Beating the Bots

Everyone complains of his memory, but no one ever complains of his judgment

BEATING THE BOTS

If you are new to playing against the computers it can be very rewarding and a lot of fun. It will surely be a great learning experience and will increase your chess skills because of the intense motivation it can create to beat this unwilling stubborn bot. If you have been playing them for a while then you may have felt powerless and frustrated at times at its uncanny ability to zap you with some surprise maneuver that either annihilates your carefully planed attack or eviscerates your king's fortress. Ripping and trashing the pawns to lay the king naked with no place to hide. And of course on the next few moves you get thoroughly slaughtered and destroyed.

We all know where the bots strength lies. It is in its superior tactics play and lightning speed being able to calculate millions of moves a sec and zooming ahead many plies to look for the next best moves. Unless you are very good and really enjoy a tough challenge, getting into a battle of tactics early in the game with very limited time is sure folly. The owners of these bots know this and this is one way that they can make a weak program much stronger. They simply severely restrict and limit the time and allow no increments. Although they may say they do this so that more people can play their computer the real truth may be more in that they are not putting this comp on line just for your enjoyment. It may be the prestige and esteem it brings them from their peers of having a very strong computer on a world wide chess server or for their ego, knowing that they have a comp that can beat the crap out of 95% of the people who play chess on FICS, including them.

So if you are just starting out playing the bots then in my opinion I simply would not play any computer that did not give you fair and decent time controls. What are fair and decent time controls? That is entirely up to you, but to me it means having liberal time increments or no time controls at all. There are plenty of other ones to play that do. Why lose on time rather than your skill if you don't have to? Take as much time as the bot will allow you and use all that time just as efficiently as you can.

The equalizing injustice of chess dictates that there is no one formula for success that one can easily reduce to the sum of its parts. But there are areas of knowledge that will help you to improve your chess skills against playing the silicon, terrifying monsters.

Finding out about the bots weaknesses you intend to play is your first primary task. Bots don't intentionally or unintentionally make bad blunders unless they have been programmed to do so. Weaker bots just play looking ahead less plies and may have a smaller databases in the opening book and end game, or have no end game or no opening book database at all. So you may not play to the blunders or mistakes very well, but there are things you can do to make the bot make poor decisions tactically.

Unlike humans bots are consistent in many of the things that they do. They may make the same mistakes over and over again, unless it has learning programmed in it and can learn from its games. Its style of play will never change and many other idiosyncrasies also will stay the same. So on your first game you may just want to play an exploratory unrated game to gather information on how this bot plays. Analyze how it reacts in certain situations. Things like style of play, how aggressive does it play. How does it Attack, how does it defend. Is it primarily concerned with attacking your king even if it means neglecting its own kings safety or other considerations?

The opening book database may or not be extensive but it may be a good idea to "Get out of Book" early in the opening forcing it to use up its time thinking about its next move rather than just pulling it out of the book. Some bots ponder on your time some do not. Look in the finger notes for info on this too. Most computers are slow learners or do not learn at all. It may depend on the comp being programmed to learn from its mistakes. Most are very greedy and will take anything you hand it, even if it leads to a mate. They love to grab pieces whenever there is a chance to do so often not seeing a simple trap you may have set for it. Many are programmed to no play you if you keep using that same trap over and over again just to gain points. But if you use a variation of that play it may not catch on you are basically using the same trap over and over again to zap it. For example, many bots do not realize that if you are giving it your queen you must be doing so for a very good reason and the bot will gladly take it with out a second thought, even if it is leading to a quick mate.

Here is a fine example in the Legall's Mate:

Legall's Mate

But be prepared that it may backfire on you too. This is where you must do your homework on each bot you play. Go to the history and examine other games to get a feel on how this bot plays. Get a notebook and use a page for each bot. Put in every thing in the finger notes and the vars.

Next play some unrated games to analyze how it reacts in certain situations. Test out its traits on such things as, what defenses does it use to defend itself, does it value bishops over knights? Put in as many notes as you can find about its style of play. How much value does the bot place on control of the center, mobility, King safety, Passed pawns, Pawn weakness. How hard does the bot try to control its positions where a pawn is unprotected and is rendered immobile or otherwise weak in the overall pawn structure. How much time does it take for certain positions? May be you can use this info to get it to waste time in similar positions. Develop unusual tactics whenever possible.

Provoke moves that it does not understand and are not in any of its databases. Make the bot feel uncomfortable by making moves that develop your position, restrict its play and limit its choices. Moves that may lead to a stronger attack or a better end game later on in the game.

If the comp is using learning then it may remember when a particular move wasn't successful and use a different one in a comparable situation. The program is then said to be dynamic. Which means that it plays by the principle of trial and error so it learns from making moves that caused it to suffer a loss in material or lose the game. But using the techniques of Aron Nimzovich's "positional play" and making prophylactic moves, and moves that are simply development moves or moves that put more and more pressure on a position or strengthen your hold on a file or diagonal make it hard for the bot to decide on a good tactical reply from its data banks. You are forcing it to play strategically and bots are not very good at doing that.

Exploiting Small Advantages, and Using Dynamic Pawn Play.
In the opening stage of chess, pawns define the redoubtable structures of Black and White fortifications. In the middlegame, pawns serve as rams for crashing through the opponent's defenses, or as armored shields to beat off an attack. In the endgame, pawns are the masters who hold the game's fate in their hands.

Pawn Structures
As a general rule, the hardest defenses to broach are the ones where your opponent has made very few concessions in his pawn structure. The simple reason for this is that the pawn is by definition the best defender available to the chess player as it is only worth 1 point. This generalization is a good rule to remember.

Creating Pawn Weaknesses
To develop quickly and seek an early initiative is not enough. In order to attack, one must first create the right conditions for it. One has to perceive a weakness in your opponents position and exert pressure against that position. You have to find, or create a positional error and take advantage of it. Perceiving the importance of pawn structures, weak squares, doubled pawns, isolated pawns, backward pawns, isolated pawn pair and other pawn weaknesses can make one appreciate the importance of pawns, especially those on the central files.

Get a Passed Pawn

If you can create a weak pawn structure for your opponent you may be able to also create a passed pawn. A pawn is passed if it is free to advance all the way to the last rank unhindered by enemy pawns. That means no enemy pawns in its path, no enemy pawn guarding any square along its path..

The real advantage of a passed pawn is that it has a real chance to become a Queen, and a Queen is almost always decisive. If there are no pawns to stop the advance of the passed pawn, the enemy pieces may have to do it, which diverts them from other actions and responsibilities. And even then, the pawn could promote anyway. If you time the advance of the passed pawn properly, you could frustrate all defensive efforts to stop it.

Try to create a passed pawn and use it to lure away the enemy pieces from other duties and activities, but don't push your passed pawn recklessly. That could lose it. When advancing it, make sure you can support it with other forces and that it cannot be won without your opponent making significant concessions.

Your passed pawn may become more important in the endgame and late middlegame, especially when your king is not endangered and can take part actively. Your game may come down to a race between passed pawns. The one who queens first usually wins because his queen can give the first check. This type of strategy is often not too well understood by the bots because it is not a tactical problem. The bot may be diverted from moving his passed pawn to queen to try to stop you from queening your pawn or pawns and if you can get it to do this you may have a winning game.

Blockade
The blockade was conceptualized and popularized by Aron Nimzovich (1886-1935), it refers to the tying down or immobilization of an enemy pawn by placing a piece, particularly a Knight, directly in front of it.

Pawn structure, pawn formation, or pawn skeleton principles is one of the aspects of chess least well handled by chess computers. So studying pawn structures will maximize your results against the silicon monster. There are four types of pawn structures: 1. Open Center. 2. Closed /Blocked Center 3. Fixed Center. 4. Mobile/Dynamic Center.

How should pawns be used to fight for the center? How does the central pawn formation affect planning for both sides? These issues are central to understanding on how to beat the bot. A closed/blocked center or a solid fixed center may make the bot make a sacrifice or take chances to open it up and also make it make tactical errors and mistakes if you defend with prophylactics both externally and internally. Prophylactic play and over-protection is one of Aron Nimzovich's original Conceptions of positional play.

Endgame
It is well known that chess programs don't understand how to play some types of endings very well. A endgame tablebase added to the chess program precisely tells the engine how to play out any position once a certain amount of pieces are left on the board. It is most likely that most bots on an ICS server have not been installed with an expensive table base. The Nalimov Ending tablebases, complete set consist of 9 CD's and is 700 MB before installation. After installation the file can grow to over 2.6 GB by adding 6 piece endings with its tablebase installer program. The complete set Nalimov Ending tablebase has all 3-4 and 5 piece endings and cost aprox $50.00.

Creating small weaknesses and then using the accumulation of the small advantages can lead to a significant advantage in the end game where bots have fewer skills and in general play a much weaker game. Time spent studying end game play will be much more useful than studying about the various openings. Also in the end game there are fewer pieces on the board for tactical play and tactical surprises by the bots.

The Closed Center

Since the bots are so vulnerable in the endgame, one plan you might try would be to create a fortress like a closed center pawn position and just sit back and wait to see how the bot will respond, maybe beating it's brains out against your solid brick wall defense. In the closed pawn center, each side has two pawns, interlocked and incapable of advancing. With the central files walled up, play proceeds at a slower pace and different principles apply. Since play cannot materialize in the center, both combatants look to the wings for activity. Each player aims for the base of the enemy pawn chain. Development does not have to be so rapid, in fact it can't be. The central barrier inhibits the pieces from coming into play. This also means that the Kings are not particularly endangered in the center. Castling may even be unnecessary. In some cases it might be prudent to keep the king in the center, possibly not castling at all. The center is often safer than the flank in the closed positions. Strategy and long-term planning assume great importance. You usually have time to maneuver against your opponent's weaknesses, something hard to do when the center is open. I once won a lost game with a bot because I tricked it to turn its attention from pushing a passed pawn that I could not stop to Queen, to try to stop me from Queering two of my pawns. Some times in the end game bots will do unexpected things and make unexpected errors in judgement. By making the bot think about strategic judgements in place of tactical play could give you a winning game.

Sometimes to defer, and postpone to playing for the end game where you may have a much better chance of wining the game may be your best strategy. Just make sure you play using plenty of time and use increments so that you may accomplish this line of play. I frequently see so many people play a 3 / 0 game, or less, with a bot and then lose it on time and then rematch and lose again on time doing this over and over again. Why do they do this if they don't have to? Do they suffer from attention deficit disorder, ADD,? Or maybe they are just hoping the bot will make a big blunder letting them win. If you do not have the patience to play longer games of chess than a 3 / 0 game then maybe playing against the bots is not for you at all because that may be the only way you may have any chance at all to win against the stronger bots.

"Be Ruled by Time, the Wisest Counselor of All" - Pericles

Creating a Fortress Position
The closed games normally begin 1d4-d5 although there are some transpositional lines from the Dutch Defense and some other openings. Playing 1d5 right away effectively rules out the advance of either e-pawn to the center. In general closed game openings lead to quieter, longer struggles. For Black, they have the advantage of being very solid.

Dutch Stonewall Defense. As the name implies Black creates a fortress of pawns, d5, f5, and c6 in the center and plays for control over the e4 square. The Stonewall is easy to understand. Black sets up a solid pawn structure keeping the position closed. The modern handling places the dark squared bishop at d6, keeping an eye on the important e5 square. The other bishop can be a problem if it gets stuck behind the pawns forever. It can go to the Kingside via d7-e8-h5 or take up a useful post at b7.

Dutch Defense Stonewall Variation

Use the Hedgehog Defense, in the English Opening to set up a black cramped formation for white and one, which is incredibly tough, and can leaves gaping holes in the rear of Whites position. 1.Nf3 c5 2.g3 and b6 initiates the Hedgehog 3.Bg2. In this position White should carefully consider his move order against the Hedgehog. Black's strategy is to fianchetto both his Bishops and leave it up to White to define the center.

English Opening Hedgehog Formation

Caro-Kann Defense. Is strategically simple. White moves e4 and Black will advance the c pawn to c6 on the second move. Next White moves 2.d4 and Black confronts the White pawn at d5. Even though players often castle on opposite wings, the Caro-Kann can not be defeated by direct attacks. The Black position can usually absorb whatever White throws at it, and complex endgame play is typical.

Caro Kann Defense Main line

Queen's Gambit. If black accepts he shows a willingness to concede the center for rapid development but this can be a dangerous strategy. If black declines there will be a vigorous fight for control of the central squares. Black should develop quickly with Nf6, Be7, and Kingside castling. The lack of open lines makes it very difficult for White to mount an effective attack.

Queen's Gambit Declined Classical Variation

King's Indian Defense. Is currently the most popular defense to 1.d4 for Black. It is based on solid principles of development and counterattack that typify the Hypermodern School of chess Many other hypermodern defenses could be used successfully like the Nimzo-Indian - Defense and the French Winawer variation but the hypermodern principles of Aron Nimzowitsch are essential to playing these games effectively.

As early as the 1940s and 1950s, David Bronstein and Isaak Boleslavsky went further than Nimzowitsch, expressing the conviction that Black should neither seek symmetry in the center nor try to control it. One should cede the center, they proclaimed, finish basic development as soon as possible, and then try to fix and undermine the opponent's center by side-blows.

The key was to fix the center, which meant to provoke a blockade, and sap the center of its dynamic potential. They relied on the simple, universal truth that whatever is fixed, immobile, has a tendency to grow weaker. It was exactly on these new propositions that new, modern opening systems were introduced, with the King's Indian Defense conspicuous among them.

Kings Indian Defense Classical Variation

How do bots think?
Many bots like to begin with E4 an open game opening. E4 is an excellent opening move because the move enhances the principles of opening play. Control the center, quickly get a safe King, and prepare for a equal middle game. It makes perfect sense. With this first one move White will attempt to create the ideal opening to control the center. It occupies the e4 square and controls the f5 square and the d5 square, the sweet center. It opens the diagonal for the f1 Bishop for quick development and also helps to prepare for the King to castle early. It opens up a diagonal for possible queen development. White has now seized half of Black's sweet center and if allowed, white will seize the other half of the sweet center with d4. If Black responds with e5 then white can attack with 2.Nf3. An excellent reply. It develops his knight, attacks the e5 pawn to destroy it right away. It covers the d4 Square and allows for his king to castle early in the opening moves. If White immediately attacks the e5 pawn with his d4 pawn and seek its removal, Blacks 's choices are very limited. Defending the e5 pawn in general puts Black at a disadvantage. If Black plays 2d6 then White can reply with 3.dxe5! dxe5 4.Qxd8+ Kxd8 If black allows this line of play White will have achieved a lot. Black has lost the opportunity to castle and his King has been forced to move into the center of the board open for more attacks. White now has a superior game.

How can black avoid this kind of play? The solution is to develop resilience in your defenses. When you play the bots you must change your style of play and instead of trying to beat the bots you must develop a new style of play and just let them lose the game, making many strategically unsound errors. (As in below what your goals should be) When you try to force the issues and play aggressively the bot will try to match you move for move and you may have to make serious concessions just like in the above example. The bot is not affected by any kind of human psychology like threats whatsoever. The computer simply plays the position. So it's up to you to create the position that the bot will play. It is vital to remember that the position on the board holds the truth.

Never assess a position that you think is a winning position by force. It is in this state of mind that can be linked to complacency and it is in this state of mind that many loses occur. You must try to force yourself to look at more possibilities, look deeper ahead in what may happen if you make that move. Try to see all the threats on the board before you move, and only then should you make your move.

The Eight Deadly Sins of Chess
Don't lose to any of the eight deadly chess sins.1.Thinking, erroneously. 2. Blinking, missing opportunities. 3. Lack of resolution. 4. Wanting. Too much concern with the result of the game. 5. Materialism Lack of attention to non-materialism factors. 6. Egoism Insufficient awareness of the opponent and their ideas. 7. Perfectionism Running short of time, trying to hard. 8. Looseness, losing the plot, drifting, poor concentration.

What your Goals Should be.
What your goals should be is to try to make the bot make many anti-positional moves. An anti-positional move is one that is unsound strategically. It differs from a blunder, which is an error or oversight that loses a piece or pawn. An anti-positional move violates the logic or direction of the game. Usually, it is a pawn move that weakens a key square or group of squares. An example of an anti-positional move would be to castle and then make the king vulnerable by moving the pawns that are shielding it. Creating these weaknesses in the castled fortress is anti-positional and another area of strategic play that computers may not understand.

Remember that creating many small weaknesses i.e., many anti-positional moves and then using the accumulation of these small advantages can lead to a significance advantage in your over all position and eventually the end game as well. The accumulation of numerous small advantages and weaknesses could cause the bot position to crash so significantly that your successful attack on the king will be assured. Its like chipping away at the castle's fortress walls a stone at a time.

Create a Bind
When one side's pawns and pieces are so well positioned that they prevent the enemy from moving freely, a bind is created. A player who is in a bind is not only cramped and unable to make freeing pawn moves but is also vulnerable to attack.

In the Sicilian Defense, when white plays c4 the dreaded Marcozy Bind, White's pawns at c4 and e4 make it extremely difficult for Black to free his position by advancing the d pawn to d5. As long as White can hinder that d pawn advance, Black will have less space than White and a more restricted game.

Accelerated Dragon Variation Marcozy Bind

Try to set up binds by using your pieces and pawns to restrict your opponent's freeing pawn moves. If you can keep the enemy cramped long enough, he may eventually be forced to make weakening or cumbersome moves that impair his position. The key to using the bind successfully is to keep the bind tight until you can convert it into a tangible, permanent advantage. Release it too soon and your opponent may be able to equalize or even seize the initiative with a counterattack.

Here are some parameters that you might want to assess in the bot that you are playing that can be adjusted by the bot's owners with a program like Chess Assistant. You should take notes on these parameters to discover more about the bot strengths and weaknesses and then lay your plans of strategy accordingly.

King safety Asymmetry -100 to +100
Avoiding Blocked Pawns 0 to 200
King Safety 0 to 200
King Tropism 0 to 200
Pawn Structure 0 to 200
In Check Extension .00 to 99
One Reply Extension .00 to 99
Recapture Extension .00 to 99
Singular Extension .00 to 99
Mate Threat Extension .00 to 99
Pondering-- on/off (thinking on your time)
Type of Opening Book
Size of Opening Book
Use of Endgame Tablebase ---yes/no
Tablebase cache size in MB 4 to 128

Some other considerations that may be adjusted in a range of 0 to 100%:

Style
Strength of Play
Randomness of Play
Book Depth
Selective Search
Passed Pawns
Contempt for Draw
Material vs. Positional
Attack vs. Defend

Positional
Control of Center
Mobility
King Safety
Pawn Weakness

Finally, don't make the same mistake that Garry Kasparov did in his first game against Deep Blue. He played right into the comp's favorite style of play. Its tactical strength lies in its awesome power of calculating millions of moves a sec, and that is just what it did to calculate all of Garry's choices and then calmly and nonchalantly went pawn hunting at the other side of the board, expending two moves. Deep Blue decided it could mate him one move before he could mate it, so it took a pawn and lost two tempos, how brash to be that self-assertive!

Deep Blue - Kasparov

I hope this article has given you some insight into how bots think and will give you an advantage over them. Good luck and may this be your new score

Carbon 1 - Silicon 0

In the final analysis I must stress the importance of paying attention to the basics of chess. If you have not thoroughly studied and learned them, then now is the time to do so if you intend to be successful in beating the bots. For this I would strongly recommend getting some help from some good books, and chess training CD's.

See How to learn the basics


                                   HOME