Tuesday, April 14, 2009

Pawn Breakthrough

Here's the second recent endgame of mine that I thought was instructive. Don't panic, though...it's not nearly as complicated as that last rook endgame.

I've noticed that I have trouble with breakthroughs in pawn endgames. I think it's part blindness, part the avoidance of calculations, and part unwillingness to muck up my tidy pawn chains.

I entered into a pawn endgame where I, as black, had a pawn plus and the opposition.

However, after White played 49.Ke2-d2!?, Black can't win through opposition alone. In some endgames the attacker keeps re-seizing the opposition, pushing the defending king back and finally grabbing material. However here, after 49... Kf3 50. Kd3 for example, White takes the opposition, and after my 49...Kf4 White could have kept up resistance with 50.Ke2!? and repeating the position. Instead, he let me take the opposition and win with the usual opposition technique: 50. Kd3 Kf3 51. Kd2 Kf2 52. Kd3 Ke1 53. Kc2 Ke2 54. Kc1 Kd3 55. Kb2 Kd2 0-1.

The fact that opposition alone here shouldn't win should have been more obvious to me at the time. The following is a classic position of this nature. Black to move:


However, in my game Black still wins in several lines by the breakthrough ...b4! For example, after 49.Kd2:

49... b4! 50. axb4 (50. cxb4 allows Kxd4) 50... a3 51. Kc2 a2 52. Kb2 Kd3 and after White captures the a-pawn Black gobbles all of White's remaining pawns and wins.

I tried to find a similar pawn endgame in my reference books, and also by ChessBase posistion searches, but couldn't find anything. I think this highlights how complex pawn endgames can be...only 4 pawns versus three, and yet this structure may be unique. ChessBase's search functions can be quirky, however. For example, if I searched my blog database, I could find this game using the position, or using the material count, but not both. Weird.

Tuesday, April 7, 2009

Two Killer Apps for the iPod

Months ago I looked into chess applications for my iPod Touch and didn't find anything that grabbed me. I checked again this last weekend and found two I was quite pleased with. One is free, and one is ten bucks. Which is for you?

Glaurung is the free chess app. You can vary its playing strength on a scale from 0 to 100. You can play against it or analyze with it (best line only shown), and can set up positions. There are several different preset time controls, and you can tweak other options such as permanent brain, play style, and book variety. One advantage that Glaurung has is that it can save games or load them from a .pgn file. I have no idea where this .pgn file is located, or if it syncs to your computer, but it at least means you can save games.

Shredder has an app available for $10. What do you get for your money? First, you can vary the play strength from 850 to 2400 Elo (I can't independently verify the accuracy of its Elo scale, but I like it better than Glaurung's 0-100). It can also be set to automatically adjust its rating to your level as you play games. You can adjust its play style, and can turn on a coach feature. The one main drawback to Shredder is that, at least in the current version, you cannot save your game. However, you can email it.

One final edge that Shredder has is that it has a database of 1000 test positions built into it that you can solve. The points you get for each puzzle depends both on accuracy and speed of solving. I like this feature a lot, although for some of the positions it says "you have solved this puzzle" after getting the first few moves correct, when I would still like to see a few moves further into the solution. this isn't a big deal, though, because you can stop the quiz and select "play this position".

So far my conclusion is: Shredder has the advantage if you're looking for a program to play against. The Elo rating with a sliding scale and automatic adjustment to match your play level is a nice feature, and I really like the puzzle feature. However, if you want to be able to save your games (e.g. if you don't have email access and can't email the game to yourself), or if you're thrifty, Glaurung is really quite nice. Both look attractive, and both appear to be hideously strong (I'm worried about the consequences this technology will have on tournament play). If you're serious about chess, and can already afford an iPhone or iTouch, then I don't think you'll regret purchasing Shredder...but you can download Glaurung first and then decide if you really want or need more.

Monday, April 6, 2009

Instructive Rook Ending

First: a shout out to chunkyrook. His endgame posts helped get my gumption level up to posting some endgame analysis of my own. I also rather liked this example of seizing a file...a fairly simple idea, yet one that was new to me.

Sometimes I get a basic ending that I thought I knew reasonably well , and then after analyzing it I realize how much of endgame play is still mysterious. I had two such instances recently where the analysis was instructive. Here is the first, which should have a dead-drawn R+P vs. R endgame. I was black:

I had just captured on c5. The defending king is able to get in front of the pawn, which should be an easy draw. A rule of thumb in 1-pawn rook endgames is the "Rule of 5": if the number of the rank of the pawn, plus then number of files the defender's king is cut off, is greater than 5 then the side with the pawn wins. There's exceptions but it beats assessing a position by coin toss. With a pawn on the third rank (counting from Black's side) the defending king would have to be cut off by 3 files to be winning according to the Rule of 5. Here, you don't even need to invoke the rule: the king isn't cut off at all and is free to get in front of the pawn, so this should be a cakewalk for White. However, after 76. Kd3 Rc1 77. Ra4 Kb5 78. Rh4 c5 79. Kd2 Rg1 80. Kc2 Rg3 81. Rf4 c4 82. Rh4 Kb4 83. Rh8 (threatens a series of checks from the rear, a theme of the Philidor position we are arriving at) Rg2+ 84. Kc1 Kc3 85.Rh3+ Kb4:

This should be a dead drawn Philidor position. For those unfamiliar with this ending, the Philidor position is one of the standard defensive techniques. All White has to do now is keep his rook on the third rank (e.g. 86.Rf3), which prevents White from getting their king ahead of the pawn. If Black delivers checks from the side, the White king just bounces between c1 and c2. If Black tries to make progress by advancing the pawn, White can play the rook to the 8th rank and deliver endless checks from the rear. If this drawing method isn't familiar to you, it's one of the first rook endgames you'll want to study.

However, my opponent played 86. Rh4?-+, probably thinking this prevented advancement of the pawn because of the pin. However, 86... Kb3 is now a win for White--but only barely! The king can use the pawn as shelter against a check on the 3rd rank. After 87. Rh8, a critical position has been reached:

87... Ra2?= If the pawn were on c3, and White was passively defending with a rook on the first rank, ...Ra2 would be involved in the winning line. This method of playing against passive defense is another basic rook endgame, but that's irrelevant to this position. Short on time, and with a 12-second increment on the clock, I was going with my gut at this point.

There is only one winning move here: 87... Rg1+! and it turns out that after 88. Kd2 c3+ white wins:

For example 89. Kd3 Rd1+ (89...Rg3+ also wins) 90. Ke3 c2 91. Rb8+ Ka4 92. Ra8+ Kb5 93. Rb8+ Ka6 94. Ra8+ Kb7.

I have not seen this winning method in any of my endgame manuals, yet it seems important. If anyone has seen this covered before, can you let me know where?

The game continued 88. Rb8+ Kc3

89. Kb1! Well played. This ending appears similar to one covered in Dvoretsky's Endgame Manual as well as (if memory serves) Soltis' Grandmaster Secrets: Endings. Dvoretsky mentions Philidor's "second defensive method": if the rook can't occupy the cutoff rank, it should place itself behind the pawn. The king goes to the "short side" of the pawn (i.e. the side with less space between the pawn and the edge of the board...in this case the left side). Because the defending king is on the "short side", the attacking king is on the "long side" and if it tries to squeak out the long side it can be subjected to checks from the side.

89...Rh2 90. Rd8 This also draws, but 90. Rc8 is the thematic "attack the pawn from the rear" second defensive method. After 90.Rd8, White needs to be careful. 90... Rh1+ 91. Ka2 Kc2 (if the defending rook were on c8, this move would just lose the pawn to Rxc4)....

92. Rd7?-+

Now we see why the "short side" defence is an important resource. Because White's rook didn't stay behind the pawn, it is able to advance. Now the only way to draw is from side checks, e.g. 92. Rg8! (or 92.Rf8!) 92...c3 93. Rg2+ (93.Rg3 also draws) and if 93...Kc1 94. Kb3.

After 92.Rd7? there are still a few tricks involved in winning the endgame. 92... Rh5 wins, but the endgame tablebases reveal this shorter and sneakier winning line: 92... Rh8! 93. Ka3 c3 94. Rc7 Ra8+ 95. Kb4 Kb2!!

with the point 96. Rxc3 Rb8+ 97. Kc4 Rc8+ wins the rook. This appears to be an important resource. Several of the lines I analyzed in this endgame resulted in this position.

93. Ka3 Rc5?=

This allows checking from the side. It turns out the only winning move is 93... Rh3+! 94. Ka2 (94. Kb4 c3 95. Rc7 Kb2 is the sneaky win mentioned above) 94... Rh8 and wins.

94. Kb4?-+ (Many moves draw, but 94. Rh7 with the idea of checking from the side appears simplest, e.g. 94...c3 95. Rh2+)

94... Rc8! Now rook behind the pawn wins. Black can advance the pawn and use it as a shield from side checks. 95. Rd4 c3 96. Rh4 Kc1 (96... Kb2 is shorter) 97. Rh1+ Kb2 98. Rg1 c2 99. Rg2 Kb1 100. Kb3 c1=Q 0-1

Even the simplest rook endgames are difficult to play perfectly. Familiarity with the basic positions (e.g. Lucena/Philidor) and concepts (checking distance; Rule of 5; playing the king to the short side; defending via repeated checks from rear/side/front) help guide you in the right direction (86.Rf6; 89.Kb1!; 94.Rh7 etc.). However, sometimes there's no substitute for calculation (87.Rg1!).