Kasparov - Deep Thought. One gate play

The story of Kasparov’s sensational confrontation with IBM’s brainchild has already been touched on GeekTimes Habré. My comment, though scored a decent amount of advantages, contains several significant inaccuracies, which I have already corrected, alas, not the strength. In order to clarify and dot the hell , it was decided to write a more detailed article about this significant match. However, in the creative process it turned out that the topic touched was much broader, more multifaceted and more interesting, and covered a much longer period of time. Therefore, the article about the match in 1997 was organically transformed into a series of publications about the extraordinary victories and defeats of the 13th world champion in the fight against artificial intelligence for 15 years.

Personal meetings between Kasparov and IBM computer monsters were opened 25 years ago, on October 22, 1989. On that day, a two-game micromatch was organized with a “chess prodigy”, the forerunner of the legendary Deep Blue - Deep Thought (“Deep Thinking”, in chess literature most often simply called Deep Cell).



Deep Sot at the time of the match was the strongest electronic chess player in the world - the reigning world champion among computer programs, as well as a participant in several "biological" tournaments in which he beat some of the famous grandmasters and even won prizes. So the meeting of the strongest "protein" and "silicon" chess players is on the agenda.

Both games clearly showed the main drawbacks of chess computers of those years: greed for the material (if they take it without fail and give and in no case give up, if they take it) and inability to play positionally. Although the computer went through millions of options as early as 1989, this did not help much in the fight against people. She thought the car was good, but she still didn’t feel the position.
')

In the first game, the man played black. In the agile struggle, Deep Sot did not know what to do. Kasparov calmly seized the initiative and methodically strengthened the position of his pieces, consistently increasing pressure on the position of the enemy king, threatening to go on to a decisive attack at any time. The car went into a deaf defense and made waiting moves. On the queenside, White marked the passage of his pawn, but there was nothing to support the demarche - all forces were tightened to defend their monarch.

When the pressure of the black army reached a maximum, Kasparov quickly and decisively transformed the positional advantage into a material one. At first, he pretended that he was going to win the passed b pawn on the queenside and thereby diverted the white queen. While the car was clinging to its “rogue”, on the kingside, the black infantryman went to the ram - 43. ... f5-f4 (see diagram). White's position immediately crumbled and had to give up the bishop. White just in case with a dozen moves played without a figure and surrendered.

Building a positional advantage without providing active tactical counterplay. Over time, the number of positional buns turns into an irresistible attack, leading to an immediate mate or achieving a decisive material advantage. That is how grandmasters won the most from computers. And now it remains the best practice.

In the second installment, Kasparov outplayed his electronic counterpart with white in the opening. While the “calculator” was wondering about the dividends from winning the rook, the white light pieces (two bishops and the knight) took up shock positions and, under the leadership of the queen, rushed into a dull attack. The black queen bravely rushed to the defense of his emperor, but he could not resist the four enemy saboteurs alone. Having given up the knight, Harry on the 17th move gives a deadly check by an elephant (see diagram). This attack can be responded in two ways. Or leave the king in the neighboring field and immediately get a mate: 17. ... 8-d8 18. 3: f8X . Or lose the queen 17. ... Fs7-c6 18. Cb5: c6 + , which happened in the game. Here you can safely give up, but the IBM employees decided that even if the car still opposes 20 moves for propriety.

This game is the second illustrative example of how masters have won over computers of those years. Computers are very eager for accepting sacrificial pieces and pawns, a material advantage in an honorable priority. If, of course, the victim is followed by a checkmate in two moves, then the machine will easily detect the danger and reject the gift of Danai. But if we are talking about a multi-path attack, especially on an undeveloped position, then the resulting full option tree contains many hundreds of millions of branches, which the processors of those years could not calculate until the mat or the obvious loss. An experienced human chess player, at a glance at the board and without calculations, everything is clear: the position cannot be held, the material should not be pursued. The computer honestly tries to find a mate (or at least material losses), and since it is beyond the horizon of counting capabilities, it comes to the conclusion that the victim can be accepted in order to repel the attack, to realize an extra piece. And this decision turns out to be fatal.

So, at the end of the 80s, the computer was still not equal to man and the match Kasparov - Deep Thought gave the current answer to "the main question about Life, the Universe and in general" for chess players - the little people are playing stronger. But in just a couple of years, robots will force arrogant humanity to reckon with itself.

(To be continued…)