The final conundrum

 It took me 18 years to find the bottom steps of the ladder to chess improvement. Often I thought that I had found the beginning. But time after time, a close scrutiny of the results of my training sessions, tempered the initial enthusiasm. Usually after a month or three. Mr Glicko and Mr Elo are relentless.

I always managed to keep my mouth shut when I thought that I was successful. Until May this year. When I discovered the importance of chess logic as the base of transfer between chess problems, and as the base of analogies in chess, I felt that I finally was on to something. I felt that I was improving. Of course that is subjective proof, but since I don't was punished by disappointment after three months, that was good enough for me. It worked. So I finally found to courage to speak up a bit.

But it wasn't efficient. It started to dawn on me that still something was missing. I started to analyze my losses, and I was working on my five new openings as a madman. I noticed the weakness of my middlegame play, and I started to work on my positional play frantically. I discovered that result = what x how.

Another breakthrough came when I started to experiment with different problem depths. In May I started with problems of 5 ply deep. That was an arbitrary choice. When I tried problems of 7 ply deep, I felt I could still work on my logic, but my training sessions lost momentum. It became more difficult and daunting to make progress. That was the reason that I started to experiment with problems of only 3 ply deep.

I had done so in the past, and I had abandoned it because I thought it was too easy. But back then, I focused on pieces and positions, and nowadays I focus on logic, which has a higher level of abstraction. When I selected problems with a rating of higher than 2100 AND 3 ply deep AND mate, I thought I had found the holy grail.

But again I was in for a shock. Sofar, I only had focused on mate in 2. The golden standard is the performance of Susan Polgar during her simul. If I'm not at least heading in that direction, I am going nowhere. For the first time I had the feeling that a tenacious training might lead to a faster seeing of mate in 2. The choice to focus on mate in 2 solely, was arbitrary too. When I tried to do other tactics in the same manner, I all of a sudden felt totally lost. Since a mate in 2 ends after two moves, no matter what. But a non mate tactic doesn't necessarily end that way. You might find yourselves in the thickets of variations that are not clear at all. I called that the final conundrum that must be solved. "How to train tactics that don't end with mate?". 

While writing the previous post I had a revelation. Blogging does that to me, that is why I write everything down. It became apparent to me that you cannot begin with logic when you have no idea what you want to achieve. And that you can only know what you want to achieve when you see it.

When I started with logic, I had suppressed the vulture. But without the vulture, you have no starting point for your logic. This means that you must split the training in two. The vulture must be trained, and the logic must be trained.

Let me try to clarify that.

White to move

q2r2kb/1b3p1p/5PpQ/3np3/2p1N1B1/2N5/1P4PP/5R1K w - - 0 2 

[solution]

Make a clear distinction between the salient cues and the ensuing logic. I said it often, I don't think it is important to find the solution ourselves. You can do so if you like, of course, but I consider it a waste of time.

In the previous post I made a difference between the executional move and the preparational move. The executional move has salient cues that must be learned by the vulture, while the preparational move must be used for training your logic reasoning.

What execution are you invited for in this position? The position invites you to play Ng5 and mate your opponent. Get a clear picture of the salient cues which gets you there if the opponent is not allowed to make moves.

Once you have a clear picture of what you want to achieve, you must ask yourself how your opponent is going to prevent it.

You will find that the knight on d5 is a very annoying beast. When it starts to move it unleashes blacks bishop, queen and rook.

You need some perseverance to get rid of the beast. When you find the move 1.Be6, you will find another salient cue. 1. ... fxe6 is impossible due to 2. f7#

The vulture might have missed that initially. Put that salient cue in your backpack.

It is a kind of Stoyko exercise, but then for tactics. In the end you will have a list of salient cues and a list of narratives. Take your time.

This is the direction in which I look for solving the final conundrum. Whether this conundrum is final or not remains to be seen. Time and again, the devil is in the details. But I feel very optimistic about it. But that never has proven to be a guarantee.

Execution

 When you have 2 move (3 ply deep) problems, one move is used for the actual execution of the wood gaining process, while the other move either prepares the execution or deals with the consequences. In 54% of the cases the first move is a preparational move, while in 46% of the cases you start immediately with the execution.

The level of the problems is between 1800 and 1850. This is the division of the executing moves in frequency of occurrence:

• B.A.D. (Barely Adequate Defended) piece 46%
• Mate 22%
• Duplo attack 16%
• Promotion 11%
• Trap 5%

In general, the salient cues that give away the executional moves are not too difficult to spot with hindsight. This means that you have to find the preparational moves or the moves that deal with the consequences. These moves must be revealed by logic thinking. By answering the question "how can I make this executional move work?".

If you don't know what the executional move is, you cannot start with logical reasoning. So you must hoover above the board spiralling in vulture mode, until you know what you want to achieve. Then the logical process starts.

The large amount of B.A.D. (Barely Adequate Defended) pieces might seem strange. But when you look it up in the database, there is usual a good reason for that. For instance you are halfway an exchange, you are harvesting a trapped piece, you punish an incorrect offer, or you sacced a piece yourself which your opponent shouldn't have accepted. Or it is simply move 39 where even grandmasters in time trouble can go astray.

Tempo battle

 The mate patterns can be conquered by doing mate in 2 exercises. The PoPLoAFun/killbox framework covers all there is to know about mates. 10 actions and 10 salient cues is all there is to know. But tactics that just gain wood is a totally different animal.

I work with 2 move problems (3 ply deep) that do not end in mate. ChessTempo considers them to be winning after you have provided the first 2 moves. But in practice matters might not be that simple. Sometimes you are just 2 pawns behind after the 2 moves. Superior piece activity is supposed to provide you a win after 10 or 15 moves. But since I'm no engine, I can't look that far ahead so easy. So albeit ChessTempo forsees a bright future for me, I delete these problems from my database for now. 2 movers are complex enough when I only have to look in the near future. Mates in 2 (M2) are way more simple, since the future is mate on the second move.

The PoPLoAFun/killbox framework certainly plays a role in 2 move tactical problems that don't end with mate (T2). But there is a lot more to it.

The core is, as expected, the duplo move, the trap and the promotion. Since these three themes are the only way to gain wood in a forcing way. You have to see the salient cues that play a role here.

But usually, seeing the theme is not enough. There is always an additional problem that you have to solve. These scenarios are common:

• Execute a duplo attack and see if you can save your pieces afterwards
• Prepare the duplo attack by a tempo move
• Save your attacker with tempo before you execute the duplo move

A common theme seems to be the battle of the tempi. Somehow I'm terribly weak in tempo play. I lack intuition in this area.

I don't believe in calculation here. Of course, you can calculate the moves. But I simply cannot belief that that is the way how a grandmaster does it. If Susan Polgar uses only 2.6 seconds pert move during a simul, she doesn't rely on calculation. She simply sees the tempi. Calculation is for people who are too lazy to learn how to see.

White to move

2r5/3r1ppk/b3p2p/8/5BnP/6P1/2R1PPB1/R5K1 w - - 1 2 
[solution]

Can you see the solution without calculation? That is the problem I have to solve: learn to see the tempi without calculation. That is the last conundrum. I hope. 

Exploring the boundaries

 Not only have I found a way to train tactics that works, but by focusing on 2 movers (3 ply) recently, I have found a way to do it in an efficient way too.

Since May 1st I focused on mate exercises solely. Since I adhere the adage that the fastest way to come to a conclusion is to make extreme choices. A 2 mover usually involves two or three logical narratives at average. So far 6 narratives is the maximum that I have found. This means that a mate in two is complex enough to study everything that is important around mates. It doesn't make sense to start with 3 movers until you fully master the 2 movers. If that is even possible.

This weekend I have been busy to inventorize the logic of 100 mate in 2 exercises and combine it in one coherent logical framework. It turns out that all logic can be captured in the PoPLoAFun system when I extend that with some logic concerning killboxes. 

There are only about 8-10 logical actions to choose from when working on a mate in 2. Think of clearing a line of attack, build the wall of a killbox, herding the king into a killbox et cetera.

Furthermore there are about 8-10 salient cues that you must learn to recognize. Think about the line of attack, the point of pressure, the defender of the point of pressure and the like. So that is pretty straight forward.

Currently I'm trying to do the same with all the other tactics. There are about 10 times as much occurrences of tactical motifs that don't lead to a mate. So it is 10 times as likely that you encounter these logical ideas in a game. Whether this means that the logical framework is 10 times as vast remains to be seen. I will keep you posted.

Fun and momentum

 The major change that I have made in my training regimen, is that I changed the focus of pattern recognition from pieces and moves towards the chess logic behind the moves. That is a shift from pure geometrical patterns to a mere conceptual approach. That solved the transfer problem of knowledge from one problem to another. A geometrical pattern can be ruined by just one pawn on a different place, while the more abstract approach of a logic pattern is relatively piece and move independent.  Luring an overworked piece away is independent of the pieces that are involved and their exact place on the board.

I wasn't completely satisfied with the 3 movers (problems of 5 ply deep). It works, but it doesn't feel very efficient. So I changed towards the 4 movers. That was still fun to do, but it felt even worse in terms of efficiency. Somehow my training lost momentum. So I decided to have a closer look to 2 movers (3 ply). I have done a lot of 2 movers in the past when I was still focusing on geometrical patterns.

I had dismissed 2 movers in the past because I had the feeling that everybody sees them on the board fast, so it isn't going to help me to win games.

But now I have added a minor yet important twist. Not only am I focusing on logic patterns instead of geometrical patterns, but I made a selection of 2 movers with a rating beyond 1700. Not only is the training more fun, but it gained way more momentum than the 3 movers.

I really might be on to something here.

How deep must your puzzle be?

 Since May 1st I experimented with tactical puzzles that are 3 moves deep (5 ply). That definitely worked. Somewhere in the summer, I decided to give 4 movers a go (7 ply). After a few months, my conclusion is, that that works too, but it is not as efficient as 3 movers.  So I'm working with those 3 movers again.

My picture of both the openings and the middlegame is pretty much complete, so I know exactly what to study positionally the coming years. The boundaries are clear:

• Piece exchanges
• Restriction of enemy pieces by pawns
• Increase piece activity by pawn moves

CM Kabadayi is my guide here.

This means there will be one loose end the coming years: the endgame. I will study my five new openings and the related middlegames positionally, and do tactics on a daily basis. Since I don't want to divide my attention, energy and time too much, I accept that I will screw up a won endgame regularly. I consider that as a luxury problem.

For now.

More about pawns

 The problem with chess training always has been that there is way too much good advice. With tactics, it took me 18 years to separate the wheat from the chaff.

The book of GM Hellsten about middlegame strategy contains 826 potential relevant variations. But learning them is like learning an encyclopedia by starting from the letter A. Nobody can do that, unless you are an idiot savant of some sort. I just have the idiot part covered, so I got to turn to some chaff separation first.

I was happy that I could separate quite some chaff, until I found the two subjects that are the place to start: exchanging pieces and pawns. It turns out that there are about 10 types of positional exchanges, and the actual study of them is pretty straight forward.

But studying pawns is a different animal. Here again, there is way too much material to study.  In the book Hellsten dedicates 475 variations to pawns.

After some time I found that the two most important areas are where the pawns have an effect on the LoA landscape (LoA = line of attack): piece activity and restriction.

You can use your pawns to increase the activity of your own pieces, and you can use them to restrict your opponents pieces.

You can do a lot more with your pawns of course, but that only distracts you from fully mastering the building of a viable LoA landscape. GM Nigel Short once said, "Modern chess is too much concerned with things like pawn structure. Forget it, checkmate ends the game!"

This reflects the difference in pawn moves that affect the LoA landscape and pawn moves that are designed for a favourable ending. You first have to master the building of a LoA landscape.

CM Can Kabadayi happens to have written the two following books about pawns:

The Art of Burying Pieces and The Art of Awakening Pieces.

The first book is about restriction, and the second book is about improving the activity of your own pieces. I'm studying both books currently. Much recommended!