Sunday, September 16, 2007

Defining the real problem

In the previous posts you could see my initial struggles with the matter. High time to formulate what the real problem is. If we find an answer to this problem, we have found what seperates the masters from the amateurs, the men from the boys.

The idea's about invasion squares are already pretty familiar to me since I've been thinking about them for months. The idea's are already crystallized to a certain degree. I realize that this subject is new to the readers of my blog. What I now want to talk about is new to me too. This means that my thoughts haven't crystallized yet. That will it make it even harder to you to follow me. I hope you will bear with me. You will find much overlap with previous posts too.

I encountered the problem for the first time to the full extent in a position you are already familiar with:

Black to move and win.

For 12 days in a row I'm struggling with this position.
I have written down the total tree of variations with moves that weren't illogical. That consists of over the 200 moves. I found that there are 3 ways to prune a tree:

This is the most common way. Since you have to make a move every now and then you have to gamble when you play a promising line. In my last two games I gambled on the right continuation. One I won, one I lost. I don't think that gambling has much future in chess. But if you lack the skills it is often the only method.

I have been too harsh in my judgement about my intuition in this position. I dismissed 1. ... Bxe4 as a bad way to take on e4 if you compare it to 1. ... Nxe4 and 1. ... Rxe4. But it is indeed the most bad way, even when the counter attack from Rg1 is ruled out.
Which raises the question "how do you improve your intuition?"
I can think of no other way than doing lots of problems and formulating lots of narratives.

Backwards thinking.
This looks the most promising method. My study of the invasion squares was meant to make backwards thinking possible. In the position above you get something like:
If you invade whites position with Qa2, the white king will be mated (plan gamma). The invasion square a2 is guarded by the knight on c3, which has to protect e4 too (plan bhèta). The other protector of the invasion square e4 is Ng3. With the exchange sacrifice on e4 I can deflect Ng3 which enables the knightfork e4 against the queen and the protector of a2 (plan alpha).

If my idea's about invasion squares are correct, then the only moves to consider in this position are the 3 moves that effect the invasion square e4: 1. ... Rxe4 1. ... Nxe4 1. ... Bxe4
This are the only possible moves that can lead to a viable forcing attack. Other moves are only to look at for accidental tactics. But forcing lines have to go along the invasion squares.

Actually my investigating of duplo attacks in the past was meant to introduce backwards thinking in accidental tactics. When you can identify two potential targets, you can think backwards to find out if there is a tactic to attack them both at the same time.

My ideas about invasion squares are just that. Logical looking ideas. The ideas haven't been put to the test seriously. Yet.

The pruning potential of backwards thinking is tremendous. Yet there is another nut to crack. I wrote about that in paralysis by analysis.
Even when there are only 3 candidate moves left my mind easy paralyzes. The alternating moves (white/black), the counterattack of white which interferes with blacks attack and the bookkeeping of values I tend to do all at the same time. Thus causing a short term memory overload error. Even after 12 days looking at the position my mind shuts down easy. All side-issues are solved and clarified, yet the heart of the calculation remains confusing for me. I'm quite sure that this is the biggest difference between me and a good player. I have seen people do such calculations in less than a minute. I'm in trouble even if I have 12 days to try.

Only if I'm able to solve this problem, if I learn how to do this calculation in under a minute, I have a chance to become better at chess. This is the real problem. And I have still no idea how to crack it.


  1. What makes the CTB level 5 problems different is that many of them I just can't solve without some kind of thinking backwards. I have to see some hidden mate, and then figure out ways to reach it. This is also what makes them much harder than the previous problems.

    Which raises the question "how do you improve your intuition?"
    I can think of no other way than doing lots of problems and formulating lots of narratives.

    Playing a lot?

  2. Zenchess has an interesting post about similar questions about thinking efficiently. This is a huge problem for me in games, one that has improved a little bit just by playing games.

    Do you find Tisdall's stepping stones method useful at all? I, frankly, find it of limited use, but I have only given it a limited try.

  3. I am still not sure if I understand the idea of complex tactics correctly, but I guess this is because my own problems are much more simple. Anyway, I think narrowing it down to 3 good looking candidate moves sounds like good progress! I just started reading Kasparov's book "How life imitates chess", which is more about planning and strategy in general than about chess. He stresses the importance of having a long-term plan both in chess as well as in 'real' life. I would like to suggest the following: you narrow it down to a few moves that are tactically good by calculation like you try now. Then you try to decide among them by looking which alternative fits best in your overall plan.

  4. Blue,
    playing a lot is of course helpful too, but the guys that I know of who play 13 hours a day still remain 1400 players.

    The stepping stone method from Tisdall is usefull in it's own right, but this is no visualisation problem. This is short-circuiting of the brains with your eyes wide open!

    I will have a look at Zenchess.

  5. Sciurus,
    I'm hesitant to formulate a lifelong plan like Kasparov. Before you know it you end up as the president of Russia!

    I look at ways to decompose the 3 tasks I tend to mix up a little further.

  6. Re: Even when there are only 3 candidate moves left my mind easy paralyzes

    My question: Is this a calculation problem or a mnemonic problem ? If you had perfect memory or were able to take notes during play would this be solved?

  7. Tak,
    that is a very good question. I find it difficult to describe what is exactly happening. The main cause is that the content of the short term meomory fades away too fast. If it could hold 100+ items in stead the usual 5-7 I probably would have the overview I so desperately need. Even writing it down on paper doesn't suffice, since I still cannot hold more than a few items in my mind so I miss the survey.

    Maybe the solution lies in the direction of forming chunks, or templates. I just don't know.

  8. Tempo-
    that might be a solution! As Russian president you will have direct access to a whole bunch of GM brainpower. Go for it, I might even cosider becoming Russian to vote for you ;-)

  9. My issue with your technique is really an issue with me. That is, I have a difficult time pulling away from the board and looking at the position in isolation. Your problem in the post is a snap shot. But in a game, you would have worked to achieve that position so you would automatically know all the roles of the pieces. They'd be like family. You put the bishops in one spot to serve a purpose, you positioned your knights just so. Further, you would be very familiar with the opponents pieces too. As such, I think you would be able to calculate through a position much faster than you think. But when you look at a position "cold" (like when you do problems) you have to factor in the whole board first.

    Just talking again here. But I guess I'm trying to outline a limitation to doing problems. Since you didn't create the position yourself, there are limits to what you can achieve with it. Yes?

  10. Pale,
    you are quite right that it makes a huge difference if you have coöperated to get the position on the board or not. Yet I have two remarks.

    The first is that in a lot of games there is a moment that complexity rockets sky high and that I just have to gamble, in order to stop the clock from ticking. Only if you are alert on it you will notice how often that happens. Meaning: complexity has grown over your head..

    The second is that after 12 days studying a single position as snapshot the pieces maybe have not become family, but it are already very good acquaintences.

  11. Re: Being familiar with positions we've played into compared with problems from the middle of someone else's game.

    I find that when I play a strong player, they are very good at complicating the position. So much so, that the familiarity of a position can wash away in the span of 2 or sometimes even 1 unexpected move. This goes hand in hand with Tempo's idea that the difference between the stronger and weaker player is the stronger player's ability to handle the complexity of the problem.

    Re: If it could hold 100+ items in stead the usual 5-7. Maybe the solution lies in the direction of forming chunks, or templates.

    This is of course one of the dream scenarios. That we don't need to hold all 100-200 moves of tree in our short term memory. We just need 5-7 chunks, or larger units. This is a believable idea if you've ever seen a grandmaster analyze a position. Somehow they see an entire line all at once. You ask them "what about this move?" and they respond in 2 seconds "no, because of this variation", and it turns out your knight is trapped in 5 moves. It seems they can't be thinking of these moves individually.

    Alas, I'm not sure we know how to replicate this chunking trick.

    I do think that Tempo's analysis of these complex positions has led to a lot of insight into how a chess game is won. The posts of this past week have been really great. I hope it leads to some improvement!

  12. RE: The second is that after 12 days studying a single position as snapshot the pieces maybe have not become family, but it are already very good acquaintences.

    -after spending 12 days with them, they'd be more like prison mates than family, LOL!