Man versus Machine: When a Computer will become World Chess Champion

Filed under General Chess, Offtopic

Not next year, when it will be either Anand or Topalov! But how far later?

In one of the 4 myths (depending on what you believe) on the game of chess, I held the view that we have not yet reached the stage where we expect a machine (read computer) to become World Chess Champion, notwithstanding that one win of DeepBlue against Garry Kasparov in 1997. But I left a question mark against that conclusion as I was not sure how long this state of affairs would hold, seeing the speed of progress in computer technology. More powerful processors, larger memory chips, and sophisticated software to utilize the hardware advances are hitting the road every year.

This prompts me to take an inventory of relative strengths and weaknesses between man and his robotic creation and try to understand how they are placed against each other in respect of our area of interest – the game of chess.

To make such a comparison, we must select the parameters on which to base our study. I thought of the following – it will be great if you can suggest some more, with your analysis on those aspects of chess in lines similar to what I tried. Another point – men can make ten identical “Deep Blue” computers but no two human players are alike. So when comparing the relative strengths and weaknesses, I am considering only the best human chess players in the following comparisons.

Chess Strategy:
Simply put, chess strategy is a plan of action. As discussed in What is chess strategy? Isn’t a chess game all chess tactics?, a strategy may be formulated even before the start of a game and may continue well into the end game. Strategy encompasses not just the game but the players as well – we examined this in the first two points of our article on 10 steps to raise your game – part 1.

Do you expect a computer to think like this? I am sure the answer will be in the negative. But let us leave aside this question and ask a more basic one: does a computer think at all? From whatever little I know of computers, I understand that a computer’s seeming intelligence is the cleverness of its programs put inside by human programmers. In fact one of the aims behind creating progressively stronger chess-playing computers is to understand the nature of intelligence and the thinking process of humans. Are our thoughts simply the results of logical process linking appropriate information from our memory of accumulated data in our brains?

That is very difficult to accept if you study the background of most theories and inventions. When Einstein propounded his theory of General Relativity in 1912, it was nothing but an interesting idea as it went against many concepts held dear by scientists till that time and something that could not even be proved. It was Eddington’s experiment 7 years later during the total solar eclipse of 1919 that finally established the bizarre idea of light rays getting bent by gravitation! This is what I understand as “thinking”. Will any computer ever be able to produce that kind of thought?

I am with you if such esoteric thoughts keep you off! Let me therefore bring our thoughts nearer home – the playing of chess. All of you know that a chess player has only 20 moves at his disposal when making the first move – 16 pawn moves (8 pawns taking either one or two steps forward) and 4 knight moves (each knight jumping forward of Bishop pawn or Rook pawn). I rarely play chess against computers because I like to see my opponent and get the feel of the physical chess pieces (not to speak of the chance, however remote, to sit opposite someone like Alexandra Kosteniuk – a computer could not care less)!

But unlike me, many of you must be playing online chess and hence my question to you – have you ever seen a computer make any of the 10 moves other than pushing the four central pawns (c-pawn to f-pawn) or Knight to c3 or f3 squares? Even pawn moves b3 or g3? Making a reply like a5 against White’s e4 opening? That is exactly what Preston Ware did in 5th US chess championship held at New York in 1880 while playing as Black against Congdon, Sellman, Cohnfeld, Grundy, Mackenzie, Delmar, Judd, Moehle, Ryan and Noa, and won against the first four players! A human player can go out of the shell at random, but not a computer (unless programmed to make random opening moves!)

What is the point of telling all these? Just that computers cannot really think independently and so bound to be week in strategy when compared with humans.

If they could do otherwise, then why bother with such mouthfuls like “QGD Slav Defense Dutch variation Main line” or “English Opening Caro-Kann defensive system Bogoljubov variation” and similar others! Just ask the computer to start successively with the permissible 20 opening moves one after another. Then let it keep working out both players’ best moves till the board situation reaches a position where a request for further analysis will make it emulate Rip Van Winkle! We can scrap all those 500 ECO codes and replace with only 20 codes – Rybka1, Rybka 2, … Rybka 20 or Fritz1, Fritz2, … Fritz 20 etc. We humans could then start where the computer went to sleep and take the game forward whatever we can! Wouldn’t life be much easier for chess players?

Chess Tactics:
Chess tactics comprise of several moves in sequence, your own and the expected responses of the opponent, the execution of which is supposed to give you some advantage – be it material, space, or time (tempo). This obviously needs a player to visualize correctly the successive changed positions on the board to make sure that the tactics are sound. With the increase in the number of moves involved in a combination, the complexity tends to increase exponentially. The capability of our brain to store these visuals in memory and our ability to recover these without error ultimately limit the number of moves we can foresee.

As far as memory cells are concerned, human brains may be superior to the best computers available now but the problem comes in recovering the stored data immediately when we need it. A computer has no such problem as its human designers have put in all the links that enable instant recall facility. The electronic circuit also allows it to calculate all the permutations of moves at a much, much faster rate. This gives a computer what we call ‘a brute force’ capability with which a human brain cannot compete. Agreed that with this method, a computer probably carries out umpteen times calculations most of which are useless and would be instinctively bypassed by a human being. But by the sheer speed of calculation, a computer can come out with the filtered output much earlier than his human competitor. All I can say, give the devil his due!

Also, keeping this aspect in mind, I would suggest that when you play against a computer, try to avoid tactical games which is a computer’s forte. Instead, go for playing those slow positional games with a lot of maneuvering for putting your pieces and pawns in strategically favorable positions. I am sure it would give you a better chance to prevail upon your computer!

Memory and Processing power
As already referred under the previous parameter, a computer’s all the memory is active memory whereas for a human being, it is only a small fraction of the total. The number of brain cells may be varying between humans but still remain within a limit and I have not seen any report that this capacity will keep growing in future. But computer memory and its processing power keep increasing every year and hence it will sooner or later exceed the power of human brain in this respect.

Physical factors
Another weak area for poor humans! If you take care that the power supply is stable and the processor is not allowed to get overheated, the computer can play at the same level 24 hours a day, seven days a week! Even when you are physically well-rested, does your efficiency remain same in the morning and in the evening, on Saturdays and on Mondays? If you try to analyze your own performance, you will surely find days in the week and time-slots in the day when you seem to perform better than at other times. If you have to challenge a chess-playing computer, you better choose your time and place!

Emotional factors
I do not know if others will agree but it is my conviction that the emotional state of the mind affects human performance probably even more than the physical factors. During a game if something unexpected happens, we get surprised. Out of surprise comes confusion and confusion makes us lose our track. We get annoyed and angry that we have lost the way and errors mount upon errors. Does the computer feel any such emotion? Not by any chance – it simply remains busy calculating, calculating, calculating….. This lack of emotion and retaining objectivity is a very strong plus point in favor of computers.

In fact, I have always wondered about the emotional state of Kasparov’s mind when he started the return match with Deep Blue 2 whose predecessor he had defeated in an earlier match. In all the 3 games he played as White, he opted for openings he had rarely played. When Deep Blue played as White, it opened with e4 but Kasparov responded with moves that he rarely used prior to this match. It was as if he was afraid that Deep Blue would be well prepared against his favorite openings and defenses (about 1/3rd of his games in Chessgames database have to do with Sicilian Defense) and he went for openings on which Deep Blue would not have enough data about his style of play!

It creates a nagging doubt in my mind that such emotional factors along with the stress of worldwide publicity made him playing to outwit the computer rather than playing to his own strengths.

This is what Yasser Seirawan had to say in his Inside Chess article on Game 5:

Garry was visibly shaken by this result. He stayed on stage at the board following the game for quite some time. … Garry wanted to win this game badly and I think he expected victory. When it wasn’t achieved his agitation increased. He will be very tense for game six with so much at stake. As he himself joked, “I hope I won’t resign in advance.”

Now that we know what happened in the 6th game, the above seems prophetic, doesn’t it? And that is why I laid so much stress on emotional factors.

Intuition
We have already discussed the role of intuition in Intuition vs. logic in chess and this is something that a human player will use, not the computer. An expert chess player takes an overall look at the board and intuitively rejects many of the moves without even bothering to calculate their outcome. In the same situation, the computer will calculate everything may be twenty moves deep and then reject those same moves as discarded by the human expert. Only the sheer speed of calculation makes it possible for the computer to come out with its moves in a reasonable time. The human player is slower in thinking, but as he is calculating only a limited number of lines, he appears to be capable of matching the computer move for move. I am sure many of you win at least some of your games against your chess playing program and that is possible because of your ability to ‘read the position’, what your program never does. This is one factor that only a human mind possesses.

Inventiveness
What is inventiveness? It is the ability to devise or contrive, to design for the first time or originate, possessing creativity or original thoughts, showing imaginative skill. Imagination is formed in your mind and nobody has claimed that a computer has a mind of its own! So this is one quality that is the prerogative of a perceptive intelligent mind and humans will score much above computers in this aspect.

All of us have read stories of how some apparently insignificant event from daily life has led to extraordinary inventions owing to the curiosity (a mental state?) of the inventors to find the explanation of those events. We are unable to authenticate the stories linking the boiling kettle to James Watt’s engine or the falling apple to Newton’s laws of gravitation. But we do not reject them also as fiction because those appear so plausible for a human mind. At least no one doubts the story of a laboratory research on staphylococcus culture going wrong because those germs died in contact with some fungus-like thing that accidentally fell into that Petri dish. A computer would reject the sample as an aberration but Sir Alexander Fleming’s questioning mind sought answers for this peculiar observation, leading to the invention of penicillin that literally saved millions of lives. Will you expect a computer to do this kind of ‘tangential thinking’?

Where does this aspect relate to our game of chess? Without trying out latest computers to analyze Levitzky-Marshall game position after White’s 23rd move to see if they come out with Marshall’s reply, I cannot comment on a computer’s inventiveness at least regarding the game of chess. I leave it to the readers to throw more light on this.

**********************************************************************

Well, that is all I could think of on this debatable topic. Probably another 10 years will take us nearer the truth on the subject of this article. But in the meanwhile, can you readers who have powerful hardware and chess software at their disposal carry out an experiment and bring the results to this forum? Please load the following position and find what should be White’s next move and the winning line.

Position after 13 moves in the game Breyer vs. Esser, Budapest, 1917 with White to play:

rnbq1rk1/p3bp2/2p1p1p1/1p1nP1P1/2pP1P2/2N5/PP4P1/RBBQK2R

Those who are aware of this game know that White’s next (14th) move is considered by chess experts as one of the deepest moves ever played on the chessboard. Just see if computer finds this move or something even better!

Gazing into the crystal ball
I understand that the designers of Rybka has thrown a challenge to FIDE rated players to win against it. If so, the writing is on the wall regarding the future of human players against the chess computer.

Nonetheless, just for fun, I would suggest the people who own it to try the following to test its mettle:

  • Play in strategic and positional lines, avoiding tactical games.
  • Computer chess seems to prefer bishop against knight. If you are adept in knight maneuver, build your game around this offering your bishop against its knights.
  • I expect a chess software to have built-in safety considerations for the King. Try to throw it off-balance by marching your King as an attacking piece (possibly after creating locked positions). Keep in mind something in the line of Chess Tactics: The King’s role in attack – part 2.
  • It is superfluous to say that you need a computer to play computer chess! Your time to think for a move does not depend on computer, but it does for the software. The chess software may think deeper than you, but how long will it take? Set the tournament rules for your game and see if it defaults on time limit.

A very (possibly the most) important part of the foundation on which the development of such chess software stands is the progress in computer hardware. When we look at the strength of future chess software vis-a-vis humans, we are always thinking of this factor. But we keep overlooking that progress in bio-technology and bio-sciences are likely to bring advantages to human players also. Through implant of micro-electrodes in brain to enhance neuron cross-connections, the brain’s active memory capacity and computing power in future may get augmented manifold. So, you need not yet despair that the days of Grandmasters are numbered!

Epilogue
I read an interesting paper by Hans Moravec of Robotics Institute, Carnegie Mellon University that was published in 1998. Things must have progressed way beyond in the decade that has gone by but it is worth noting what he said in connection with the celebrated chess matches between Garry Kasparov and IBM’s Deep Blue computer.

Now, the team that built Deep Blue claim no “intelligence” in it, only a large database of opening and end games, scoring and deepening functions tuned with consulting grandmasters, and, especially, raw speed that allows the machine to look ahead an average of fourteen half-moves per turn. Unlike some earlier, less successful, chess programs, Deep Blue was not designed to think like a human, to form abstract strategies or see patterns, as it races through the move/countermove tree as fast as possible.
 
Deep Blue’s creators know its quantitative superiority over other chess machines intimately, but lack the chess understanding to share Kasparov’s deep appreciation of the difference in the quality of its play. I think this dichotomy will show up increasingly in coming years. Engineers who know the mechanism of advanced robots most intimately will be the last to admit they have real minds. From the inside, robots will indisputably be machines, acting according to mechanical principles, however elaborately layered. Only on the outside, where they can be appreciated as a whole, will the impression of intelligence emerge. A human brain, too, does not exhibit the intelligence under a neurobiologist’s microscope that it does participating in a lively conversation.
 
In forty years, computer chess progressed from the lowest depth to the highest peak of human chess performance. It took a handful of good ideas, culled by trial and error from a larger number of possibilities, an accumulation of previously evaluated game openings and endings, good adjustment of position scores, and especially a ten-million-fold increase in the number of alternative move sequences the machines can explore. Note that chess machines reached world champion performance as their (specialized) processing power reached about 1/30 human, by our brain to computer measure. In coming decades, as general-purpose computer power grows beyond Deep Blue’s specialized strength, machines will begin to match humans in more common skills.
(the underlines are mine)

Though I am no expert like Mr. Moravec, but my common sense (or is it nonsense?) makes me disagree with some of the points raised by him. As mentioned in the first paragraph, I do agree that the distinctive feature of human thinking is the ability to form abstract strategies and see patterns. But the implication of his second paragraph appears fallacious to me. If I have understood him correctly, the issue can be rephrased as under:

A human brain will not exhibit any sign of intelligence under the neurobiologist’s microscope but human interactions show this intelligence.

A computer’s innards do not show any intelligence to the computer scientist, but the computer’s external responses give an impression of intelligence.

Therefore computer intelligence and human intelligence are similar.

Is that the conclusion and is it valid?

Mr. Moravec laid a lot of stress on the reaction of Kasparov about his feeling a sign of intelligence in Deep Blue after he lost the match. But that may just have been an instinctive reaction to salvage his pride (an emotional issue!) after the shock of getting defeated by a computer. Read his reaction together with what Mr. Seirawan (who was a commentator during the match) had to say and you may judge if my views are totally unfounded. I would go so far as to say that Mr. Kasparov was defeated not by the superior speed and memory and calculating ability of Deep Blue, but by the weakness of a human under emotional stress which did not exist for Deep Blue. Most of you may have played out those six games of the match. Would you agree that those games hardly show the real Kasparov we have revered, a far cry from the type of game he played against Topalov at Wijk Aan Zee in 1999? In my view, emotional factors will remain the Achilles heel of humans when playing against computers!

I would like to end my ramblings with a silly question based on a real-life story of mine.

Many years ago in a gathering of friends, the discussion somehow veered towards English as a language and someone asked if anyone knew the longest word in English. He claimed this honor for ‘floccinaucinihilipilification‘ (29 letters!). We had to check the dictionary to verify the existence of this word (it is there!) but since no one could come out with a longer one, we had to accept his word for it. But then the joker amongst us claimed that ‘smiles‘ was the longest! How could it be? He replied with a straight face that a mile separates the two end letters, so no other word could come anywhere near!

Give this question to a computer and in less than a minute it will search all the English dictionaries in existence and come out with the word which may indeed be the one quoted above. But will it think like that joker friend to give another interpretation to “longest”?

 

3 Comments

  1. yanquis1972 says:

    the hypothesis is -very- misguided, & has been for years. computers, even on very basic hardware, are -much- stronger than humans. the only reason a computer is not world champion – even if they were limited to 1 cpu laptops – is because we’d have about a dozen computer programs, followed by some humans, were they allowed to play as humans do. no one would want to see that. the reason we don’t have computer-human matches is because the results are already decided. computers won awhile ago.

    the most recent performance by a computer against humans was just a few weeks ago, when pocket fritz 4 scored 9.5/10 against a field of international masters — while running on the equivalent of an iphone.
    http://www.chessbase.com/newsdetail.asp?newsid=5719

    rybka 3 on a decent 64 bit core? forget about it. it might take years to lose a game.

  2. says:

    Rybka 3 is indeed a very strong engine. Even the 32bit version. But it is not correct to say that it will win every single game certain openings will lead to a draw. But then again you never know. I do agree that currently Rybka 3 can easily beat every GM out there might get a draw but I dont see it losing.

  3. The advancement of artificial intelligence might one day make man slaves of machines like in the Matrix.


Trackbacks/Pingbacks

  1. Chess Lessons – Improve Your Chess » Blog Archive » Man versus Machine: When a Computer will become World Chess …
  2. IT Corner » Blog Archive » Man versus Machine: When a Computer will become World Chess …
  3. Just Chess » Blog Archive » Kasparov Computer Chess – Play it the Grandmaster Way
  4. Chess on Net » Blog Archive » Computer Chess Game Benefit in Today’s World
  5. Chess Software News and Reviews Home of Chess Programs and Chess Engines » Computer Chess Clock Makes the Game Even More Exciting
  6. Computer Chess Online | The Chess Club
  7. Playing Computer Chess – Valuable Tool that Helps You Learn at Chess
  8. Chess Guide » Chess on Computer – An Opportunity to Learn How the Game is Being Played
  9. Chess Blog :: Blog Archive :: Computer Chess Test Online