Saturday, April 30, 2016

Condensing logical reasoning to knowledge

While I solve a puzzle, I observe which subjects are consuming the most time. That are potential the subjects where I can improve the most. In this post about time usage I already made a list of a few time consumers. What stands out, is that there are a lot decisions in the list. In chess you must constantly make decisions, and when you base them on logical reasoning, they tend to take an awful lot of time.

It are these time consumers that must be taken on for a speed overhaul. If I can devise a set of rules for certain positions, then I can condense the logical reasoning into knowledge. Knowledge that might be automated later on. In the following diagram, I will give it a try.

[Disclaimer]
As usual, I don't know beforehand what I'm going to write. That means that maybe it isn't even possible what I propose. But that is a result in itself. Besides that, it might give you ideas, which can be helpful when you share them. The position might be trivial to you, great, maybe you can provide a few general tips that help me. I'm sure you have your own time consuming little buggers, btw.

White to move
1r3q1k/1p2bpp1/p4B1p/P1p2Q1P/R4P2/3r4/1P4P1/1R5K w - - 0 2
solution

This position is halfway a winning sequence. That means, I was already familiar with this position when I stumbled on this diagram. I have a decision to make here, which piece am I going to take? 2.Qxd3; 2.Bxg7+ or 2.Bxe7 ?
It took me considerable time to make a decision. I didn't measure how long, 30 seconds, a minute or maybe more. When I looked at this position post mortem, I realised that it might be possible to speed up this kind of decisions by devising a set of rules. Such decisions must be made time and again, and speeding them up should make a considerable difference in time usage. Let's see if we can find a few rules.

Role playing elements
Which elements play a role in this type of decision making?
  • Value of the targets
  • Own pieces under attack
  • Value of your own pieces under attack
  • Your attacker being under attack after capturing a target
  • Capturing of a target poses a new threat
  • Longest chain of captures
What is de longest chain of captures?
What chain of captures has the highest value?

After trying a few idea's and testing them in practice,  I found that it is not so straightforward. I must investigate a lot more positions in the hope to devise a set of rules with a broad application. I'll be back.

15 comments:

  1. The decision: which of these 3 takes is the best, is a matter of calculation. There is not much to calculate, can be done in "notime".

    ReplyDelete
    Replies
    1. Somehow I have found a way to spill my time here. I have no idea how.

      Delete
  2. I strongly disagree with Aox's statement. We cannot and should not always calculate everything. It does not make any sense. We are NOT computers - we are much more than that!

    I think this position has a few hidden elements. What I analyse when looking at this position?
    1. The Black Rook is completely hanging (attacked and not defended).
    2. My white bishop is attacked by the pawn (and by the Bishop, but the pawn's attack attracts my attention much more).
    3. My Rook at a4 looks a bit awkward - I mean it may be lost when it is at such a strange position. It does not give me a feel of harmony between heavy pieces.

    AFTER the initial observations is done - we can compare the variations (not before!) and check out which one is the most promising.

    V1: QxR BxB, it is ok as I won the exchange, but maybe there is something more in the position? Take notice that my Bishop can execute a "one move check" before it will be lost.
    V2: Bxp+ QxB QxR, and now I gained a pawn more than in the previous variation. It looks much better. Is any kind of better way to "lose" my Bishop BEFORE I take the Rook for free? Can I sacrifice it for any other target (the stronger/more valuable - the better)
    V3: Now I have finally found BxB! The Queen is attacked and the Rook is hanging - BOTH at the same time! Now the final check - is it possible to get away from being captured and defend the Rook in a single move? Where the Queen would have to go to realise this goal? It looks like the square d8 is the only one... but after that Bishop would take the Queen! There is no other reasonable moves and therefore QxB and QxR.

    And now the problem has arisen. Qh4+ and it attacks my King and h5-pawn. In a first look - I analysed something like Kg1, Rd8 and Qe2. However while writing this post I discovered a very simple way to defend against such dangerous variation. The sufficient enough is Qh3! (after Qh4+ was played). This move not only defends the pawn, but forces to exchange Queens or gives the time to defend against Rook entrance (Rd8-Rd5-Rh5, etc.).

    What was helpful to me at this position? The strong priority (hierarchy) at "when you are forced to lose a piece - try to make as much out of it at possible". In other words - when you are familiar with the term "desperado" you understand my point very well. I rarely miss desperado motifs in my games (especially when I have the time to ponder over the position).

    I will be glad to provide you any kind of tips or pieces of advice. In case you want to know my thinking or analysis (or evaluation) just ask as many questions as you need. This way I could answer these and help looking this position through my eyes and simultaneuosly to think it over with the help of your suggestions (questions). This way is probably the easiest one and may benefit everyone. What do you think about it?

    ReplyDelete
  3. It is a bit difficult to explain. It is not about the position, it is not about a geometrical pattern, it is not about the hidden subtleties of the position. It is about applying "standard logic".

    If your king is in check, you have to move your king, capture the attacker or shield the line of attack. That is standard logic. There is no need to consider other moves. What I do, and you guys do to, I dare to guess, is thinking about a whole lot of things that have nothing to do with the current problem.

    This redundant thinking is consuming lots of time. It looks like having difficulty to orientate, as if you forget to ask "what next?", every time.

    The move Qxd3 looks attractive because it seems to win a rook "without further ado". But since a piece of you is hanging, you loose a piece.

    The "standard logic" says "you can do anything before taking the rook, as long as you maintain the initiative". The free rook doesn't disappear (well, you have to check a zwischenzug, but that is standard logic too). Before taking the rook, it is logical to try to apply some damage control for your hanging piece. By taking on e7 AND threatening the Queen, you get your piece back AND you maintain the initiative. This logic should be obvious, but it is not, observation shows.

    ReplyDelete
  4. I agree with Aox here: if you've gotten to the point where you're just considering the captures you mentioned, it only takes a few seconds to evaluate all of them. You just visualize the move sequences dictated by standard logic in your head. (But what do I know, I'm over 100 points below you...) Are you getting more sophisticated and worrying about additional threats, or is it just difficulty in counting?

    I wonder if doing a whole ton of easy hanging piece / counting puzzles at CT would help for situations like this. Even though they may seem beneath you. - mfardal

    ReplyDelete
    Replies
    1. It's not about getting too sophisticated or worrying about additional threats or difficulty in counting or hanging pieces.

      At is about logic concerning captures and maintaining the initiative. It should be standard logic in my arsenal, but it isn't. So I have to invent the wheel again and that takes time.

      Delete
    2. Then could training only Zwischenzug in salt mines style help? Either at CT, or combining all the tactics books you can find that separate out that category?

      It's perhaps not an exact match to the issue you're raising, but it seems pretty close. And it seems like you must have the necessary knowledge stored in your brain already, it's a question of producing the answer quickly. - mfardal

      Delete
    3. I forgot to mention Capturing Defender and Distraction as CT tags that relate to this issue and could be trained salt mines style.

      Regarding condensing logical reasoning to knowledge, are there any books that do a good job of this for someone at your relatively lofty level? There are a ton of books on tactics for beginners, and some of them do a good job of pointing out the cues that should lead a player to produce a particular tactic (for example the online one by Farnsworth). But you probably have that kind of basic-level knowledge in your brain already.

      The "hook-and-ladder" pattern is a case where the actual tactical sequences aren't necessarily at the simplest level, but there are very simple cues for it. If there are a lot of patterns that have been similarly classified we could all save a lot of time. - mfardal

      Delete
    4. My two month long experiment with the salt mines convinced me of the fact that salt mine style exercises have no positive effect on solving blitz problems at CT, at least not for the problems suited for my level. And probably not below 2200.

      The idea behind your idea is important though.

      You said "I agree with Aox here: if you've gotten to the point where you're just considering the captures you mentioned, it only takes a few seconds to evaluate all of them."

      The point Aox and you speak about is an important point. Once that point is reached, further calculation is handled in no time. The problem I describe occurs before that point. Not beyond. It is a bit difficult to explain, since it is all based on the observations of what is going on in my mind, and if you don't have the same problem at exactly this position, it takes a lot of imagination to understand what I'm talking about. But I am sure you guys are familiar with the same problem, yet only in different positions:

      To suspect problems which aren't really there.

      My knowledge of maintaining the initiative is partly theoretical. That means that to apply it, I must use logical reasoning. Which takes time. What I learned from this position, is that I can ignore certain parts (Qxd3)as long as I can maintain the initiative. The ignored part is going nowhere, in the mean time. This neglecting frees my mind from unnecessary worries, which helps me to get earlier to the point where calculation is done in no time.

      There are many positions where this knowledge can be applied. The only concern is to apply it consciously. There are often parts of the board that can be ignored, as long as you maintain the initiative. The only thing to worry about is the zwischenzug, which might swing the initiative to the other side unexpectedly.

      Delete
  5. PART I:

    I think it is about "heuristics." These are "chunks" of "standard logic" that can be applied (or not, depending on the position). Note that I am using "chunks" in a non-standard way.

    For instance, Aox (in a previous comment) gave the following excellent "heuristics" for examining a tactics problem:

    My general pseudocode up to the critical decision is:
    1. Count material.
    2. Think a moment about opponents last move.
    3. Decide it the puzzle is about Checkmate, gain of material, or pawn promotion.


    As simple as it sounds, often we miss the significance of those three "obvious" steps to get an orientation to the requirements of the position. Surely we just "see" the material balance. Surely we just "know" the opponent's last move. Surely, it is "obvious" as to the goal of a particular problem. Or, to make matters worse, often we don't know the opponent's last move, so we dispense with any consideration of what has changed with the opponent's last move. In essence, we skip blithely over those steps because for most tactics problems, we make UNWARRANTED assumptions. As a consequence, we do NOT "burn" those steps into our "first impressions" scanning process, and then we spend excessive time using T&E trying to get oriented to the position. What we miss is that if we had followed those steps each and every time, we gain immensely over time. Once that process is exercised religiously until it is no longer a conscious process, the subconscious takes over, and gives us a good "first impression" of what needs to be done. That focuses our conscious attention on the critical squares/pieces/relationships AND it takes considerably less time than methodically plodding through a piece-by-piece examination of the entire board.

    My beginning heuristic is "Think wide before thinking deep." This encapsulates Aox's three orientation steps at a slightly more abstract level. Note that it is the same heuristics, encapsulated, to make it easier for ME to remember. Each of Aox's steps are still there, just at a somewhat lower level of abstraction.

    ReplyDelete
    Replies
    1. The "orientation" heuristic is slowly becoming a habit. Only sometimes when my attention is caught by a very promising detail of the position, I forget.

      Delete
  6. PART II:

    I believe there is a critical thinking step in learning that must be done by each and every player. YOU MUST MAKE UP YOUR OWN HEURISTICS BASED ON Y-O-U-R EXPERIENCE! Or, as an alternative, you must make heuristics that you have been provided by others (directly, from books, etc.) into your own heuristics. Regardless of the heuristics you ultimately use, you cannot just copy them from somewhere else and expect the "deep magic" of the subconscious to utilize them. We confuse "familiarity" with practical usefulness.

    When trying to develop heuristics, you have to do the HARD WORK of assimilation. NOT JUST KNOWLEDGE OF THEM, BUT INSTILL THEM, "BURN THEM", DRILL THEM, INTO YOUR SUBCONSCIOUS! Do it consistently, over and over and over. . .then repeat it again and again.

    It is trivial to memorize any set of heuristics. It is NONtrivial to make them part of your subconscious. At first, you have to consciously analyze what the heuristic is trying to tell you to observe. YOU have to do the hard work of finding examples of the heuristic at work, then determine in which situations the heuristic might be applied, and then in which situations the heuristic is NOT applicable. Most of us UNCRITICALLY memorize heuristics (we think), and then either "forget" to apply them, or don't know when to apply them and when NOT to apply them, or mis-apply them.

    As an example, consider Dr. Tarrasch's aphorism, "A Knight on the rim is dim." It is drop-dead simple, and obvious to the point of triviality - or is it? How many times have you looked for specific examples of when that aphorism can be applied, is applicable, and cannot be applied? I'd be willing to bet (based on MY personal experience) that the answer to "How many times?" is close to zero.

    It is a matter of establishing context for the subconscious. If you uncritically "adopt" ANYTHING without serious effort to "make it your own," then it is NOT yours! You will NOT have it ingrained into your subconscious and (for all intents and useful purposes) it is a worthless waste of your time to have "learned" it in the first place!

    ReplyDelete
    Replies
    1. I'm still working to devise the heuristic in the first place. "Condensing logical thinking into knowledge". It's evident I can do that, even without a coach. It just takes a lot of time to investigate and to sort it all out.

      The assimilation of the knowledge is the critical part of this story, as you justly remark. Once, I looked at my next opponent, who was teaching a youngster how to develop during the opening phase of a game. He possessed clearly all the knowledge. When we started to play later on, I played the king's gambit against him. He made all the opening mistakes he had just told his pupil to avoid. That made it very clear: knowledge in itself is nothing, because you still need logical thinking to apply it. And it is easy to forget to apply it for that very reason.

      Delete
  7. I've only just now realized you're back, Temposchlucker - that's awesome! Good to see you posting again :)

    ReplyDelete