Saturday, June 25, 2016

I'm not seeing it. Yet.

In a previous post about seeing vs calculation, I showed you a few diagrams of puzzles where I find it difficult to see the solution. The following diagram is from the first puzzle of that post.

After 1.Bxd4 Nxc5. White to move
It shows the position after 1.Bxd4 Nxc5
The winning move is 2.Bxg7 and I can see why it is winning. Not only does the move capture the bishop, it also threatens the black queen. It is a discovered attack. In the time that black needs to save his queen, white can bring his bishop into safety.

What I don't see though, is why 2.Bxc5 is not winning. Apparently the move does the same as 2.Bxg7, a discovered attack taking a piece and threatening the black queen.

As usual, this post is not about this position. I use the position as a vehicle to show you something. We look at CCT, which describes what to look for in order of forcingness. That order is handy because it prunes branches with lesser forcing moves, preventing you from calculating unnecessary lines. But now I'm looking from the perspective of what is going on. A position like this is about gaining wood. That means, that capturing pieces is paramount, and you can only postpone a capture if you have a very good reason for it. Like a check or a threat. If you postpone a capture, your opponent must too, otherwise you are not going to gain wood.

Aox gave a nice rule of thumb for positions with mutual captures: "A sequence of takes where you take first and last is usually benefitting". Whether it is beneficial or not, is based on the value of the pieces that were taken. I discovered this rule in 2007 or so, and it is very useful. But now I see that this rule needs extension, for whenever one of the parties postpones a capture by flicking in a check or a threat. I already devised this rule: "In case a capture is met by a check or threat in stead of a recapture, you win a piece when you can capture the threatening piece.". But in this case, there is more going on.

The puzzle at hand rates 1952. Which is kind of reassuring. Albeit I'm a patzer, at least I'm not alone. Let there be no misunderstanding. I can calculate why 2.Bxc5 is not winning. But that is not good enough. The average solution time is 57 seconds. Calculating a solution is both time consuming and error prone. To solve this position within 57 seconds correctly, I need to see the main ideas. I can never calculate this correctly within 57 seconds. To speed up calculation is no option, since that introduces errors, as we found out.

I feel that the only reason why I am not a 1950 rated tactician, is because I cannot see the solutions in this kind of positions.

There are two more things going on in this position. After 2.Bxc5 Rxc5 it appears that white's rook on c1 is overworked. It cannot defend both the bishop on c5 and the rook on d1. In itself that would be no reason for the move 2.Bxc5 to be not winning a piece. [UPDATE] After 2. ... Rxc5 3.Rxd8 Rxc1+ white still would win, now the queen. If it weren't for the check, which gives black the opportunity to recapture the rook on d8. I hope you can still follow me.

I hope that you can see what I'm after. I want to tell the narrative of this position in terms of captures and postponement of captures. I want to devise some rules that govern the story of the initiative. And what is even more important, I want to be able to see these rules at work. My posts about the initiative generated remarkably little comment. But the story of a combination is told by the vicissitudes of the initiative. I want to see it at work.


  1. [3qr1k1/2rnppbp/p1n3p1/2N3P1/1p3P1Q/1P2B3/P4PBP/2RR2K1 b - - 0 0]

    What I love the most is your questions, uncertainity, curiosty and exploring all the things you are really interested at!

    I hope my small contribution gives you some light at the end of the tunnel my friend.

    My explanation goes into 3 ways.

    1) STATISTICAL explanation (discussion)

    a) Rating Blitz:1952
    Average time (Blitz):00:57 s.
    Number of attempts (Blitz):189
    Correct solutions (Blitz):54.5%

    In my opinion it is not a task properly estimated. It could have been at least 2050-2100. And not because it requires a deep look into the variations, but a very good EVALAUTION. I will discuss it later.

    Average time is about just one minute! It means blitz players had to think deeper. There are only 200 attempts and it is a small number to make (strong) conclusions. Anyway only HALF of the players solved it correctly! I was wrong too (even if I could solve rest of the puzzles correct!).

    b) Rating Classical:1755.9
    Average time (Blitz):04:36 s.
    Number of attempts (Blitz):849
    Correct solutions (Blitz):54.06%

    This time rating is set only to 1755! Are you kidding me? It is understimated as hell! Anything (much) below 2000 is not a good estimatation as this task is too hard (I will show you soon).

    Average time is close to 5 minutes! How can you think over the puzzle that is 2-3 moves deep for such a long time? Yes, you can! Especially as you COMPARE other variations (not just play what looks good!). There are close to 900 attempts and we can make (pretty strong) reasonable conclusions. Please take a look: EVEN after 5 minutes thinking only HALF of the the players solved it correctly! Did you notice how much close this score is to the blitz solvers?

    We can state that blitzers are probably (much) stronger at fast (and correct) solving and that's why they needed only 1 minute. Slow thinkers (most players under 2000) needed 5 times more. Even though it did not improve the correct solution number (% of correct solutions). Why? Can you guess it?

  2. 2) LOGICAL explanation (discussion)

    Before I told you this task is too hard. Now I am going to explain it. First look at the variation I had to check out before believing this task was NO easy at all!

    [FEN "3qr1k1/2rnppbp/p1n3p1/2N3P1/1p3P1Q/1P2B3/P4PBP/2RR2K1 b - - 0 1"]

    1. ... Nd4 2. Bxd4 Nxc5 3. Bxc5 Rxc5 4. Rxd8 Rxc1+ 5. Bf1 Rxd8 6. Kg2 Rd2 7.
    f5 Bd4 8. fxg6 hxg6 9. Bxa6 Bxf2 10. Qxf2 Rcc2 11. Qxd2 Rxd2+ 12. Kf3 Rxa2 The final position is at least -7,2 (extremally easy to win to all the players). Especially as there is not any counterplay.

    I gave the engine this position to crack. I was really curious what is the evaluation after ...4. Rxd8 (?) Rxc1+ 5. Bf1 Rxd8. Take notice all of the moves are forcing - you do not have to check anything else - just evaluate.

    After a few minutes (I have a slow machine) of thinking - at the 26-28 deep level - it shows -5,0 to -5,5. It means it HAS TO BE lost completely (especially as it is not a fortress neither the advantage you cannot finally transform into mate).

    Now white has to get out of the pin as doubling rooks is a killer.

    The rest is a very easy to understand if you KNEW it was that bad for white.
    6. Kg2 Rd2 7. f5 Bd4 8. fxg6 hxg6 9. Bc4 Rxf2+ 10. Qxf2 Bxf2 (and 11. Kxf2 Rc2+ 12. Ke3 Rxa2... and the rest is obvious).

    What is the conclusion? Black was winning (after ...4. Rxd8) because it has no time to make a counterplay (counterattack). The King was cut off at the two last ranks and White could not defend these ranks nor key square f2 (due to Rd2+Bd4 attacking twice)

    3) MY own explanation (the one I wanted to share, but decided I expand it with the previous points).

    If you are at least 2100 player with tactical strenght 2200 or better - you can evaluate such positions without much problems. Why? Because you HAD HAD TO analyse such positions before you ACHIEVED a rating (at least) 2200 or more. And the secret does not cover pure tactics, but it includes positional pressure (lack of necessary counterplay for White). Such elements are the part of chess education from the level of 2000 (i.e. after the student achieves this rating he has to learn these elements much deeper).

    And that's the reason this task should be rated at least 2050-2100. It is not a pure tactical puzzle, but a "semi-positional one".

    What is suspicious? The value of the number after the White's mistake: 2...Nxc5 it is +4 for White (24 deep) and after White played 3.Bxc5 it drops up to -0,7 (24 deep). But after the final blunder it goes from -0,7 (3...Rxc5) to
    -4,2 (26 deep) - when you play RxQ (4.Rxd8??).

    What I thought previously? It is a normal mistake and the task was easy. You helped me to change my mind! This task was really hard one! Especially as my mind wanted to refute it with the help of (after 3.Bxc5) 3...Qc8! and if you play 4.Bh3 I will answer 4...e6. I estimated the Bishop is lost because you cannot defend it good enough (4.Rd5 e6!). The only problem was the hidden (?!) defence - Be3 and everything is fine (White is winning). I simply overlooked it!

    BTW. It is a very good idea to provide FEN of the positions you analyse, because this way we can set it up immediately. It is especially helpful if we can analyse it with engines (and see the refutations).

    1. You are making matters way too difficult. It is not about this position, remember? It is about what I want to learn from this position. Since my gut feeling tells me there is something important to learn here. I'm not interested in the positional aspects of the position, but in the difficulty my mind has to see in what situations I win a piece and when do I not. I can only educate my mind, when I find out what this difficulty is.

      The first question is: why does 2.Bxg7 win a piece, and why does 2.Bxc5 not?
      In other words: why does the discovered attack works in the first case and not in the latter?
      The answer is: it works in the latter case too.

      1.Bxd4 Nxc5 2.Bxc5 Rxc5 3.Rxd8 wins the black queen. Compare to:

      1.Bxd4 Nxc5 2.Bxg7 Kxg7 3.Rxd8 wins the black queen.

      The difference is, that in the case of 1.Bxd4 Nxc5 2.Bxc5 Rxc5 3.Rxd8, black has the follow up 3. ... Rxc1+

      If the king was on h2 in stead of g1, there would be no check, and white would continue with a 4 point advantage. Even better, since he doesn't have to parry the check, he has time to take the rook on e8 either. It is the check of 3. ... Rxc1+ that turns the tables.

      This leaves me with the question: what exactly is the effect of rook c1 being overworked? I can calculate it, but I don't see it. And I know it should be seeable.

  3. BTW. You can see (analyse) White's position after 5...Rxd8. The WSB is pinned, Queen is blocked and you cannot prevent doubling rooks at the first rank or going Rd2 with Bc5 (forcing the exchange Q for R+B). It is a valuable position to show "the power of making threats you have to deal with (or you get mated) :). Anyway it is not a pure tactical consideration, but a positional one (at least it is my opinion).

  4. I updated the text (in blue) since the original analysis was nonsense. It is still pretty unreadable though, which is a clear indicator I don't quite have a clear picture what makes the combination tick. It's now gibberish in stead of nonsense. I'm working on it.

  5. ehhh,
    the position is silly easy
    After 1.Bxg7 the black queen is hanging so black has not time to recapture

    On key to exchange sequences is: reduce your opponents amount of possible takes and improve your amount of possible takes.
    Here by Bxg7 Black cant take anything because they have to take care of their Queen somehow

    By the way seeing
    The "mate in 1" / "Board vision" experiment did show that an improvement in vision is ( seemingly ) only possible in extreme simple tasks . A tactic pattern more complicated than M1 should be completly unimprovable

    1. Yes, I feel silly as a Brexit voter after the discovery that demagogues sometimes lie :D

    2. oops i did not read your post correct, you want to know why 2.Bxc5 is wrong, its because after 2...Rxc5 the queen is not realy hanging, the rook eyes to the white backrank so black gets 2 rooks for the queen AND the piece back

    3. "silly as a Brexit voter after the discovery that demagogues sometimes lie"

      realy silly: either the demagogues for "in" did lie OR the demagogues for "out" did lie, they cant be both right.. i think....

      But it will be very interesting to see wich side was right IN REAL and what will happen next.
      I would think that the EU might get enough push by the brexit to become more transparent, democratic and subsidiarity what is desperate necessary in my eyes.
      And maybe the other 27 nation will now be without any mercy to the UK to make them suffer as much as possible to keep all others in the boat? maybe Shottland will leave the UK which will become the the NotUnitedKingdom (NUK) or LittleBritan (LB)

      We will see, very very exciting!

      AAAnd next? The nexit?

    4. To say it different:after 2...Rxc5 3.RxQd8 the situation is just the other way around, now black can take the rook with check and white has no time to take care of the hanging Rd8. In both cases the key is to force the opponent to waste a tempo instead to continue to take. Either by check to the queen or by check to the king. The line is just deeper and the visualisation therefore less easy.

      By the way, its possible that the high speed for this puzzle is a effect by the "not absolutaly precise" ratingsystem of ct. The "average time" is calculated only by the average times of the correct solver of the problem.

    5. So far we have seen two referenda in Holland. In both cases, the referendum was hijacked by people with no interest in the subject of the referendum whatsoever. They abused it for their own political agenda. The same is true for Cameron, who abused this referendum to settle a conflict in his own party. As long as it is allowed to abuse referenda, it is an ugly undemocratic tool.

      The "free beer for everybody party" AKA the populist party here would love to misuse a referendum for their own political agenda and organize a "Nexit". At the moment they are not given the opportunity. We are living in interesting times. . .

  6. "What I don't see though, is why 2.Bxc5 is not winning.

    To answer your question definitively requires calculation and analysis of the various piece relationships. You know you can do that, but your preferred mode is to avoid that (unless or until it is absolutely necessary) and just "see." Consequently, I'll address what I "saw" rather than what I calculated afterwards.

    From my perch on the vulture's back, I have this gut feeling: White is trying to attack on the Kingside. Without any calculation, my instinct is to eliminate the formidable defensive piece, the Black Bishop on g7, rather than the relative tame Black Knight on c5. Otherwise, there is no possible scenario in which White can penetrate into the Black Kingside. There was an immediate "sight" of the discovered attack on the Black Queen after either 2. Bxg7 or 2. Bxc5, but (again) it was a gut feeling that there might be some adverse tactics for Black if I lose sight of the primary objective, so I didn't consider 2. Bxc5 until much later - in concrete analysis.

    Maybe it is an illusion, but it seems to work for ME in ambiguous situations like this one. One of the hardest things for ME to do is to "trust" my intuition; I'd much prefer to confirm every possibility with concrete analysis. After reading GM Rowson's books, I've really tried to rely more on my intuition as a first approximation. At the same time, the Ronald Reagan aphorism "Trust - but verify" always comes to mind. It seems to work best for ME when I can get a good clear grasp of what the subconscious is trying to tell me BEFORE checking it out using my conscious mind.

    I have no idea if this is the idea you are seeking. These are not the droids you are looking for came unbidden into my mind.

  7. Aox said "To say it different:after 2...Rxc5 3.RxQd8 the situation is just the other way around, now black can take the rook with check and white has no time to take care of the hanging Rd8. In both cases the key is to force the opponent to waste a tempo instead to continue to take. Either by check to the queen or by check to the king."

    What I want to learn is to keep track of the initiative. CCT is about the initiative and (not) wasting tempo's. I think that understanding of what happens with the initiative, or the tempo's, is paramount when you are trying to develop combinations. Somehow between 2.Bxc5 Rxc5 and 3. ... Rxc1+, it is somewhat vague to me what happened with the initiative. The move 2.Bxc5 is weak, since the bishop looks protected twice on c5, but the rook at c1 is overworked, so it is deceptive. But it does not exactly hands over the initiative to black. The transition of the initiative is caused by the check on c1. I don't see what happens to the initiative during move 2.

  8. I did never like the concept of intiative, maybe this is a reason why i dont get better?
    In my eyes there are ( potential ) weaknesses and there are methods to make use of them. When you have both and your opponent not ( that much ) then this is called "having the initiative".

    In this puzzle there are the weaknesses Rc1,Rd1,Rc7,Qd8 (and many others) which are attacked once and defended once. Well.. theoretically is the Qd8 attacked twice but the method to make use of this fact is so far..

    Now Bxc5 gives black more access to the weakness at c1.

    I dont know if you can translate that back to the initiative-speach

  9. I note for the record that Stockfish evaluates certain variations (after a couple of days running nonstop):

    0. … Nd4 1. Bxd4

    1. ... Bxd4 2. Rxd4 (+- [4.38])

    1. ... Nxc5 2. Bxg7 (+- [5.63])

    1. ... Nxc5 2. Bxc5 Rxc5 (∓ [-0.90])

    Surely if the evaluation changes from White advantage (+5.63) to Black advantage (-0.90) based on capturing on g7 or c5, then it cannot be because of some subtle positional factor which we amateurs fail to grasp. Positional factors rarely amount to more than the equivalent of a Pawn. As Tempo has rightly pointed out, the difference between the two moves is the initiative. We all acknowledge that as "fact," but we seem to be lost at "see" (pun intended).

    Paraphrasing Justice Potter: "I don't know what the initiative is, but I know it when I see it.

  10. Robert

    We can say that intercepting the initiative is due to the STRONGER treat(s). Take notice how all of the Black pieces GAIN the time to attack vulnerable (sensitive) points at White's camp. If there would not be the threat against the King - in this case mating ones - you would not have seen such an evaluation. Is the same if you could capture some of my pawns and pieces, but in the final variation you would not be able to defend against a mating threat (let's say - mate in 4 or 6).

    Recognizing the initiave is not that hard. However what is hard is the FINAL of the variation as a concrete position (who is better and why). Take notice some amateurs play in a very aggresive way and they look like having the initiative ALL the time. When you look closer you will see the ONLY way to stop their aggresive moves is to play the moves that evaluate better for us. I have played against amataueurs (1200-1700) at least about 20-30K games (fast time control like 5, 10 or 15 mins per game). I had to learn how to intercept the initiative as even IF I had know they play incorrect - I could not stop them (their aggresive moves) against my army. After I had learnt how to do it - I started winning against them in 90% of our games (previously only at 15-20%!).

    And I see the concept of initiative is much better to discuss and analyse the positions from the positional point of view. If there are no CLEAR forcing moves (like chess, captures and threats) you may be lost in the arrays of variations - and often you will not be able to recognize what A is better than B (without extreme differences and engine's help).

    The best example of showing what initiative means is the gambit play. You can check out the position is (close to) equal - unless the side that sacrificed the material creates threats or regains the material. Otherwise if you slow down with doing these things - you are probably the side who obtained the lost position. And the most impressive sign of the full compensation of lost material (let's say being a Queen down) due to the gambit play - is the concept of perpetual (check or move repetition). The side that stands much better (in a material) cannot avoid move repetition - no matter if it is perpetual check (only the weaker side can voluntarily chose to stop doing this) or perpetual repetition (I attack your queen with N and you have the only squares and you are forced to repeat the moves - otherwise I will capture your Queen and the final position will be much better to me).

    Any questions or comments?

  11. See my comments on the blog post on Monday, June 27, 2016 "Elaborating on the initiative."

    In the broadest sense, the initiative is about controlling the tempo of play via threats. That is the point of GM Beim's definition of the initiative:

    "Possession of the initiative means being able to create threats faster than the opponent, and the aim of developing your initiative is ideally to use your threats to forestall the opponent's activities, defensive as well as aggressive. From this it follows that fighting for the initiative always means trying to be ahead in a race! Thus we can see already the concept of the initiative is inextricably linked to that of time and speed."

    In the theory of Steinitz, the initiative resides ultimately with the player who has an UNCOMPENSATED advantage. It does not matter what form that uncompensated advantage takes (whether it is in King safety, Material, Pawn structure, Mobility, or Combinations). Thus, it is possible to have the (long-term) initiative even while under heavy attack or in the middle of a mind-boggling tactical mêlée.

    The complexity of chess resides (among other things) in trying to evaluate relative gains/losses in different forms, not in the same form. For instance, if I have more material than the opponent and the opponent has no compensation for that material, then I hold an uncompensated advantage and I have the initiative. Or, if I have a preponderance of attackers on a particular square AND I can capture the piece on that square without (eventual) material loss, this is also relatively easy to evaluate by counting (adding/subtracting up the material gains/losses) and I have the initiative. On the other hand, when there is a tradeoff, this becomes several orders of magnitude more difficult to evaluate! A gain of time at a cost of material is a typical tradeoff. If we are assuming that the tradeoff is balanced (no uncompensated advantage to either player), then according to Dr. Lasker, balanced positions should lead always to balanced positions. The balance can be static or dynamic or a combination of both. The initiative belongs to the one who can initiate threats that force the other player to respond to them. That does not mean that the player with the initiative always has an uncompensated advantage. The initiative is an advantage ceteris paribus; it may be compensated for by any number of other things.

    I think we are looking here for a restricted view of the initiative in terms of how to "see" the components of tactics without having to engage in deep calculation. But, I may have missed the point entirely. . .sigh.

    1. I think we are looking here for a restricted view of the initiative in terms of how to "see" the components of tactics without having to engage in deep calculation. But, I may have missed the point entirely. . .sigh.

      The goal I was after, was to prune the tree of analysis, by adding a hierarchy to the moves. (Heisman seems to suggest he works that way.) Consider the more forcing moves before the less forcing moves. That is a way of forward thinking. The problem with forward thinking, is that you have no idea where you are heading. Any pruning would be welcome. But since it is entirely possible that it are not the 8 most forcing moves you need to consider but that the 9th is the right one, a lot of redundancy is introduced.

      I concluded that backwards thinking is better, and that we should learn to see the end idea of the combination first, before we even start thinking about the moves that lead to the execution of the idea.

      We might well consider moves to probe the position, but just as a method to help the mind to recognize the idea.

    2. I found that forward probing, by means of considering the most forcing moves first, to be not effective in most positions. It didn't lead me to the main idea behind the combination. And when it did, I often had to work myself trough considering moves that didn't work first.

  12. Hmmm. . . Th gist of your last two posts seems to fall in the general direction described by Dr. Lasker under "The Plan" [Fourth Book: Position Play]:

    The methods followed in the analysis of a given position by combination and by the creation of plans are determined by the direction of the underlying thought. The combination-player thinks forward: he starts from the given position and tries the forceful moves in his mind; the position-player thinks backward: he conceives a position to be arrived at and works toward that position of which he is more conscious than the one on the board. He sees successive stages of the position aimed at and he visualizes the stage in a reverse order. If one position, according to his plan, is to follow another he sees the one that is to follow first and he deduces, as it were, the anterior position from it.

    Skipping a considerable amount of justification for planning:

    ". . .the position on the board must show a sign, a characteristic moment, which tells us whic: "h plan to follow and thus relieves us of the necessity of searching through an immense mass of variations."

    I take it that this is the conclusion you have arrived at: ". . . we should learn to see the end idea of the combination first, before we even start thinking about the moves that lead to the execution of the idea."

    If we can catalog the "clues" preceding and directing a particular type of combination (of motifs and themes), I think it makes it much easier to "see" what must be done (the individual moves and the order to play them) to realize the "idea" of the combination.

    1. I have had a good look at the puzzles in my database, while asking myself "how could I arrive at the solution?". The simple ones can be found by forward thinking. But at the moment there is a slight bit of complexity, the forward method is too time consuming, or will even not lead to the answer. The backwards method is the only logical method. As found by excluding all other possibilities. We must start with defining the targets. The weaknesses, as Aox put it.