Tembusu is a college of many talents, including that of a ridiculously talented chess player who happens to be my suitemate, Terry Chua. Sharing the same love for competitive gaming, we often discuss related topics, searching for the concept of genius and understanding meta-cognitive frameworks in gaming as well as famous chess players.
If you haven’t heard of Gary Kasperov and the famous man-versus-machine battle with Deep Blue, check out this video. If you have, just watch it anyway.
Now as with the nature of most things on the internet, I shall attempt to present my opinion as fact.
I do concur that Kasparov lost rather than Deep Blue winning in that situation.
Do correct me if I’m wrong but I do believe that there is a misconception about the idea that a chess player’s ability is tied to his ability to think X number of moves ahead of his opponent.
Top tier players in any strategy game have a very strong study and understanding of the game that gives them various frameworks of goals, positions and objectives to work around. Through those frameworks we then see the supposed 15,16+ steps ahead that people so often project. It’s actually scarier than that: through identifying your goals by considering optimal plays given your positioning, they probably have the concept of the entire game figured out and will be able to tell you your course of play and their counter to either trap or aggressively take a victory.
There is a very meta-cognitive aspect in competitive PvP strategy games, more commonly known as the concept of playing the player. People have preferences and tendencies to follow certain courses of action in which they believe are more optimal, this assists greatly in the formation of the frameworks mentioned earlier, especially during the midgame of chess when there are various difficult decisions that are equivocal in nature.
To call Kasparov v.s. Deep Blue a Man versus Machine would not be a 100% accurate statement. While Deep Blue is technically a machine perhaps with Machine Learning features, it is to a certain degree an optimization process of its creator’s ideas on how chess should be played as they have their frameworks of optimization. This could be finding moves and positions that give the most number of options, or perhaps a model of piece value calculation and assessment that the machine uses to make its moves, maybe even a historical collection of plays that have made value against Kasparov and attempt to recreate them in amalgamation with any of the above concepts. Coding is science after all, not magic.
Now, my conjecture is that Kasparov understands that Deep Blue and other ‘super chess computers’ have such a framework in place and his real goal is to try and figure out what is this framework and model that Deep Blue has been based off on. This is where the glitch in the system comes into place in throwing him off, leading to his defeat.
The mode of optimization that he set Deep Blue on probably was challenged when the unusual move took place, giving him a piece of data and evidence that did not fit in with the read that he had placed on Deep Blue. I would believe that this sowed some doubt in his mind and as the video puts it, ultimately led to his loss to Deep Blue.
This made him believe Deep Blue was cheating after he lost the second game.
Kasparov had been put on tilt by the glitch in the system from game 1, and not the emotional temper-throwing kind we see the media characterize tilt as. He had been thrown off his game and his meta-cognitive view on the situation had been affected. In the second game Deep Blue perhaps could have shifted its framework and approach towards the game, a task easily done by a super computer, but this shift away from Kasparov’s initial projected read on Deep Blue, combined with the doubt from the glitch earlier, put Kasparov in a position where he was unable to identify what Deep Blue’s framework was. This resulted in both sides playing a slow game and, I believe, Kasparov trying to figure Deep Blue out within the constraints of the round.
The eventual resignation was a sign of the compounded tilt that had led to a lapse in decision-making on Kasparov’s part and the dip in performance which he faced throughout the series. Going back to the meta-cognitive concept, playing as the favorite, he believed that he had Deep Blue figured out, citing that no one else understands chess better than him. However, faced with information showing that he may not have understood Deep Blue’s framework and a loss in hand, this mentality switches to a learning approach where instead of pro-actively tackling the perceived weaknesses of Deep Blue’s framework; a more reactionary defensive mode of play is developed in trying to buy more time to reconcile his understanding of the situation.
The next three games are draws as a result of this, as Kasparov was no longer playing to win but rather, to not lose.
Machine does triumph over man in the 6th game of the series. Kasparov shows that he is human after all, makes a mistake and Deep Blue capitalizes and wins the series.
The video does bring up an interesting point that machine has triumphed over man in today’s day and age, taking close to clean sweeps against grandmasters all around the world while failing to drop games at all. What is interesting however is that the processing power of these super chess computers have actually gone down.
This makes me believe that the development of chess has plateaued out for now and the player with the cleaner and more optimized decision-making will triumph, in which case the computers are favored. For man to triumph over machine again, he will have to find a new framework by which the computer will not have an effective response to.
To view the actual video of Man vs Machine, click here.
Image is a screenshot from the shared video.
Senior Editor: Shubhendra Agarwal
Editor: Alison Chew