Monday, May 23, 2016

Spiralling vultures

Hitherto I have found that the main problem with tactics lies in the inability to see a combination as a whole (level 3). There are a few temptations that distract you from seeing the big picture.

Temptation 1: trial and error
The first temptation is dubbed trial and error. This means that you see an interesting move, and want to see where it might lead to. Effectively, this is gambling. As in real life, gambling usually costs more than it yields. It depends completely of the position if the gamble pays off. In the metaphor of the missing keys which lie on your bed, if you are searching in the garage, you are wasting time. Your mind is gadding around at level 1 (moves).

Temptation 2: tactical motifs
When you see an interesting pin, you might try to exploit it. At the same time, you loose sight of the big picture. You can't know beforehand, if the pin is actually an element of the required combination. Effectively, it is a gamble, which might or might not pay off. You might be looking in the garden. Your mind is gadding around at level 2 (motifs).

Temptation 3: heavy duty logical reasoning
When you yoke up a logical reasoning, you can't know beforehand where it will lead you to. You dive in a tunnel, and while following the tunnel, you are not aware of what is outside the tunnel. Effectively, it is a time consuming gamble, which might or might not pay off. You might well end up in the limbo looking for ghost keys. Logical reasoning is quite error prone, so it takes a lot of time to check the results too.

Spiralling vultures
Initially, I wanted to develop a checklist to help me to guide my attention at level 3 (seeing the big picture of the combination). Due to the comments and the discussions with Robert Coble, I realized this might even be not necessary. What if I just persist in viewing at the board from a distance? Disciplining the mind to retract from any temptation that tries to distract?

This idea is consistent with an idea I had long ago about backwards thinking. I have been spiralling around this idea for long, and more and more it looks as if it is the best way to go. Ones you see the idea, finding the moves that execute the idea is usually no problem.

So now I'm spiralling above my enemy like a vulture, looking at the position from above, rejecting any temptations to go in for the kill too prematurely. Only when all important elements of the combination have revealed themselves, I dive to the ground to pick my opponents eyes out. There is another vulture in the air too, which is spiralling around my fortress in the hope for a surprise attack. I try to look from the perspective of the hostile vulture too, in order to prevent any devastating counter attacks. But usually that picture is somewhat out of focus.

Keeping the overview. As usual, the enemy vulture is out of focus
My first experiments with this way of surveying the position proofs that everything can be seen from a distance indeed. It only takes a lot of time to see it. But that shouldn't come as a surprise. That is what we found long ago: first do it right, then do it fast. The idea of seeing is consistent with what I would expect concerning the brain parts that are involved in acquiring chess skill. Experiment to be continued. . .

Saturday, May 21, 2016

Is logic logical?

So far I found two types of positions that cause me trouble: positions where I overlook a tactical element and positions where my logical reasoning goes astray. And of course there are combinations of those two types

Logical reasoning reinvestigated
After doing a lot of problems with my previous post about logical reasoning in mind, I can't help but wondering if logic reasoning is actually the best tool for the job. I remembered a lot of muttering reactions long ago when I proposed to exchange logical reasoning for something that you actually see. The most grumbling was founded by the idea that human thinking is so much more than pattern recognition, and chess is so much more than a simple game, that we need at least to engage the whole human mind in thinking about a chess position, with a big role for the human reason.
Albeit that might be true fore a complex middlegame position (I commit the heresy to doubt that), the battering of our human vanity by computers the last two decades, and the fact that chess tactics at CT with a below 2200 rating are simple in essence most of the time, gives me the courage to ask this question again.

Intermezzo.
I insist in calling the below 2200 problems simple, since that is how they look with hindsight. The fact that these problems are high rated is due to the fact that there are a lot of dumb counterparts out there that play chess like an idiot just like I do, and spoil the statistics in doing so. But the fact that broddlers can't solve a problem is no reason to call a problem complex. It is only a reason to call us broddlers. From time to time I get a problem that remains complex, even after solving it, but I make a note of that to study them later. If I was a master, I would see the problems as simple beforehand.

Think like a grandmaster
When you were of grandmaster level, most of the below 2200 rated problems would look quite simple. A grandmaster would apply very little logical reasoning, since most things are evident for him.

I noticed that when I start logical reasoning during solving, it is as if I dive into a tunnel. Being only aware of what is in the tunnel. When I come out of that tunnel a few minutes later, I have either confirmed the line I'm investigating, or I have falsified it. Since logical thinking is quite error prone, a lot of checking must be done to verify the result. Which is another time consuming activity.

So if a master doesn't need much logical reasoning, why should I? When I took a closer look at the position of the previous post, I realized I could do well without logical reasoning. The missing tactical theme can very well be seen.

In fact, when I apply logical reasoning, I should be very suspicious, since it is a tell-tale  sign that I'm in tunnel mode. I should zoom out and start scanning again at level 3 (total combination)

Chess heuristics

Robert Coble wrote a long article about applying the methods from training pilots and karateka to chess. I made a little survey of it for myself. It inspired me to start from scratch, and make a checklist that aims specifically at seeing the combination as a whole. I will let you know how it goes.


Thursday, May 19, 2016

Chess logic

The position below is of a special pedigree. It was easy to spot all construction blocks of the combination. The only thing I had to do is to put them in the right order. For that, a little logical reasoning is necessary. In tests, I usually score extremely well when it comes to logical reasoning. But here I could not get the building blocks in the right order within 4 minutes. And when I saw the correct answer, I slapped my proverbial forehead.

Black to move
6k1/1p4p1/3PN2p/p7/1n1R3P/1q4P1/5PB1/6K1 b - - 1 1
 solution

This kind of logical reasoning is both time consuming and error prone. It need to be worked on separately. It is quite different from the position in the previous post, where the problem was that a certain tactical building block was overlooked.

I'm a bit surprised to see that I'm so bad in logical reasoning when it comes to chess, while it usually is my forte. If I could only overcome these two flaws: poor chess logic and missing tactical building blocks, I would make a quantum leap forward in tactics.



Monday, May 16, 2016

Finding a remedy

After 51 days of doing blitz exercises at CT, the diagnosis of what's wrong is pretty clear. Yet I'm not one step closer to a remedy. Let's summarize what we have found so far.

The diagnosis
  • Logical thinking takes too much time and should be minimized. 
  • I wander too long at level 1 and 2 (individual moves and individual tactical motifs)
  • I have oversights at level 3 (combination, how the pieces work together)
  • Guiding my attention to find the oversights at level 3 has proven to be elusive so far.
  • Decision between different choices takes much time
  • Confusion increases time usage in an exponential way
What has been tried so far
  • Creating a formal thought process.
  • Applying a formal thought process.
  • Slow post mortem to ingrain the geometrical patterns into the brain.
  • Slow post mortem to ingrain the logical patterns into the brain.
  • Solving at the highest possible speed. 
  • Guiding the attention at level 3.
  • Categorizing the problems
  • Analyzing the time consumers
  • Focussing on the squares where attackers converge
  • Focussing on the defenders
  • Slow problem solving 
  • I might have forgotten a few experiments
A few facts
  • Random trial and error can be replaced by logical thinking
  • I can prevent almost all oversights by taking more time. The effect is that both the red (error) and the green (correct) colours at the summary of my performance statistics at CT turn into yellow (correct but too slow).
  • The best metaphor found so far is that of the lost keys which lie on your bed. The problem is not that you do not recognize your keys, but that you forget to look in the bedroom.
  • Confusion is highly personal. It is related to complexity. It causes a memory overload due to processing high numbers. The high numbers might be the result of roaming at a too low level. There are more letters than words in a text, so processing letters causes a memory overload.
  • The nett result of  51 days training is close to zero, if I don't reckon with an initial improvement the first days due to adaptation of the exercises.
  • I have got more knowledge and understanding of certain positions. But not more speed.
  • You can't find something at level 3 by looking at level 1 and 2 only. Only by accident will that happen.

Remedy
Given the amount of remedies that have been tried, finding a remedy is not so easy as it looks. None of the experiments gave me a definite feeling or even a hint to be on the right track. Robert Coble has given a few beautiful examples of training pilots and karateka's which hint to how we can go from serial to parallel processing, but is not clear how that exactly should translate to chess tactics.

Somehow, the solution has probably to be sought in visual patterns. Maybe, the fusiform face area has something to do with it. But some critical knowledge about tactical improvement is still hidden in the dark.

Example
There are a lot of different positions which can be categorized in certain types. Not every type needs the same remedy. But there is one category that stands out: the one where you fail to see the final combination because you look only at the lower levels. Below you find a perfect example of that.

Black to move
 3q1rk1/1p2Rppp/2p5/1bP5/3b1P2/1N4PP/6BK/4Q3 b - - 1 1
solution

 During investigation of this position, I overlooked how the black attackers converge on certain squares where they attack the enemy targets. After 5:34 min I saw the essence of the combination. How to speed that up? How should the attention be guided towards d1, where the black queen executes a duple attack on the white king and knight?


Thursday, May 12, 2016

Zooming out

The following diagram is a beautiful example of the problems we are trying to tackle. It is by no means a complicated position, so there is no confusion. This means that confusion is not the first problem we have to fix (!). All geometrical patterns are perfectly well known, so no problem in that department. There is not much visualization required, no problem there. I dabbled around on level 1 and 2. I.e. I looked for moves and for tactical motifs or themes, or whatever we call it. I knew I should look for level 3, the combination. I can't say that any logical reasoning really took off. You can't find a combination by just scanning level 1 and 2. If you are lucky then you get an idea of level 3, but if you are not lucky you are pretty much toast. You can't work your way up. You can't comprehend a sentence by just scanning the letters and the words ad infinitum. Even a checklist that in itself is based on questions that don't expand beyond level 2 is not going to give you an idea of a higher abstraction level.  Somehow you must zoom out and look at the bigger picture.

White to move
r4rk1/pbp4p/2q3pB/2p1p1N1/8/6P1/PPP2QP1/3R2K1 w - - 1 1
solution

After "thinking" for 21 minutes, I decided to have a look at the answer. Although I tried to zoom out mentally, I didn't see the main idea behind the combination.

What is needed to find an idea on level 3? The checklists that reveal intricacies about level 1 and 2 are no longer conscious in use by me, since they don't address the main flaw in my approach, and it is a lot of work for something that doesn't work most of the time. I use a checklist for level 4, the orientation phase, and that works well.

What I need is a checklist for level 3, as a kind of training wheels. Had I for instance asked in this position "how can the white knight and the white queen collaborate together?", then I had seen the main idea behind the combination immediately. Once the idea is found, the accompanying moves are usually found fast. That is not the problem. I'm going to try to develop a level 3 checklist. I'll be back!

For your convenience:
Level 1 = chasing letters = t&e = chasing individual moves
Level 2 = chasing words = recognizing motifs
Level 3 = sentence = combination
Level 4 = text = solution
(maybe someone can formulate it a little bit less informal)

Tuesday, May 10, 2016

Another look at missed patterns

So far, investigation  has revealed two types of error:
  • Missed patterns
  • Flawed logic

Today I had a closer look at the missed patterns. I found out, that if I take more time for a problem, I stop missing patterns. The price tag for this is twice the amount of time than the average solving time of other users for a problem. This means, that if I take my time, then sooner or later I will find all patterns that I used to miss. This implies that the missed pattern problem is actually a slow logic problem. Which I'm very glad to find out. All my failure is logical reasoning related.

Only once in every 25 problems (4% of the cases) I make an error that is due to bad logic.
I'm glad that both problems are related to logical reasoning. I will make an in depth study of the 4% flawed logic errors. What remains is 64% failure due to slow logical reasoning and 32% correct solved problems in time.

And so the key question remains: how to speed up logical reasoning? After 45 days consecutive error solving, I'm still no closer to an answer to that question. Yet in this case, my gut feeling tells me that it should be possible to find an answer and to improve accordingly.

This one took me 41 years, and I still failed.


Saturday, May 07, 2016

Attack your confusion!

After I made an error, or use excessive time while solving problems on  CT, I always take my time during the post mortem to analyze what has been going wrong. Once in a while, I bundle my errors, and try to categorize them. I found that I made two types of errors.

Missing a pattern
Often a pattern is all too familiar, yet I miss it. A knight fork, a simple mate pattern, that kind of things. To be honest, I'm not too worried about that. It is caused by the fact that with CT, you must time and again acclimatize to a position. When I play a game at a slow tempo, I rarely miss these patterns. On the other hand, to become better at CT, I will have to fix this. I belief that should be perfectly doable. For the time being I focus my attention to fix a more nasty kind of error. Just for the record, I already explained that I don't belief that salt mine style exercises are of much help here.

Confusion
From time to time, I bump into a position that confuses me. When the mind is confused, time usage grows exponentially. I don't miss familiar patterns in this case, I miss the patterns because they are not thoroughly familiar. The positions are always related with exchange sequences, and subtle intricacies concerning maintaining the initiative. I have written a few times about the initiative the past weeks, and I noticed that these posts generated very little page views, and almost no comments. I'm very surprised by that.

It is very hard to imagine that this isn't a major cause of errors for most readers of this blog. But maybe in positions that are different than the ones I show you here. With hindsight, it is hard to see why I had so much trouble to fully understand the following diagram. It has taken me hours to grasp the simple essence of the position, and I have written out the whole tree of analysis for that. The position exposed a few major flaws in my chess thinking. If you too think that the position is dead simple, you should look for positions where you are confused yourself. The timer of CT shows you the way.

 
Black to move
6k1/pp4pp/2p1brq1/3pn3/4P2P/2N1PBP1/PPP1Q1K1/5R2 b - - 1 1
solution

I played 1. ... Nxf3 here, which is wrong, and I was genuinely surprised to see that I should take with the rook here. What is the combination about? The heart of the combination is the duple attack 1. ... Rxf3 2. Rxf3 Bg4, which pins the white rook to the queen. Normally this shouldn't work, since the rook on f3 is twice attacked and twice defended, but due to the high value of the white defenders, black outnumbers white on f3. The first move is a preliminary move, which is designed to lure the rook into the pin with a capture. The reason why the black rook should execute the preliminary move, is that when the knight takes on f3 (1. ... Nxf3 2.Rxf3 Bg4) the white rook becomes a desperado, with a clear target on f6 and g6. When 1. ... Rxf3 2.Rxf3 Bg4, then the white rook has no such profitable target.

As said, the position is dead simple with hindsight, but it took me a few hours to iron out the confusion in my brain. Why was this position so confusing to me? Somehow I wasn't able to see what the pieces were doing in this position.