repeatedly asked. I found this more than a little irritating since by all rights he should have played in the top slot not me. He then challenged me to a game. As we played he continued to remind me of my stupidity in my loss. As the game progressed I saw a simple way to set up a mate in one. I was sure he would see it. I tried to camouflage it a bit before I set the trap. It was not a good move on my part, from a purely tactical point of view, but I couldn’t resist. To my amazement he didn’t see the trap. I won. That was the most enjoyable game I’ve ever played. It wasn’t a very good game, but I relished being able to ask my opponent at the end how he could have possibly missed a mate in one.
I imagine learning chess is quite different today because of the growing power of computers. I played against and easily beat the first popular computer chess toys. This wasn’t because I was so good but because these programs were so poor. That’s all changed. Today several inexpensive programs exist that would easily win against even my best efforts. More than this, a specially designed computer has been able to defeat a world champion chess player. That special design is relevant to this post. Experts in artificial intelligence in that day had thought the best way to program a computer to play chess was to make it play like a human being. They looked for algorithms to express things like chess theory and strategy. Their computers didn’t perform very well against good human players. The switch came when some IBM engineers designed a program to play like a computer and not like a human. Deep Blue relied on blazing quick processing speed and a huge database of openings. It favored brute force which means that rather than analyzing a position’s strengths and weaknesses to find the best move it looked at as many moves as possible. In theory, a computer could look at all possible chess games and find the best moves by avoiding any path that leads to a loss. What makes this difficult in practice is the vast number of possible games, a number in the trillions. As computers get faster and faster, however, this difficulty becomes less of an obstacle. Deep Blue was the first computer to beat a world champion.
All of this suggests a thought experiment. Let’s imagine someone learning to play chess, developing strategic and tactical skills, learning to avoid blunders and find the best move, without ever playing against a human opponent. The tutor would be the latest and greatest computer. What would result? Given a modicum of natural talent, our imaginary chess player could become incredibly skilled and proficient. Given brute force processing, faster processors and a huge database of games the computer could, in theory, always find the best move. The computer wouldn’t miss anything. It would be totally objective. It would focus on outcomes. Its results would be deterministic and, therefore, repeatable. The computer wouldn’t need theory, beauty or elegance. Trained by such a tutor, our imaginary chess player would be only a perfect memory away from being able to play perfect chess. Perhaps, but I’m not so sure.
In fact this scenario haunts me. It seems to threaten to destroy everything human about chess. Is our imaginary chess player really learning to play chess? The best play would involve following a list of computer generated moves. These moves would need to be committed to memory. Can following a list of moves committed to memory pass for playing the game? Why would our hero play? What would draw him or her to the game? What if we eliminate the human element altogether? Can a computer play chess? What would be its purpose in doing so?
This scenario haunts me not just because of my enjoyment of chess. It haunts me because it seems like a picture of our modern world. On the one hand we see and experience the tremendous, almost miraculous results of science. On the other hand we find our humanity to be more and more alien to the very world we’ve created. I don’t think it’s an accident that two of the most totalitarian regimes in recent history, those of Hitler and Stalin, lauded science. What scares me isn’t science, it’s forgetfulness. It’s forgetting that doing science, really doing science, is a human activity. It requires people. It’s social. Its purpose is to serve not obliterate human beings.