Computers in chess... Good or Evil?

  • GM Gserper
  • | Feb 12, 2013

Computers are an essential part of the modern world and it is impossible even to imagine our everyday life without them. As a matter of fact, you wouldn't be able to read this article if there were no computers and the miracle called "Internet". But let's talk about the computer's impact on our beloved little world of 64 squares.  Is it positive or negative?

Please do not rush to call me the Luddite who hates Progress. Let's just analyze the positive and negative sides of the computer's invasion.

It is a well-known fact that computer is an excellent tool to study chess. And I am not even talking about the unlimited amount of chess related material you can find on the Internet. Simply by using a chess engine you can significantly improve your chess.  Just ask Hikaru Nakamura who claims that he hasn't read a single chess book and credits all his achievements to the countless hours he spent with a computer. So, you play with a computer, analyze with a computer and at some point you start think like a computer... at least according to Nakamura. Hikaru likes when he gets compared to a chess engine and therefore in the following game we can witness the battle between the best human computer and one of the best silicon monsters of that time, Crafty. The game is remarkable because the majority of computers don't have the word 'resign' in their vocabulary and play till the bitter end. Playing with a lonely King against   4 or 5 pieces and pawns of your opponent is not the best strategy against one of the World's best players. We discussed this problem here:  ( )   Hikaru found a very elegant solution.  He ...well, I am not going to spoil the fun, judge for yourself!

Still, I wouldn't recommend the Nakamura's way of learning chess to everyone. If you don't have his enormous talent, I would advise the traditional approach of learning using chess books. I still believe that one comment of Capablanca or Fischer is more valuable than a dozen of games against a chess engine. Moreover, in my opinion, had Nakamura in his childhood combined his purely computer-based training with old fashioned chess books, then the number one player in today's rating list could have been different. I find it very symbolic and significant that the current number one Magnus Carlsen is not a 'human computer' in any way. He is a superstar, he is genius, but he is not a computer! When we analyze his games we can see reflection of old masters: the technique of Capablanca, calculation precision of Kasparov and the Fischer's desire to win.  But I digress here...  

I still think that a computer is invaluable teaching tool. For example, when I analyze my student's games and see that he mentally gave up at some point, I ask him what happened and he usually says that the position was completely lost and the game was over anyway.  Then I suggest him to play this position against a computer and see if he would be able to beat the beast at least once. Or when you analyze a very dry, technical position with a computer, it constantly surprises you with a maze of unexpected combinations. But there are certain limitations you should know about before you start working with a computer.

You can sum up all these limitation with one short statement: 'computers do not understand chess'! Yes, they can beat the World Champion and yet they don't really understand chess. It is all about brute force or by other words an amazing ability to calculate, calculate, calculate. It is for this reason Botvinnik called a chess playing computer a 'tireless idiot'. If you are confused, let me show you a simple example:

This is a basic position of a Bishop of a wrong color.  the position is a 'dead' draw.  How dead?  If you have a friend who doesn't know the rules of chess and just tell him how the Chess Kings, Bishops and pawns move , then explain him to move Black King back in force keeping it in the corner 'h8', he will be able to make a draw against the World Champion even if he doesn't know how the other pieces move! ( we discussed this position here: )
However, a chess engine would evaluate this position as completely won for White (assuming it doesn't have an endgame library or table bases).  Here we can see the phenomenon known as 'horizon effect'. A computer can analyze a certain number of moves ahead, but it is not enough to see that the position is  a draw. That's why a chess fortress is a completely foreign concept for any computer! Therefore, no computer will ever be able to solve the next relatively simple puzzle:
The final position is an excellent example of how helpless computers become in this kind of a situation! Any chess engine will indicate an easy win for Black, while most of human players won't have any difficulties to understand that it is a draw.
To be continued...


  • 4 years ago


    RHoudini... What you say right now proves that you rarely play real chess tournaments... The tourneys on are educational but cant be compared to the real deal.. And chess engines donno anything or even know the concept or the meaning of the word "Psychology" of the game... Your argument is invalid due to less experience on the actual field... And if you really did have enough experience in the field you would realize how important it is... So before you post here, saying all this "disappointment" and such... Please go play at tourneys where Masters play and see what ACTUALLY happens... 


    Also, these past comments have gone astray from the point of this article. This article is about whether chess engines are the best method to learn chess... The answer is a clear and resounding NO... If you disagree, prove it... But dont post something stupid please

  • 4 years ago


    I have also studied using both methods (computer and book) and can only say that my game improves much more from one game annotated by a master (Lasker is my favorite) than hours of mulling with a computer.

    The human-master insight is priceless.

    Also; I thought it was common knowledge that A.I. Programmers tried to make computers play chess like a human AND DID NOT EVEN COME CLOSE, so they gave up and went with brute force.

  • 4 years ago


    Chess will survive as a game for a while, but the popularity will continue to wane given all the alternatives these days (video games, etc.). As a sport, chess has already been ruined because one can never know for sure when an opponent is cheating (even in tournaments!). This is a real shame...there is no sport in facing a cheating human.

  • 4 years ago


    Very nice article.

    Wish to see the continuation soon.

  • 4 years ago


    computers are more powerfull in every single aspect of chess than humans, and "they" will be better. Humans dont want that to be true. Believe wathever you want, but chess will always be a game so i enjoy it more playings vs real people.

  • 4 years ago


    I've gotta agree with RHoudini here, the idea that chess engines are still simply brute force calculation of lines isn't true anymore. I remember reading that there are now chess engines that can play as well as Deep Blue did in 1997 but with far less moves calculated.

    Another important point is that we don't realize how much human brains skilled at chess employ a kind of brute force calculation. This is not the same as the method currently employed in engines however - research suggests that skilled and Master chess players' brains compare the position on the board with many thousands that they have memorized, all in parallel. Of course nobody feels like that's what they're doing, as with much of what our brains do we're not conscious of it. And of course it is combined with other forms of chess knowledge, like general concepts and strategies.

  • 4 years ago


    Interesting.  2 moves before 50 moves rule Fritz 13 realized that "he" should try to give a rook... :) And, of course, it didn't find Ba4+.

  • 4 years ago


    @RHoudini > Indeed, crafty 1.6a x64 finds Ba4+ in a split second. What is interesting is that Houdini 2.0c x64  doesn't, even after several minutes... It's weird, since Houdini 2.0c x64 is supposed to be (a little) better than crafty 1.6a (according to this list : Do you have an explanation? Does Houdini 3 find Ba4+? (I don't own it)

  • 4 years ago


    amusing and elegant..

  • 4 years ago


    The real game of chess has nothing to do with computers. Using computers to imporve your game is silly. Chess is an art. It can only be learnt by playing. All this computer fuss is so ridiculous...

  • 4 years ago


    I think computers can get a lot better at playing like humans (playing weaker, but more similar to human choices and human mistakes) That would make them more worth playing. Also they can get a lot better at giving good advices other than just the evaluation in form of +X and giving the best lines. Lucky us (or at least for chess coaches) we are not there yet and advices from strong human players are a lot better than computers, but there is no telling about the future. If computers would manage to set a good rating strength depending of the quality of the human play and give good feedback with explaining the mistakes in analyses and give some advices to improve and give similar motifs to study there is no telling how good computers could be at teaching chess.

  • 4 years ago


    Computers can't read human mind like human do i can create traps or leave a piece un protected to create a trap to make a instant check mate or capture a better piece computers are just calculators i can do empathy and figure out my human enemy's next move i can put my self in his/her situation but computer can't. A creative human can beat a human trained by computer sooo easily. And yes computers are good for training "best moves" but when it comes to memorising computer's moves it will ruin your creativity and i say you can be better with combining workouts against human and computer.

  • 4 years ago


    I am no chess genius or at least I do not consider myself to be one, but it doesn't take much of a genius to know that computers and computer software are made by humans. Not other computers. It is humans who have discovered the technology to build the computers and to create the software. I guess playing against a computer is good if you want to find a way to learn how to play the game without letting your own personal emotions get in the way, because that is about all the good comes from playing against a computer. If the computer wins or loses, it doesn't throw some kind of tantrum like an eight year old who is told they cannot have dessert unless they finish all of their dinner. I will prefer to play hours of chess against another human as opposed to playing against a computer any day of the week.

  • 4 years ago


    I don't like computers in chess at the grandmaster level because it extends to horizon of "playing on your own" way too far, and to me winning as a GM becomes more about "preparation" than talent. The strategy of chess begins before the game now!. What variations is my opponent good at/ likely to play, what can I prepare for them, etc. The game is begun well before the board, which it always had been, but before computers it was the mind of the player alone creating the prep work. Thats why I think GM games are "boring" now, because the postitions reached are not where a human would take the game and therefore harder for an amatuer or spectator to appreciate. 

  • 4 years ago


    Btw, Rybka found this crazy variation where White actually wins in the last position if Black doesn't play carefully.

  • 4 years ago


    The article and some of the comments repeat a number of ideas that were true in the past, but with the current state of computer chess have  become misconceptions.

    Engines have considerable knowledge. For example the bad bishop position shown in the article is part of the knowledge of nearly every engine. An engine like Houdini 3 has more endgame knowledge than the typical 2500 player.

    Engines are no longer materialistic. Current top engines will easily play 20 moves with a pawn less if they believe they have sufficient compensation. For example, please verify the Houdini 3-Rybka 4 game recently played in the TCEC tournament (see Houdini 3 sacrifices a pawn around move 14 and maintains an advantage throughout the game despite being a pawn down.


    As a general comment, this article would have been fine in the year 2005, but in 2013 it's pretty much outdated. My comments here reflect my disappointment at reading such a poor article at

  • 4 years ago


    Sorry, I am far more inclined to believe the GM who has spent years studying chess than a guy claiming to be Houdini's creator from Belgium named Robert. 

  • 4 years ago


    I still think we know more about chess than computers do.  Computers are materialistic and don't understand pyschology like we do.

  • 4 years ago


    @Tjornan, below Critter 1.6a analysis, finding Ba4+ in less than a second with a draw score:

     12/33 0:00 -5.97 1.exf6 Ra8 2.fxe7 Rxd8 3.exd8Q Bxd8 4.bxa3 bxc3 5.fxg5 Re7+ 6.Kd3 c4+ 7.Kxc3 Re3+ 8.Kb2 Bxg5 9.a4+ Kb4 (2.444.255) 2916

     12/33 0:01 0.00 1.Ba4+ Kxa4 2.b3+ Kb5 3.c4+ Kc6 4.d5+ Kd7 5.e6+ Kxd8 6.f5 Rd7 7.Kd3 Bb6 8.Ke3 Rb7 9.Kf3 Ke8 10.Ke3 Bd8 11.Kf3 Bb6 (2.862.898) 2828


    As I said in my earlier post, it is rather clueless to assume that current engines cannot solve fortress positions.

    BTW, I am the author of the Houdini Chess Engine, I hope you will pardon my misconceptions regarding computer chess :).

  • 4 years ago



Back to Top

Post your reply: