"...a Chessmaster should be a combination of a beast of prey and a monk."

Saturday, October 11, 2014

Attacking failure

When I say I need to work on my attacking skills this is the kind of game I think of.  Here I am attempting to review the game since I noted in my last post that this seems to be a big weak link in my training.  I used the engine to analyze it.  I feel mixed about this.  A lot of people say you should analyze without the computer but of course the nice thing about the computer is that it is actually much more objective than I am.  One of the things that I mentioned I wanted to be able to do was compare my game thinking with a more objective evaluation.  Perhaps the engine is a crutch here but perhaps it's valuable as well.

This game was played at 90-30 against the Chessmaster character Trudy.  It's a good example of me misjudging an attacking situation as much more crushing than it actually was.  In actuality I have a strong advantage but nothing like a "mate in x".  Also the ending of the game is a good example of something I have seen written about by strong players and that's the difficulty of switching gears particularly when you are on the offensive and need to switch to the defensive.

Trudy - Nathan

Result: 1-0
Site: ?
Date: 2014.10.11
Game analysis
Engine(s): Houdini_4_Pro_CA_x64A
Analysis time: 0:43:33
[...] 1.b3 ♘f6N
1...g6 2.♗b2 ♘f6 3.g3 ♗g7 4.♗g2 d6 5.d4 c5 6.d5 ...1/2-1/2, Adams Michael 2610 - Shirov Alexei 2710 , Hastings 1992 It (cat.14)
1...e5 2.♗b2 ♘c6 3.e3 d6 4.g3 f5 5.♗g2 ♘f6 6.f4 ...1-0, Ponomariov Ruslan 2704 - Pomes Marcet Juan 2371 , Albox 6/ 4/2005 It (open) (active)
1...d5 2.♗b2 ♗g4 3.g3 c6 4.♗g2 ♘d7 5.h3 ♗h5 6.♘f3 ...1/2-1/2, Adams Michael 2630 - Anand Viswanathan 2725 , Las Palmas 1993 It "Gran Canaria" (cat.16)
1...c5 2.♗b2 ♘c6 3.e3 d5 4.♗b5 ♗d7 5.♘f3 ♘f6 6.O-O ...0-1, Morozevich Alexander 2721 - Rashkovsky Nukhim N 2525 , Moscow 1992 Memorial M.Tal
2.♗b2 g6 3.♘f3 ♗g7 4.c4 O-O 5.e3 d6 6.d4 ♘bd7 7.♘c3 e5 8.♗e2 ♖e8 9.O-O e4 10.♘g5?! I was actually prepared for this. Perhaps because it's the only agressive looking reply to f4.
10.♘d2!?10...♕e7 11.h3 a5 12.a3 c6 13.♕c2 b6 14.b4 ♗b7 15.b5 c5 +0.33
10...h6 I'm looking forward to her knight being stuck on h3 already looking to play g5 as Houdini agrees in the variation to move 11 to complete the job. She can of course get the knight out, but it will take some shuffling of her pieces to do it. (10...♘f8 11.♕c2 ♕e7 -0.28) 11.♘xf7? I had looked at this but written off as a blunder and Houdini agrees. (11.♘h3!11...g5 12.♔h1 +0.01) 11...♔xf7 12.♘b5?! A second blunder, here of a tempo. I think the game up to this point gave me a false sense of security that I was well ahead of the computer and from here on out was starting to look for more directly winning lines. (12.f3!?12...♔g8 13.♕c2 -1.46) 12...a6 13.♘c3 c6 I was worried about the results of her playing d5 here and maybe plopping a knight there but decided it didn't really work out for her. 14.♗g4 ♘xg4 15.♕xg4 Here my big debate was between Nf6 with a double attack on the queen, or Qg5 hoping to trade queens since I'm ahead material. I ended up deciding on the Nf6 because I wanted to release the light squared bishop and I wanted to have the Nf6 so I could play d5 and it seems I have reinforced my pawn on e4 and with it a space advantage. 15...♘f6 16.♕e2 d5 17.cxd5 cxd5 18.♖ac1 ♗g4 Here's where I really started thinking in terms of an attack on her king. 19.♕d2 ♕d7 20.♗a3 ♗f3 Here is what they call "count the pieces" or "local superiority". I've got my queen, my bishop, and my knight all able to attack the kingside quickly. She's got a rook defending. It seems like she lacks defenders here, so it seems like an important condition for an attack is given. Of course the pawn that can slide into f3 to support the attack is another big help. I feel there should really be something here and am really trying to find it. 21.gxf3 exf3 22.♔h1 I hadn't considered this move though I had noticed that she had a tempo to spare here. 22...♕h3 23.♖g1 ♘g4 24.♖xg4 ♕xg4 I went over Smyslov-Stahlberg from round 4 of Zurich 53 today and there winning the exchange as the result of an attack on a king was considered sufficient reward and was winning. Maybe I should have been happy with that. Still, considering I was a piece for a pawn up, seems like I lost ground rather than gained it if the attack doesn't pan out. 25.♖g1 ♕h3?? At this point I am thinking I am still good to go and that all I have to do is get a rook onto the h-file and I'm won. Obviously, Houdini doesn't think so. Still, at this point if I play the way it wants me to I'm still 2.5 to the good.
25...♕f5!26.♖g3 g5 27.♗c5 ♖ac8 28.a4 ♗f8 29.h3 ♖e6 30.♕d1 ♖f6 31.b4 -2.48
26.♘xd5!26...g5 27.♘c7 ♖ad8 28.♕c2 ♖c8 29.♕c4 ♔g6 30.♕d3 ♕f5 31.♕xf5 ♔xf5 +0.42
26...♖ad8?? At this point I'm starting to realize that it might be a bit more work to wrap this up than I thought just a move ago, but I still see this as totally won for me, and just a question of getting a rook onto the h-file. I am just assuming that the computer is only left with futile shuffling of the pieces and that I can proceed totally unhindered in my designs. I believe I glanced at pieces the queen attacked but totally neglected to register that the rook on g1 is actually an active offensive pieces as well as defending against mate on g2. Interesting though that if I just play to trade off queens I still have a significant advantage according to Houdini.
26...♕f5!27.♕xf5 gxf5 28.♘xd5 ♖ad8 29.♘f4 ♖c8 30.♗c5 ♗f8 31.♗xf8 ♖xf8 32.h3 -1.59
27.♕xg6 ♔g8 28.♕xg7#
Powered by Aquarium

Tuesday, October 7, 2014

Some notes about training methods

Just been thinking about some things lately and wanted to write a post about them. These are just my ideas and I certainly don't have any authority on the subject. Still, I think it can be valuable for me to write out my thoughts, and it's possible someone else might be interested. In one of my early posts I believe I raised the question of chess skill being more a question of knowledge, or more like the skill of a musical instrument. I definitely believe it to be a mixture at this point. There's plenty of knowledge elements where progress is made by learning ideas, like certain key ideas in king and pawn endgames or variations of opening theory. However, there's a huge component that is more like a skill where practice, much like practicing playing an instrument matters. Jonathan Rowson in his "Chess for Zebras" talks about the comparison with the practiced skill as well.

I agree with the commonly held view that at my level the most value in terms of improvement is to be gotten from studying tactics. This makes a lot of sense to me.A really awesome thing about chess training these days is the wealth of available material for doing tactics training. I have huge stores of problems on my computer and in my collection of chess books. There's also really great sites like chesstempo. There are actually two related things here though. One is more traditional "mate in x" problems, and the other are the gain material type problems. I have heard and it makes sense to me that it's wise to have a balance between these two and that it should favor the tactics. I myself have been doing more mate problems lately because I spent a lot of time focusing on the tactics and maybe not enough time on mates.

I think there are some interesting issues to think about here. One question is the sort of Michael de la Maza question. Can tactics training alone get you to an advanced level (say 2000)? Another question is the relationship between calculation and memory with tactics training.

Regarding the first question, I am obviously not qualified in any sense to make a statement. I can however make a statement about what I want from chess training and how I approach the question. As I said above it makes sense to me that this should be the bulk of training at my level. I think there are some appealing things about tactics training. The tasks are bite sized, with clear condition, so there's a very direct sense of satisfaction in being able to say "I did x number of tactics today." There's also a very direct feedback, so there's no wondering about shades of grey or complex concepts. You either solved the tactic or you didn't. I think these things make tactics training very quantifiable and satisfying. There also seems to be a pretty clear correlation at least from a lot of word of mouth between training this way and playing performance increase. I generally like tactics and mate puzzles, (though I do at times get a bit bored), but I think it's a bit of a mistake to discount other methods of studying the game because they aren't so quantifiable.

The flip side of the coin here is that I like chess because it's very complicated and there are a lot of ideas. I like tactics training but if that's all there was to chess I don't think I would feel as drawn to the game. I obviously don't know anything about de la Maza's life, but it seems interesting that from what I have heard he gave up chess after meeting his goal of getting to 2000. Part of me feels like that kind of makes sense in a way. If all chess is, is getting good at tactics then it makes sense that it wouldn't have that drawing power to make you keep coming back. To get to the point, that's not what I want chess to be for me, so I seek to learn about many facets of the game, from opening theory, to strategy, to endgames, and I also like the history and culture of the game. For me due to these concerns going over historical games is another big part of my training.

The other issue that I raised with tactics is the idea of calculation vs memory. Some people argue that the point of doing tactics training is to cram patterns into your head. Dan Heisman for instance suggests that an important part of tactics training is the familiarity with tactical patterns. They suggest that if you don't solve the problem in a fairly small amount of time (by my standards) you should give up and look at the answer, because learning the pattern is more important than calculating the answer. One argument that make is that strong players unquestionably have a storehouse of such patterns that they are able to call upon as an aid to their calculation. I don't really like these arguments. For me, tactics training is calculation training. The way I optimally solve a problem is to look and see all the wrinkles of the problem before I input the solution or consider myself finished with it. Obviously I am human and there are times when I do more guesswork or when I go off half cocked. I'm talking about my optimal idea that I strive for. My reply to the argument about the strong players having a storehouse of patterns is that it seems possible to me that they built up those storehouses through practicing calculation, so the patterns got ingrained. If that is true, then it's possible that the attempt to cut corners and simply memorize the patterns will backfire because you aren't actually putting in the work that will build that storehouse. There's something about that way of looking at it that just seems a bit confused. Like, it's possible that you are completely right about the strong players having these storehouses of patterns but still wrong about the way that they should be built up. It doesn't follow logically to me that you should just be able to short circuit the process and jump to the conclusion. Anyways, at the personal level I feel that's true for me. Memorizing doesn't work for me unless I have many repetitions or I have a narrative that I can fit the ideas into. For the record, here is a link to a post by a very active member of chesstempo and apparently an FM. Note his first point is essentially in agreement with me.

Other than that, I do a number of things. One big one that I like a lot is going over historical games. I very much enjoy going over commented games. Right now as I mention in a previous post I am going over the games from Zurich 1953 using both Bronstein's and Najdorf's tournament books. I am enjoying it quite a bit and feel I am learning from the commentary. I also intend to start studying some commented Lasker games in the near future. I obviously am not plumbing the depths of these games and there is much I don't understand but I'm ok with that. The other thing is uncommented games. I haven't done as much of this but the Anderssen games I have posted here are what I mean and I would like to do more. The idea is to find uncommented games and analyze them. I have found the ones I did quite enjoyable and I think it's quite valuable as well. For this, I think it's important to pick games that are comprehensible to you. I've mentioned this before, but for me to jump in and try to analyze the games from the Kasparov Karpov matches just isn't going to bear much fruit because it's just too far beyond my understanding level. I don't mean seeing all the details here. I just mean even being able to understand the themes of the games. That's why I started at the beginning of chess history. I have looked at the Greco analysis/game records before and a little bit a the La Bourdonais McDonnel matches. Adolf Anderssen seems to be a good place for me now. I also don't expect to be able to understand all of his games. And when I look at a game I don't expect or claim to understand all of what I am looking at. However, I think the process of looking at different games and picking ones that seem comprehensible and then practicing my analysis skill on them are valuable.

Going along with the historical games idea, I have heard it can be a good idea to have a chess hero, who you emulate. I'm less sure about this one and it may be too artificial to pick one for purpose of picking a chess hero, but I have decided to give it a try and picked Emanuel Lasker. He's a world champion and he's said to have a universal style. This appeals to me because I generally agree with the people who say improvers like me shouldn't try to have a style, rather we should seek to improve our chess and eliminate weaknesses. I think that makes a lot of sense. So that's why if I am going to have a chess hero it would be one who was know for being well rounded. Also, we both studied philosophy, and I also haven't heard really anything bad about Lasker as a person.

Openings is another area that is often talked about. I agree with the common consensus that for my level it's not very important to study openings. However, I don't think that means I shouldn't study them at all. I think it makes a lot of sense to look up the openings that occur in your games and try to understand where you were out of your own book and what that means in this particular case. I do study lines that I haven't seen but to be honest my memory isn't that good for rote memorization. If I have played a variation or if there's an idea I can latch onto, then I am much more likely to remember the variations, and that's why learning through my own games seems like a good thing. That's part of why I like to save my games and review them in the tree structure in chess assistant. I was just today looking at some variations in the Ruy Lopez and I learned a fair bit by reviewing the games I had played as well as reading from the resources. As far as openings go, I would definitely recommend Paul van der Sterren's "Fundamental Chess Openings". This book is really amazing and is strong contender for best or at least most useful chess book I have so far bought. Very clear explanations of the ideas of the moves often relating to the history of the variations. I won't say it's the only opening resource you need at a low level but it's quite valuable.

I've also been doing more work on chess endings lately. Particularly solving some simple studies. These have mostly been of the King + Pawns variety. As I mentioned recently Laszlo Polgar's "Chess Endgames" is a good source. But obviously this has involved some other things like the bishop and knight checkmates that I posted about recently. I've also tried to learn some basic rook endgame concepts. I think though this is another area where doing has to work with learning. I definitely appreciate descriptions of some key concepts but I need to practice these things to really get better.

Another thing I do is keeping a chess journal as well as this blog. It's pretty basic but I write down the stuff I do every day relating to chess. I find this helps me to both have a sense of accomplishment and to have an accurate idea of how much I am actually doing. I also think the blog has been helpful in spurring me to do some things and to be a bit involved in the chess improvement community.

There's also some things I would like to do more of. Primarily this is analyzing my own games. I believe the arguments of people who say this is a crucial method for improvement. I guess it seems like when I analyze my games I write down a lot of what was going on in my head but I'm not so good at actually looking at the variations. I think that's a mistake. I think the point is to be able to compare what was going on in your mind with some more objective conclusions about what was actual. I just find it easier to pry into the variations with an Anderssen game. I feel when I review my own games I often just can't see beyond my own ideas. One option here is just to let some time go by and look at the game then. This can definitely help, but sometimes I feel too far removed then, like the moves just seem dumb. Anyway, Chessadmin's blog "Path's to Chess Mastery" is kind of my guiding light here. It seems to me he does a pretty good job looking at his games and comparing the things in his mind with more objective conclusions.

Well, that was a very long post, but I do feel it's helped me to crystallize some things. Here's hoping someone else can get something from my musings.

Friday, October 3, 2014


Tomasz asked me to post an example so he could critique it.  The example html won't fit in the comment so I will post it here.

Nathan (1600) - WeakDelfi (1200)

Result: 1-0
Site: Local computer
Date: 2014.10.01
[...] 1.♔b2 ♔f5 2.♔c3 ♔e4 3.♔c4 ♔e5 4.♔d3 ♔d5 5.♗e3 ♔e5 6.♘g3 ♔d5 7.♗d4 ♔c6 8.♔c4 ♔d6 9.♘f1 ♔e6 10.♘e3 ♔d6 11.♔b5 ♔c7 12.♔c5 ♔b7 13.♔b5 ♔c7 14.♘f5 ♔d7 15.♔c5 ♔c8 16.♔c6 ♔b8 17.♘d6 ♔a8 18.♘b5 ♔b8 19.♘c7 ♔c8 20.♗a7 ♔d8 21.♘d5 ♔e8 22.♔d6 ♔f7 23.♘e7 ♔g7 24.♗e3 ♔h7 25.♗g5 ♔g7 26.♔e6 ♔f8 27.♘c6 ♔g8 28.♘e5 ♔f8 29.♔d7 ♔g8 30.♔e7 ♔g7 31.♔e8 ♔g8 32.♗h6 ♔h8 33.♔f7 ♔h7 34.♗f8 ♔h8 35.♘g4 ♔h7 36.♘f6 ♔h8 37.♗g7#
Powered by Aquarium

Wednesday, October 1, 2014

K+B+N vs lone K

Today I successfully checkmated the black king with just a bishop and knight three times in a row against the computer without consulting any tutorials during that time.  It's not easy to do but it doesn't really take so long either.  For me, what I did was I read about it in several books.  (I often find that reading the same material presented in different books really helps me to learn something).  After that I tried and failed a bunch.  Then I consulted the books and a video again.  After that I started trying and failing again.  Then I started to notice some particular situations kept cropping up, and I consulted the books again and learned what to do so I could progress to the point where all that is left is like a regular tactic.

There are really some pretty crucial things.  The first is just that you will ultimately have to get the opposing King to a corner that's the same color as the bishop. 

Pandolfini's workup on this is pretty good because it breaks it into chunks so first he teaches you how to checkmate with the bishop and knight when the king is already trapped in the correct corner.  Then he tries to teach you how to drive the king from the wrong corner to the right corner.

So basically when I was trying it, I would set up the white king, knight, and bishop spaced out on the first rank.  One thing I did that makes it a bit more complicated but makes sure that you have the technique rather than just some specific moves is to switch up which color bishop you are using.  Then the black king is on some random square in the middle of the board.

It's quite tough to learn how to drive the king into a corner but after doing it a bunch that part starts to get worked out.  Once you get the enemy king into the corner, you have to force him along the side of the board to the correct corner.  This is tough but especially for the first part if you try it a bunch you will see some things keep happening the same every time and it's easy to start building up the sequence that you will need.

Another tough part is that when you get about halfway along the edge of the board, there are some slightly counter-intuitive moves that let the enemy king come off the edge a little bit, but still keep him trapped.  These are the ones I had to learn by rote (although I can perform them going either way).  Once you've got that down then it's pretty obvious after a few tries how to make the box smaller and eventually give the mate.

Apparently a lot of people question whether this is worth knowing.  For example, Silman in his "Complete Endgame Course" skips it because he says it's not useful as these positions are very rare.  There's a couple of reasons that I wanted to learn this.  Part of me just feels like it's kind of a mark of distinction that it's not an easy thing to learn, and I feel like it sets me apart just a little bit to have learned it.  Secondly, I do believe Jesse Kraai's argument, that learning this isn't so much for the fact of using it in a tournament game and being able to score that extra 1/2 point on the quite rare situation it might come up.  The point is it helps you learn to use the pieces together in a simple setting, and that's something that translates to other situations for sure.

On the other hand, I don't feel that I have mastered it.  In that I feel I kind of have it memorized right now, but that I could lose it.  One possibility for me would be to continue practicing it until it becomes sort of second nature to be able to do it right.

This probably won't translate into an apparent strength gain but I am happy to have accomplished it anyway.