Tactical insanity: blitzin' on subpatterns

I am doing 150-300 tactical problems a day. As I've said before, it is intense. Aside from the sheer mental and temporal load it requires, doing hundreds of problems a day is a very different experience than doing 10 or 20 a day. There are a couple major differences I've noticed.

First, I am now pretty much in blitz mode during the tactical training, and this has started to adversely affect my thought process in real games. I am playing impulsively and making stupid mistakes. Quickly solving 300 problems creates an illusion of tactical prowess on my part. I need to consciously remind myself to switch to a lower gear when I get back to standard time controls.

Second, very similar sub-patterns now emerge in the same session, but these sub-patterns contain important differences that require some caution. Since I am now working in blitz mode, it is always tempting to just make the first move that pops into my head, but it is important to check that all the key elements of the memorized pattern are there, not just some of the key elements. For instance, consider the following two positions from Chess Tactics for Beginners (CTB):
Each position contains the same motif along the g-file, but they exploit this sub-pattern with very different consequences. Even at the second stage (out of five) of exercises, CTB has many such problem-clusters, often three or four problems with the same subpatterns exploited in different ways. This conscious recognition of subpatterns embedded in quite different contexts is something that I hadn't really experienced when doing only 10-20 problems a day. I guess if I were solving the problems by tactical motif, this would have been more obvious.

	# Circles	Percent Correct
Problem Set 1	14	98-99-100-100-100-100-100 100-100-100-99-100-99-98
Problem Set 2	14	90-93-96-99-99-99-99 99-99-99-99-99-99-99
Problem Set 3	0
Problem Set 4	0
Problem Set 5	0

NOTE: Circles done with CTB.

The endgame is hard

I no longer lose all my games because of tactical blunders in the middlegame. In fact, I have started to fairly consistently out-tactic my similarly-rated opponents. On the other hand, because my opponents are getting better, my material advantage (and mating prowess) is usually not enough to mate before the endgame, so we wind up in the damned endgame. The endgame is my new Official Weakest Area. I typically proceed to blow it, often losing theoretically won games.

Because of this I've starting to study some endgame material, starting with Pandolfini's Endgame Course. This book is typical Pandolfini: he doesn't provide a coherent, clearly articulated, conceptual framework for the endgame or any aspect thereof. He is generally a lazy writer. For instance, he doesn't name key positions like the Lucena (which is in there I think), and what he calls the Philidor position is not what other people call it.

But like his other lazy, unsystematic books, in practice it seems pretty helpful. It consists simply of two hundred thirty-seven positions, one per page, that he thinks are important for beginners to know. His treatment of the BNK vs K mate is quite nice. He builds up to the key position (Position 24) after providing eight useful intermediate positions. It took me about an hour to master this mate using his book, as opposed to about ten hours the first time I learned it using a slightly different pattern.

I am presently working on QK vs RK mates, and have spent about a half hour on the following position which he calls the Philidor Position (white to move and win):

For this position, we get the following from Pandolfini: 1. Qe4+ Kg8 2. Qa8+ Kh7 3. Qe8 (1-0). The evaluation is because we are now back at the original position, but with black to move, and there is nothing good for him here. Pandolfini examines a couple of possibilities: If 3....Kh6, then Qf8; if 3...Rg8 4. Qh5 mate; if 3...Rg1, then 4. Qe4+ Kg8 5. Qa8+ Kh7 6. Qa7+. And then Pandolfini says "Whatever else black tries leads to a similarly dismal result." While this may be true, I am finding even this apparently trivial problem challenging. After a half hour, I still don't feel I have fully mastered it, much less the more general problem of QK vs RK, which apparently is quite difficult in theory.

The endgame is all about slowing down and thinking through concrete variations, and that's what Pandolfini's book is aimed at. The book will be a good complement to a more wordy, explanation-heavy book that provides general principles in addition to concrete examples.

One bit of advice for other patzers embarking on endgame competence: enter the positions in Fritz and play them out there. Fritz has found many errors in my thinking that I don't find on my board: I'm just not good enough at the endgame to trust my intuitions. Also, this lets you slightly tweak the positions to determine how different the board can look for the general solution to still work (e.g., in the above position what if the king is not at f6? What if black doesn't play one of Pandolfini's moves: he acts like the solution is trivial, but you might be surprised: e.g., what if black plays 3...Ra7 or Rc7?). Also, it helps to practice the same position at different locations on the board, to try to see the pattern regardless of where it is on the board.

Note the above position received a nice discussion from Scirius here.

I'm hiring a chess coach

Hello, my name is BDK, and I'm an blunderholic. Can you direct me to the closest BA meetings? I need a sponsor to help me.

I noticed at ICC that one of the Carolina Cobras, International Master Jon Schroer ('schroer' at ICC), offers lessons, and lives in a nearby town. He won the North Carolina championship five times between 1995 and 2002, and is one of the highest-rated players in the state. Perhaps more importantly, he has lots of experience teaching chess at the scholastic level, so probably has worked on explaining things so that chess numbskulls like me can understand them.

Added 12/21/06: One of the main reasons I want to try this out is that I've noticed amongst the chess bloggers that it's the people with some kind of chess mentor or mentors in their life that seem to have improved the most. That is, people they trust, people who are willing to go over games, answer questions, give them advice on how to improve, etc.. In particular J'adoube and Patrick stick out as mentioning helpful discussions with people who are better than they are, and they are two bloggers who have improved very quickly.

I plan to start working with Schroer in January. I'll let you know how it goes.

Update 12/22/06: Our first meeting is scheduled for Thursday, January 4!

Chernev's Logical Chess is sweet

I've had Chernev's Logical Chess for a long time now, but have finally started to work through it. For those who don't know, he works through 30 games, giving annotations on every move, aiming to describe why the move was played. For bad moves, he explains why the move was bad.

Man, what a great read. It is nice to see all these concepts I've been learning applied in real games. I am also learning some new ideas. If you haven't read it, and are a patzer like me, do it up! Like Pandolfini's Russian Chess, there are ample diagrams so it is pretty easy to follow without having a board handy. Also, it is fun that he peppers many moves with quotes from the greats.

I have read some criticisms of the book. For instance, I have read that Chernev is a slave to the outdated "rules" of chess (e.g., open with a center pawn). However, he frequently emphasizes that different principles conflict and that you need to carefully think out what to do based on the concrete position in front of you. For instance, page 20: "Chess is not a game to be played mechanically. Usually, moving a piece twice in the opening is a waste of time, but threats must be parried before continuing development." Also, some criticize him because he was never a Grandmaster. This kind of shallow potshot is just lame, as it says nothing about his ability to teach chess and offers nothing substantive in its criticism. Give me a clear-writing and thinking competent 1800 player to a vague English-impaired slogan-throwing dogmatic Russian GM any day of the year.

This is starting to get intense

I am up to 200 problems a day in the second stage of Chess Tactics for Beginners. I am solving about half of them instantly (i.e., within two seconds), and most of the rest with a little thought (less than 20 seconds). This is pretty intense, as it takes about an hour to finish. After tomorrow, I'll be doing all 300 problems in this stage every day until I can solve every problem easily.

I can't imagine what it will be like when I do the final circle: 1200+ problems in one day...

	# Circles	Percent Correct
Problem Set 1	14	98-99-100-100-100-100-100 100-100-100-99-100-99-98
Problem Set 2	9	90-93-96-99-99-99-99-99-99
Problem Set 3	0
Problem Set 4	0
Problem Set 5	0

NOTE: Circles done with CTB.

Thought Process Trilogy Volume III: Yoda Returns

I asked Heisman about his thoughts on the topic of thought processes in chess. He has written about it, so I knew he'd have an opinion. This will be the final post in my Thought Process Trilogy (see my previous two posts for the first two Tolkeinesque volumes).

Blue Devil Knight: Do GMs consciously go through an explicit step-by-step thought process, or is it something more appropriate for newer players?

Heisman: No GM goes through an explicit thought process. Imagine using an explicit thought process to help you ride a bike: "OK, left foot push forward, and now right, etc." To stop and think about how to do it makes you do it worse. You want to develop good thinking habits for use during a game, which means you ultimately will not consciously think about thinking.

On the other hand, you can't jump from using a bad thought process to a good thought process without working on it, and you can't work on it without thinking about it. This is where the explicit, step-by-step thought processes that people write out come in handy. A good time to work on your thought process is at home in exercises where you can practice getting it right so you don't have to think about it as much in games.

Once you have truly learned a good thought process you don't sit and consciously think 'Time to look for all the checks, captures, and threats.' You simply look for the checks captures and threats automatically, without thinking.

I was glad he used the bicycle example, as it reinforces the 'training wheel' analogy in the previous post.

Note I asked him the above question during his webcast of 'The Renaissance Man' this week, so the above is a rough transcription from the notes I took during his answer. To hear his longer unedited response, there is probably a replay of his show sometime this weekend (note I actually asked him two questions about thinking processes).

Thought process: chess training wheels

Thinking more about the back-and-forth on thought processes in chess (as exemplified in my previous post and in J'adoube's recent post here), I think I've come to a better understanding of what a thought process will add to a beginner's game (note I am not talking about GMs, but to people who believe there is something missing from their game other than just pattern recognition).

There are certain things you must do if you are going to play well against players at your level in slow time-control games. Most relevant are:
1) Consider alternatives. You need to have the restraint to consider multiple candidate moves, not just the first one that pops out at you.
2) Analyze. That is, when considering candidates, you have to be able to think through continuations to see which candidate leads to the best position.
3) Evaluate (implied from 2): you need to be able to determine who is doing better (otherwise you will pick stupid candidate moves because you have evaluated the resulting lines incorrectly).

If you aren't doing 1-3 on some moves, you are either playing blitz-style chess, or are in positions you have played before so are operating on pattern recognition. In those common middlegame positions where we don't see any obvious single patterns and need to actually think, the above three things need to be done, and beginners should be told this.

A chess thought process, by which I now mean those explicit lists of steps to go through when finding a move, is an extremely helpful device to force beginners to appreciate what they need to do to play well in slow games against their peers. We don't just naturally do 1-3. Beginners tend to impulsively want to go with the first move that pops into their heads. They are well advised to slow down, stop, and think about what they are doing. Providing a neat little algorithm is a great pedagogical tool, which is like a ladder that they will discard once it becomes automatic and second nature. Making it explicit provides a common language for discussion and evaluation of their performance, and gives them things to work on (e.g., their visualization skills).

As I said in my Chessplanner document, the fact that most thought processes are written as a serial set of steps, like a computer algorithm, should not be taken that seriously. There, I wrote (and I still completely agree)

[I]s it important to go through the steps in order? No. While it is important to apply each step on each move, the order of application is not essential. Though the Chessplanner decision procedure is written as a serial thought process, this is largely an artifact of the medium used to describe it (namely, written English), and things are rarely so tidy in practice. For instance, when a material threat pops out in Step 1, I typically consider multiple defensive moves right away, and eliminate some of them as obviously foolish (officially part of Step 3). This seems to work for me, and trying to impose a serial order on such a spontaneous and useful process would be foolish. Pattern recognition is not independent of board evaluation (Step 2) or understanding move consequences (Step 3), so with practice, many aspects of Chessplanner will fall under the purview of "pattern recognition." I welcome this as a sign of developing chess expertise.

So, in sum, an explicit thought process is like training wheels on a bike. You don't expect to need it forever, but it lets you build up good habits and provides a safety net so you don't crash and burn. There is nothing sacrosanct about the linear, step-like nature of the thought processes. What is important is that you do the thinking, not the order in which you do it or whether you are consciously working down a set of boxes you are checking off in your head.

Finally, I'll end with a quote from MDLM where he describes his experience after completing the Seven Tactical Circles:

When I first developed this study program there was no third step. I thought that the first two steps [vision drills and the seven circles] would be enough. Unfortunately, I soon discovered that transferring my tactical ability to OTB games was quite difficult. Repeatedly I would look at the board for a few seconds, decide that there was no tactical shot and then make what I thought was a reasonable move. Time and time again I would turn out to be wrong and lose my queen or give up a mate. As a result, I decided that I needed to develop a more structured way to think about my move during the game. I experimented with several recipes and finally arrived at an eight-step procedure.

Even for Michael, knowing the patterns in a computer tactical puzzle context wasn't enough for the knowledge to show itself in real games. In my experience, without a thought process, the knowledge I've gained about chess is much less likely to show up in games. If you are beyond that stage, then congratulations. You have graduated to a two-wheeler.

Chess thinking process

Note: slightly revised 12/12/06.

A couple of bloggers have recently been discussing the need to come up with a thought process (Nezha, Tempo and Fierabras discuss it here, here, and here, respectively). Also, a while back Patrick over at Chessforblood wrote up his own thought process, VICE(TM).

While quite a few players poo-poo the idea of using a thought process, my attitude is this: everyone uses a thought process. Some people apply it consciously and explicitly, others unconsciously and automatically, and some are in-between (perhaps in quiet middlegame positions they consciously go through various board-evaluation criteria such as pawn structure, looking for a good plan). Further, some people do just fine without an explicit thought process, and some people will be helped more than others by making their thought process more explicit (what if Kramnik had had a simple blundercheck step in game two against Deep Fritz?).

We all know a lot about chess. A thought process increases the probability that the knowledge we already have will express itself in real games, decreasing the likelihood that we will miss obvious tactics or mates. Most coaches suggest that their students initially follow a set procedure for finding moves during a game. Since they are better than me (and probably most people reading this blog), I'll bet that this is because conscientious application of a thought process helps their students get better faster. With experience, the process should itself become largely automatic and unconscious, and not feel like such a stiltifying pain-in-the-butt to apply.

Ultimately, most thought processes usually boil down to the following steps:
1. Evaluate the present position to determine the strengths and weaknesses of each side.
2. Pick candidate moves that will increase a strength or eliminate a weakness.
3. Analyze the variation trees of the candidate moves, and pick the candidate with the best evaluation in all possible continuations.
4. Blundercheck.
5. Move.

A few months ago, I spent a lot of time writing up a thought process, which I called Chessplanner. It basically consists of the above five steps, but the first step in CP is unique as it explicitly says that right after the opponent moves, don't think, just allow your pattern recognition machinery to work its magic, and see what tactical or other interesting moves pop out for you. As I said in the document linked above, "In previous versions of Chessplanner, I didn't explicitly make accommodations for pattern recognition, but it was an obvious shortcoming when I tried to apply it in real games. Previously, the first step was to evaluate the board, but it was counterproductive to start right in with such explicit calculations as it impeded the operation of my pattern recognition abilities."

In retrospect, it isn't only pattern recognition that this first step makes room for. It is much more inclusive than that. It allows room for all those powerful spontaneous, unconscious processes that we possess, built up via experience. It is while playing from this reservoir of unconscious knowledge that we achieve that 'flow' experience that Nezha and Fierabras discuss in the posts above.

Personally, though, just being in the flow isn't enough for optimal play. I often find even better moves when I consciously look for them, considering possible discovered attacks and the like. As I improve, though, these simple tactics pop out without effort. This is why I am not dogmatic that people need to consciously use a thought process: most players are so much better than me that things I need to think about just occur naturally for them. I bet that GMs hardly ever consciously apply a thought process. I have found my thought process especially helpful in forcing me to avoid bad habits (especially playing too quickly and not blunderchecking), and this may be the main reasons many coaches suggest that novices use a process.

There is a lot out there on the topic of thought processes in chess. Material I have seen that includes substantive discussions of thought processes are:
1. Kotov's Think Like a Grandmaster is the locus classicus of most subsequent work, though it actually doesn't lay out an explicit thought process. He focuses most on how to evaluate a position and the importance of being efficient in thinking through variations (e.g., don't analyze the same line multiple times). His approach of thinking through vast game trees is now considered somewhat unrealistic for use during games in today's relatively short time-controls.
2. Various Heisman articles such as A generic thought process.
3. Soltis' How to choose a chess move. He doesn't suggest any single thought process, but reviews different types of thought processes and evaluates their strengths and weaknesses in different positions. It is the most recent major revision of Kotov's work, and is quite good. I probably won't read it closely, cover-to-cover, for a couple of years, though.
4. Harding's Better chess for average players, a wonderful little gem I just found that is meant to be your 'second' chess book. It includes a cool 10-step thought process in Chapter 6 (but it basically reduces to the above five-step process).
5. Silman's How to Reassess Your Chess.

If anyone knows of any others that explicitly discuss thought processes in chess, please comment it in! In the next year, when I revise Chessplanner, I'll want to incorporate quotes and insights from all the major resources out there.

Finished Pandolfini's book: Review

Russian Chess was a great read. Here is my mini-review.

I loved this book. One reason I like it so much is that it is the only 'move by move' analysis book I have that I can read without a board nearby, which made it great airplane and bedtime reading. He shows a board, on average, every four moves, which is easy to follow (at least if the picture is on the same page as the subsequent four moves). The six games are well annotated with lots of strategic gems. He asks a question (or two) and answers it after every move. The questions come in many varieties: some beginner-type queries (e.g., what about this move?), some are more didactic (who is doing better right now?). A really nice feature is that he peppers the book with sidebar quotes featuring chess wisdom-nuggets from the Russian masters. Get this out of print work while supplies last (they are available very cheap, used, at Amazon).

My criticisms of the book are minor. Only one game begins with 1. e4. I guess for a book on strategy this makes sense, but e4 is still played by GMs I think. Also, the book advertises itself as analyzing the chess of the 'Kasparov generation', but there isn't one game of Kasparov in there! Third, while strong on the opening and middle-game, there is little to no endgame analysis, so if you are looking for help finishing a game, you won't find it here. He can't really be faulted for this: authors have to make sacrifices in one area to focus on others. Finally, it's almost like he wanted to make it accessible only to people with a college-level vocabulary. Instead of 'cowardly,' he'll use 'pusillanimous'. Instead of 'knowledgeable,' he'll use 'erudite.' It reminded me of grading students' papers who had clearly used their computer's thesaurus function a few times too many. I found this kind of funny.

Finishing this book made me realize I have finished only a fraction of the chess books that I own. I need to dig in...

Circles Update: Finished minicircle 2.6 after a little break in Canada. I was doing 60 problems a day. Now I'm doing 80 a day in the seventh circle. I am typically getting one or two wrong out of 300 while my speed continues to increase. Soon I'll be doing all 300 every day until they are so easy I can do them while listening to talk radio.

	# Circles	Percent Correct
Problem Set 1	14	98-99-100-100-100-100-100 100-100-100-99-100-99-98
Problem Set 2	6	90-93-96-99-99-99
Problem Set 3	0
Problem Set 4	0
Problem Set 5	0

NOTE: Circles done with CTB.

Greetings from Whistler

I'm attending a conference in beautiful Whistler, Canada. The gondola is right outside my hotel. The 2010 Winter Olympics will be held here, and I can vouch for the fact that it was a good choice. One thing I've noticed: Canadian people are ridiculously nice. It is almost unsettling how gentle and sweet they are, even the scary looking ones with nose rings. It makes me want to raise kids here.

No tactics here, but I started reading Euwe's Master vs Amateur. What a great idea for a book. Instead of analyzing master versus master, look at games in which one or both players is a normal club player and makes normal club player mistaks, and show how a good player should exploit those mistakes. It seems so obvious, but then again most GMs aren't known for their pedagogical prowess. They need to release this book in algebraic notation. Also, I will finish the last game in Pandolfini's Russian Chess on the plane ride home.

On the plane ride here I was playing on my pocket chess set, and this adorable little girl kept peeking back at me between the two seats in front of me. At one point I looked up and she was holding up her own portable chess set between the seats, smiling. I moved to sit with her and her mom and we played a couple of games. She was about six and kept trying to give me her queen for free. I gave her a couple of tips (e.g., don't open 1. h4 2. Rh3) to help her game. It was fun. She was one of those kids that makes ya think having kids might not be all that bad.

Zenchess is a mensch

Zenchess seems to be some kind of chess analyzing machine. He has provided a wonderful video analysis (!) of my second game from the triumverate of games I recently posted.

Once I've studied it I will leave some detailed comments at his site.

Update: I just watched the video, and it is awesome. Some highlites: a helpful analysis of when to (or not to) push those rook pawns in the opening, instruction on how to develop more actively in the opening, helpful analysis of pawn structure and pawn moves in general with an eye to space, tactical motifs, piece activity, and some useful endgame tidbits. This hour-long analysis of a game between two amateurs will be useful to anyone struggling to get better at this mighty game. Thanks again, Zenchess. This is the kind of analysis people would gladly pay money for...

Time to learn about the English...

I don't have any standard response to the Dreaded English. Sometimes I play symmetrically (1...c5), other times I play the center response (1...e5). After that I just try to trust my judgment.

Well, I did that last night and got demolished. I spent half the game on the first fifteen moves, and then ended up getting caught in a tricky attack from his queen as I started to rush my moves. My opening wasn't the problem, but chewed way too much time off the clock. It has been my experience that it is very helpful to have a skeleton of a repertoire in response to all the major openings. This saves me time, and lets me build up wisdom each time I see that opening in a real game.

So far, the Dreaded English hasn't come up very much. But now, alas, its time has come. After reading through the options in Sam Collins' wonderful book Understanding the Chess Openings, I've decided to go with 1...e5.

I am at the point now where I don't fear learning new opening theory, but kind of enjoy the process of looking for a response to something that dinged me. It's like going out on the dating scene, trying things out, and then settling on The One for me and living happily ever after.