Category Archives: Endgames

A Reminder About The Strengths And Limitations Of Chess Engines

Last week, I wrote an article about methodically building an endgame fortress, given a fascinating position that arose in a student’s game. One thing I did not focus on in the discussion was the fact that he has inquired about confusing evaluations by chess engines during endgames. Often, in positions that we know a correct and foolproof drawing technique, a chess engine will show a large evaluation score incorrectly suggesting that the superior side has such an advantage that it should win. He knows about tablebases, which are exactly computed evaluations of simplified endgame positions in which exhaustive search has proved the result is a win or draw, along with the full variations leading to the end result. So he asked me why the computer can be unreliable.

The reason for this has been known for decades, but is worth periodically remembering given how this period of human history has shown a huge explosion in the successful use of computation to make progress in various domains, including chess, Web search, and shopping prediction. This success has led many to believe (either with excitement or worry) that computers are on the verge of becoming super-intelligent. I don’t believe this is the case at all (but this general topic is outside the scope of this site).

Here I just want to point out something very concrete: that endgame fortresses are still difficult for computer engines to “understand”. Coincidentally, GM Daniel Naroditsky recently wrote an article about breaking fortresses. I highly recommend studying it, and experimenting with “throwing an engine” at the positions he gives. You will get an idea of how a computer engine “thinks” as it tries to search forward for some kind of forcing variation.

The main difficulty for chess engines is that they are largely programmed to search forward, whereas we human beings in the domain of chess are able to use meta-reasoning rather than just huge, long chains of “if/then/else” in order to generate ideas, plans, and test them out by working backward (instead of forward). Breaking a fortress involves taking in the biggest possible picture of what is happening on the board, eliminating what must not possibly work (usually, this means shuffling pieces around is not enough), and trying to see if something might work (usually, this means some kind of temporary “weakening” through Pawn advances in order to eventually create a Pawn break). We humans can with care determine what the “critical” positions must look like, in which everything has to be accurate else counterplay is achieved, and then zoom in to see if we can create a forcing line (usually based on some kind of Zugzwang or switch of attention from one side of the board to the other) that is tactically justified.

Chess engines are tremendously useful for this kind of work, actually, despite what I’ve said: on their own they may not be able to crack fortresses, but they can supply the assistance to a human who is “directing” the problem solving on the meta level and assigning specific calculations and verifications to the chess engine. The usefulness of computers as assistants is very exciting and real, but should never be confused with claims that they will achieve autonomous “intelligence”.

Franklin Chen

Share

Methodically Building An Endgame Fortress

A student showed me a fascinating game of his in which he was fighting for a draw as White, being an exchange down (Rook down for a Knight) for a Pawn. The position looked precarious, but the more I looked at it, the more it looked like he missed a fortress draw (he blundered quickly instead). Upon analysis, the fortress idea appears to work, but just barely. Below I explore the construction of the fortress and a subtlety that shows how a single inaccuracy could cause White a lot of trouble.

Features of the position

The starting position has unusual features that give White a fighting chance to draw at all:

  • White has a Queen side Pawn majority and a King side Pawn majority. This helps prevent Pawn breaks by Black, although Black may be able to try a minority attack on the King side.
  • Black’s b6 and e6 Pawns are extremely weak. If White could win one of them, that would ensure a lot of counterplay, probably good enough for a draw.
  • White’s Knight on d4 is a monster. Most critically, it prevents any Black King invasion via c6, b5, or f5, so Black can any possibility of winning only by using the Rooks and King side Pawns.
  • There is only one open file for any of the Rooks, the a-file. If White takes it, White should probably be able to draw by perpetual check and/or winning the b6 or e6 Pawn (especially the e6 Pawn, in which case White would have a passed e-Pawn ready to march to e6 and e7).
  • One of Black’s Rooks happens to be very poorly placed. It will take time for this Rook to get to the a-file and join up with the other Rook to try to advantageously trade one Rook and then aim to knock off any weak White Pawns that cannot be protected by White’s King or Knight.

Ideas of White’s fortress

Making a list of the features of the positions gives many clues about how White could possibly draw this position, as well as how Black can try to win it. Of course, general considerations are not enough: very careful tactical calculation is required especially when White has the opportunity to go all out to abandon everything and try to get to Black’s seventh rank with a Rook: if the attempt at a perpetual check (or other draw by repetition) and/or Queen promotion fails, White will obviously lose. In this article I don’t focus on the variations in which Black allows such penetration, but on the fortress itself, under the assumption that Black does not allow the penetration.

The first thing to do is to imagine that Black does trade off White’s remaining Rook. Black can always force a Rook trade if desired, so we have to at least be able to hold the draw if White’s Rook can no longer defend the whole range of White’s position, from Queen side to King side.

  • Black’s King cannot make progress as long as White’s Knight stays close to d4 and attacks the e6 Pawn.
  • If Black sacrifices the Rook for White’s Knight, that should not achieve anything because Black’s King is not close enough to do anything useful in the King and Pawn ending.
  • The c-Pawn must remain protected: this requires either the King on the b, c, or d files or the Knight on e2.
  • The e-Pawn must remain protected: this requires either Ne2 blocking a Rook on e1, or f4 creating a Pawn chain.
  • The f-Pawn must remain protected: if the g-Pawn has been forced to advance to g3, then f4 creates a Pawn chain; if the g-Pawn has been forced to advanced to g4, the f-Pawn is best protected at f3 by the Knight on d4.
  • The g-Pawn must remain protected: it has to go to g3 or g4, because otherwise it is too far away from White’s King and Knight, which ideally remain no further than the e-file, in order to guard against possible loss of the c-Pawn or possible invastion by Black’s King.
  • The h-Pawn must remain protected: at h3 it is in big trouble because we assume the g-Pawn has to be advanced; at h4 it might be OK, protected by a Pawn at g3; at h5 it might be OK, protected by a Pawn at g4.

How might Black breach the fortress?

The main thing to notice is that if Black can get a Pawn down to h3 safely, without trading any Rooks, White is surely lost, because Black can first tie up White’s pieces on the Queen side, then trade a Rook just in time to get the other Rook attacking White’s defenseless Pawn on h2. Therefore, Black has the plan of g5, h5, h4, h3.

Also, if Black can force a Pawn trade of the g-Pawn and open a file on the King side (say by White being able to play f4 only after Black has already played g5), White is surely lost, because of the power of a Rook crashing through White’s position through that file and winning one or more remaining weak White King side Pawns with the help of the other Rook.

So the main variation below, which succeeds in setting up a defensive fortress, has White hurrying up to distract Black’s Rook away from the King side to defend the a-file, then playing h4 to permanently prevent the h3 plan. Note that it involves saving time by not defending the attacked h2-Pawn at all.

An interesting side variation, which may lose, involves White playing g3 to protect the h2-Pawn currently under threat, but permanently weakening the h-Pawn. Black can try the g5, h5, h4, h3 plan. If White just waits passively, the game is lost. There is a fiendishly complicated variation in which White abandons the fortress idea and tries to get counterplay at the cost of sacrificing the f-Pawn after redeploying the Knight to d6. This is scary-looking and I don’t actually know if White can draw with computer-perfect play, but it is White’s best try after starting the mistaken g3 idea.

Annotated

Franklin Chen

Share

We Had an Even Finnish to the Chess Game

My opponent in this chess game is from Finland.

The first 11 moves of this chess game were pretty much what I expected. I was surprised by White’s 12th move, but it was in my database of chess games.  Once again, I was a little surprised by White’s 15th move. V. Golod has commented that this position gives Black an isolated d pawn, but it is otherwise equal. Blacks’ 16th move was based upon the results of games in my database, the evaluations of various chess engines and upon Golod’s comments. The Golod analysis is included in my subscription the ChessBase Magazine.

Up until move number 18 we are still in my database of chess games. White’s move number 19 is a novelty. The more usual move here is 19.Bd3. After that I was on my own. With a Knight versus a Bishop I thought that it was best to keep as much of my material as I could on dark squares so that this material could not be attacked by the White Bishop.

When I was given the chance to grab the White Bishop on move number 21 I took it. After that, there was some maneuvering of pieces and pawns with neither side getting any advantage. Because the material was even and our ratings were close I offered a draw and he accepted. At the time that I am writing this, this is the only game in this section that is completed.

Mike Serovey

Share

The Danger Of A King Out Of Play In The Endgame

In a hard-fought game my student played that ended in a draw, when we were looking at it, I observed that his opponent missed a win at one single critical moment. This was a result of an accumulation of positionally questionable decisions that, although in themselves still led to defensible positions, led to a single blunder that could have been punished.

Three mistakes

Allowing an outside passed Pawn

The first unnecessary concession was made in the late middlegame when Black captured a piece on a5 allowing a recapture with a Pawn bxa5 resulting in White getting an outside passed Pawn. Granted, this being a Rook Pawn made it not as useful, but still created unnecessary danger.

King out of play

The second unnecessary concession was moving the King from g8 to h7, out of the main action. It was best to moving the King toward the center and toward the Queenside, with the goals of safeguarding the Pawn chain from c6 as well as, more critically, aiming toward White’s a-Pawn, either to capture it or at least prevent it from Queening. Granted, Black had a plan to get the King to f4, but it is slow. In fact, it ended up working in the game, but only because White did not act more quickly and decisively to try to Queen the a-Pawn.

Creating another outside Pawn for the opponent

The final concession, which in this case was a big blunder, was to accept White’s sneaky offer of a Queen trade, resulting in transforming White’s c-Pawn into an “outside” b-Pawn that could have been used as a Pawn break to lead the way for White’s King to invade the Queen side and successfully Queen the a-Pawn. A calculation shows that Black’s attempt to also Queen a passed Pawn is too late, because White’s active King can get to Black’s King side Pawns in time to ensure that after White gives up the Rook in turn, the resulting King and Pawn ending is an easy win because Black’s King ends up out of play and White can just push a passed Pawn to victory.

Lessons

The main lessons to learn are that even in a drawable position, it is wise to keep the draw simple by not giving a passed Pawn to the opponent, not giving a Pawn break to the opponent, and keeping one’s King ready to prevent Queening of a passed Pawn if it does exist.

Franklin Chen

Share

Another Comedy of Errors

This is a game that I played back in July of 1990. This is one of four chess games that I played against Rick Christopher back then. I won three of those games and lost one of them. This game is one of my wins.

Rick was a player that I didn’t take seriously because I was rated much higher than he was and because he never wore shoes to any chess tournaments that I can remember, not even in the winter! In this game I got a little lazy and did not see some of my opportunities to win more quickly and Rick (White) missed some opportunities to equalize. I basically waited for Rick to blunder and then won the endgame after he did blunder. This strategy does work against weaker players, but it is better for my game play overall to force errors.

Mike Serovey

Share

Anticipating The Endgame As Part Of Understanding The Opening

The 2014 World Chess Championship rematch between Carlsen and Anand kicked off with Carlsen playing the Grünfeld as Black, an interesting choice since he does not usually play this opening, and in fact Anand is the one who prepared the Grünfeld as Black in 2013. The game proceeded along a path in which Anand as White lost an opening initiative and got into some trouble but held an unpleasant endgame.

Since detailed commentary from many strong players is already available and will continue to be provided as the match progresses, so why should I write out it here at The Chess Improver? My goal here is to describe the big picture that players of many levels can relate to and hopefully apply to their own play.

The goal of the Grünfeld Defense opening

Black’s goal in playing the Grünfeld Defense is to try to destroy White’s center, by targeting White’s Pawn on d4. The asymmetrical Pawn structure that arises when White’s c-Pawn is exchanged with Black’s d-Pawn gives Black possible chances to contain White’s d-Pawn and counterattack with a Queen side Pawn majority.

White has a choice of goals in return, and has to make a decision. (Take note if you are following the match, because we may see the Grünfeld pop up again with players making different decisions.) The three basic choices are to:

  • Grab the big center with e4, advance with d5 eventually, possibly make a passed d-Pawn for the endgame.
  • Forget the endgame, go all out with an attack on Black’s King based on h4, h5, etc.
  • Forget the big center, protect the d4 Pawn with e3, block in Black’s Bishop on g7, and try to make headway on the Queenside.

What happened in this game

What actually happened was Anand played as though aiming for one of the first two, but was inconsistent in followup. He got the center and then played as though to attack Black’s King: Qd2, allowing his Knight on f3 to be captured by Black’s Bishop permanently messing up White’s Pawn structure (doubled f-Pawns, isolated h-Pawn), castling Queen side. But he never did attack Black’s King after all, and the Pawn on d5 didn’t get any further.

So Black’s defense, based on destroying White’s Pawn structure and surviving any attack, with the aim of reaching a superior endgame, worked out. Anand had to be careful to hold the draw in face of his isolated and weak f and h Pawns.

The main thing I want to point out is that it was not automatically bad for White to allow the weakened Pawn structure. Before the endgame, there is the middlegame. It is a valid, aggressive idea for White to decide not to try to win the endgame, but instead the middlegame. It just didn’t work out in this particular game.

Franklin Chen

Share

Amateur Versus Master: Game Twelve

My opponent is this game is a Senior Master and is the only 2400 rated player that I have faced in Over the Board chess. Gary has won the State of Florida Chess Championship at least once and has also run the state championship as the senior tournament director for that event at least once. The state championship several years ago was the last time that I saw Gary in person. Gary is a year or two older than I am and he also has some chronic health problems. Gary has managed to keep is USCF chess rating over 2400 points for about 40 years now.

I learned the Botvinnik System from a USCF Life Master who did not know what it was called at that time. He advised against playing this system as Black, but I often get away with it and Botvinnik himself played it as Black. In this game I missed a shot at an upset victory on move number 12. Gary most likely would have found the correct line of play, but it may have rattled him anyway.

I walked into a Knight fork on move number 13 and lost the exchange of a Rook for Knight. Things went downhill for from there.

Mike Serovey

Share

Playing Better Rook Endgames

You can learn technical rook endgames using any good endgame book, but what I am going to share is based on my experience and little reading. If you deploy following points while playing rook endgame it will definitely help you.

An active rook is your hero: Rooks love to attack in the endgame. Here are two simple examples which will help you to understand what I mean by that.:

With these examples I am not claiming that an active rook will always secure you a win or draw a pawn down, but it will definitely provide you better chances to win or defend in worse conditions.

In order to keep your rook active you should know the Tarrasch Rule which is to place rooks behind the passed pawn, whether it is your passed pawn or your opponent’s.

Cutting off the enemy king: This can happen a lot in practice and often decides games (a very useful tool to obtain lucena position!)

In this position Rd1!! is a forced win for White and no other move will do. Here the Black king is cut off by a file, and if you want to check how effective a rook is when it cuts off enemy king along a rank, please study the Philidor position.

Rook works well when weaknesses are fixed rather than mobile, something I have learned by studying Capablanca’s rook endgames. And in order to target those weaknesses you must have an entry point into the enemy camp. You can do that by working hard!!

Ashvin Chauhan

Share

The Common Problem Of Following A Pattern Without Understanding It

Last week, I wrote about the importance of learning and teaching through comparing similar but different situations. Again and again this theme pops up, and is easy to miss if one is not careful. It is easy to memorize a pattern without understanding its context and purpose, or more charitably, to have understood it once but getting it mixed up with another pattern during the heat of battle. What is the solution? Sometimes the solution is just to review concrete details. Sometimes the solution is to remember a higher-priority pattern that gives real force and justification to the pattern at hand.

Here’s an example I recently saw, involving the elementary Lucena position which is a win for the side with the Rook and Pawn versus Rook, if one understands the fundamental concept, which is “building a bridge” in order to block the opposing Rook’s checks and therefore ensure Pawn promotion.

Lucena position

The standard easy win for White is to

  1. Chase Black’s King further away from the Queening square by checking.
  2. Lift the Rook to the 4th rank in preparation to “build a bridge”.

However, White in eagerness to “remember” the key pattern, that of the Rook lift, failed to perform the first critical step, and the result was a draw by mistake! Building the bridge is pointless if it only results in Black’s King reaching the advanced Pawn and gobbling it up.

The solution to this mistake is to remember that the primary goal in this position is not to build the bridge. The real goal is to successfully Queen the Pawn, and getting Black’s King far away is the most important part of that, not the bridge building. The bridge building is not the goal, but the means to the larger goal. Without remember this, it is too easy to just vaguely remember one aspect of what the winning technique is, and use it outside of the larger context.

Franklin Chen

Share

San Pedro Escapes the Four Knights of the Apocalypse!

According to Wikipedia, the free encyclopedia, the original meaning of apocalypse is an uncovering, translated literally from Greek as a disclosure of knowledge, i.e., a lifting of the veil or revelation, although this sense did not enter English until the 14th century. In religious contexts it is usually a disclosure of something hidden. Christians changed the meaning to ‘end of the world’ because the Apocalypse of John is about the end of the world.

In this case, what was revealed is that my opponent does not know how to play the Four Knights variation of the Sicilian Defense and is weak in middle games. However, he avoided the blunders that would have allowed me to win this game. I settled for a draw against an inexperienced player while I was up two pawns. On move number 34 I was inspired to look at an idea, but I got impatient and I rejected it before I realized that it actually wins. I was preparing to move out of my apartment over Labor Day Weekend and I wanted to end this game before I moved out and took a time out from my remaining games. If I had been more patient I would have found the winning ideas. Mr. Generoso was generous in giving me those two pawns and he may have thought that it was the end of the world while he was struggling to draw down material. ;-)

I took that lazy man’s shortcut and played the way I had played in two previous chess games. The first time that I had an endgame with my Rook on the queening square and my opponent’s Rook behind my passed pawn was at the State of Florida Chess Championship of 1986. If I remember correctly, my opponent was a 1200 rated player. He blundered by moving his King to the third rank and that allowed me to move my rook off the queening square with check and then queen the passed pawn. The second time I had this kind of endgame I played more than 60 moves before I realized that I could not force a win and that my opponent was not going to blunder. After this game I am going to endeavor to avoid having my Rook in front of a passed pawn again!

This game was my second draw and Pedro’s only draw so far. What is even more embarrassing for me is that Pedro has three losses so far in this section. At the time that I am writing this I have four draws and no wins or losses in this section. I need to win at least one of my two remaining games in order to get second place in this section.

Mike Serovey

Share