One response to “Shipov’s live commentary on the London Classic, Rd 6”

  1. Hi Sergey,

    I have seen a YouTube video on this game: ‘Chessworld.net Instructive Endgame: Bishops not great at monitoring passed pawns! Kramnik – Carlsen’. At move 69 the video suggests White has a win and gives the following line: 69 g5, hxg5; 70 g3, Kd4; 71 Kg4, Ke5; 72 Ba2, Ke4; 73 Kxg5, Kf3; 74 Kh4. At this point Ke3 is suggested for black and the video goes into several lines of analysis using various engines without coming to any conclusion. I myself prefer 74… g5+ and offer the following lines of play.

    (A) 75 Kh3,Kf2; 76 Kg4,Kg2; 77 Bd5+,Kf2; 78 Kh3,Kg1; 79 Kg4,Kf2; 80 Kh3, Kg1. Or here 79 g4, Kf2; 80 Kh2,Ke3; 81 Kg3, Ke2; 83 Kg2, Ke3.
    (B) 75 Kh3,Kf2; 76 g4,Kf3; 77 Bd5+, Kf2; 78 Kh2, Ke3; 78 Kg3,Ke2; 79 Kg2, Ke3.
    (C) 75 Kh3,Kf2; 76 Kh2,g4; 77 Bd5,Ke3; 78 Kg2,Ke2; 79 Bc4+,Ke3; 80 Kg1, Kf3; 81 Kh2,Kf2.
    In each case Black draws by a repetition of moves.

    So it seems your analysis based on general principles and positional understanding is correct and better than the best efforts of computer programming. But please don’t get me wrong: I am no cynic and am very much in favour of technology where it can help to give us a different perspective. After all, I have worked in IT all my life and understand how it can be used to provide tangible benefits. However at the same time I am very aware of the limitations of technology in situations where the normal rules or ‘parameters’ do not apply and an element of judgement must be used based on previous experience. The game Kramnik – Carlsen illustrates just such a situation. I am also reminded of the game between Topalov and Shirov (Linares, 1998) where black although two pawns up in the endgame played an unexpected bishop sacrifice to create a positional advantage and won with just king and pawns. At the moment I cannot imagine a computer coming up with such a move, especially if it has no precedent to draw on. It is far easier for computers to find sacrifices in the middle game where there are many examples in its database. To find such moves in an endgame requires the willingness to abandon material for positional or strategic gain. The sacrifice must be correct, otherwise it will almost certainly be fatal with very few chances to recover. Such sacrifices become more difficult for a computer to consider the longer a game goes on because it has to change its inbuilt ‘rules’, exactly the kind of thing that computers are not designed to do. Unexpected moves require imagination and a level of ‘lateral thinking’ which computers are not programmed for. Computers generally thrive on performing laborious tasks super-efficiently and not on coming up with new and original tasks.

    Will computers ever be able to emulate human beings as far as chess goes? They are certainly getting better all the time, so who knows, perhaps one day they will. To reach that level, though, they will have to be able to ‘reprogram’ themselves, and I don’t know if this is possible. It means that not only will they store rules and factual information, but also assimilate positional situations and ‘learn’ general principles from them so that they can reassess their own rules and identify the situations in a new way. The key to this will be the quality of analysis engines and their ability to identify patterns and create new links within their knowledge store. So quality, not quantity or power, will become the determining factor.

    I hope this has been useful.

    Regards,

    Antony Katcharyan

Leave a Reply