Wednesday, February 11, 2015

Calculating Forcing Moves: Defensive Idea 1, Part 1 (Capture the Attacker)

When your opponent creates a (real) threat, or you are considering creating one against your opponent, there are five defensive ideas to consider.  This post will explore the first idea starting out with a simple position, and I'll add more complex positions over time.

The general process is to use the ideas to identify specific candidate moves.  Next, calculate each move to the end and evaluate the final positions (material and activity).  As you progress through the candidates, keep in mind your "king of the hill" (KOTH) candidate move to help you pick the best at the end.

The first defensive idea is to do something to the piece creating the threat: capture, pin, or deflect it.  Of course when you are in check you have no time to pin or deflect the piece, but any other kind of threat gives you more time and therefore more options.

Black just played Bd8 to block white's d-pawn and attack the queen.  White to move.

In this postion white is up one pawn (a passed pawn), but black is threatening to capture white's queen, giving check and attacking white's h4 bishop.  Let's review the five ideas to identify actual candidate moves, and then calculate each candidate and pick the best one.

Idea 1 (capture, pin, deflect):  Qxd8; can't deflect or pin the bishop to anything of equal/greater value (the queen or king).
Idea 2 (block):  none
Idea 3 (move):  Qf5, Qe5, Qd5, Qf4, Qe3, Qd2
Idea 4 (defend):  Does not make sense to defend the queen here.
Idea 5 (counterattack with an equal or greater threat to WIN, DRAW, DEFEND): an equal threat would be to attack black's queen; a greater threat would be to attack black's king.  Rd6 attacks the queen; h6 might be interesting since it threatens a capture-check.

That gives us nine candidate moves to calculate.  Let's start with white's most forcing move first, and look for black's possible refutations: 1. Qxd8:

White is now up a bishop and a pawn, and is threatening to win black's rook with 2. Qxg8+ Kxg8 (only) 3. d8=Q+.

1...Rxd8 2. Bxd8.  Material is now equal (R/B/PvsQ), however, white's threat is to promote his pawn to a queen in two moves, leaving white up an entire R/B.  Black has no way of stopping the pawn from promoting without giving up his own queen.  Black should reject 1...Rxd8.

A quick search for other candidate moves for black yields nothing:  1...Qh6+ allows the simple block 2. Qg5.  1...Qa8 (defending the rook) allows the simple 2. Qxa8 (any) 3. d8=Q.  1...Qe6 (again defending the rook) still allows 2. Qxf8+ (any) 3. d8=Q.

This makes 1. Qxd8 our KOTH.  Now, in an actual game with time controls white can stop calculating at this point because our KOTH forced the win of a lot of material:  capturing the attacker with 1. Qxd8 simply wins a bishop (at least) by complete force.  Black has no moves to refute it.

However, for the sake of this exercise let's look at white's other defensive candidate moves.

There are six square the queen can move to in order to escape the attack.  However, the bishop on h4 is hanging so any move must either defend the bishop or create a greater threat.  The only move that comes close to that is 1. Qd5 (we can reject the other "move" candidates):

White's threat is to capture black's queen.  In a more complicated position, we could also apply the same five defensive ideas here to consider how black can best respond to white's threat.  But here black's best defensive idea is also to capture the piece, which refutes white's move:

1...Qxd5 2. Rxd5 Bxh4.  Now black has two defenders on d8, and can rather simply move his king over, capture the passed pawn, and comfortably be a full bishop ahead.  Since this evaluation is much worse than our KOTH, we can reject 1. Qd5.

There are two counterattacking moves that appear to create an equal or greater threat.  Let's start with 1. Rd6:

1...Bxg5+ 2. Bxg5 Qxd6.  Completely refutes 1. Rd6, so no need to look any further.  Since this evaluation is much worse than our current KOTH, we can reject 1. Rd6.

The second counterattacking idea 1. h6:

Is an interesting idea but the "threat" to play the capture-check on g7 is too slow.  White's biggest threat is still to capture on d8.  So 1. h6 simply allows black to avoid losing his bishop with the defensive idea of pinning the attacker (white's queen) to something of greater value (the king) with:

1...Qxh6 2. Qxh6 gxh6 and now white has lost a pawn (so equal material), and can only initiate equal trades.  Now white's d6 pawn will likely be captured.  So black need not look any further at defensive replies to h6 since he is now at least equal.  Since this evaluation is much worse than white's current KOTH, we can reject 1. h6.

That's it.  The goal is to do this calculation in your head in seconds.  Many of you probably found Qxd8 in a couple seconds, and did not even consider other replies for white.  Or maybe you considered other replies for white first, and then looked at Qxd8 out of desperation.  Future positions will be more challenging, I promise!



    (Quote) "Future positions will be more challenging, I promise!". I am not sure if you REALLY know how challening this example is! It is one of the richest (in ideas) example of "simple" chess position as I have EVER seen!

    This way you helped me to uncork all the questions inside me (related to simple tactics) with just one example! Try to imagine how grateful would I be if you provide more of these! Currently I am trying to refute most of your comments. Believe me, it is really hard task to me, but extremally important. I am still trying to look at the position to the core and understand its internal connections.

    There are many questions that come to my mind. Here are just a few (you can answer them anytime you wish):
    1. How can you know WHEN you should stop looking for a refutation?
    2. How do you define "a refutation"?
    3. What moves (from all legal ones) you do NOT consider (and why?)
    4. How can you define "a threat" (the sound and unsound one)?
    5. What is the method to find and (in a later stage) select all the candidate moves?
    6. Is there any specific order you consider the moves and calculations?
    7. What elements do you take into consideration when assessing the position?
    8. How can you define priorities? (when looking for threats, defence, refutations)
    9. What is a "force move"?
    10. Should we always start (looking for) at force move(s)?

    I think for now it may be enough. I do not want to overhelm with my questions. Take notice I ask all the questions (even so called "stupid ones"). If there are unappropriate ones, let me know. I do it on purpose as I am looking for the deepest side (core) of your method (of looking for tactics and calculations) - that's why I ask "strange" questions. Sorry in advance - I hope you know what I mean.

    Looking forward to your reply! :)

  2. Hi Tomasz -- thanks again for your questions, and for your interest in my blog! There's no such thing as a stupid question, only a stupid answer. I'll try to answer each question...and here's hoping none of my answers are stupid!!

    Refutations (q1 and q2). A refutation is "proof" by your opponent that your candidate move is bad. Think of a forcing variation (or chess in general) as an ongoing argument between you and your opponent, and the specific language used are concrete variations. So in the example above white is saying to black "1. Qxd8 is a winning move", and black tries to prove white wrong with a specific variation. Unfortunately for black, white's "argument" is better as proven by the specific variations. However, white's other candidate moves are another story. For example with the counterattack candidate move 1. Rd6, white is saying to black "you take my queen, and I take yours." But unfortunately for white, when black takes the queen it comes with check, after which black simply takes the rook. So black's argument (if 1. Rd6 Bxg5+ 3. Bxg5 Qxd6) is the winning argument, and so black has refuted white's counterattack candidate.

    You need to calculate deep enough that you can convince yourself that the forced variation is over ("quiescence") -- there are no more moves that immediately force a change in the material (and activity) balance. So in the 1. Rd6 Bxg5+ 3. Bxg5 Qxd6, white only has a bishop left on board (vs a queen and rook), and white has no immediate mate. Even though the position is not truly quiet because white has a capture (Bxd8), it would be a waste of one's time and energy to keep calculating the variation because the material balance is so clearly in black's favor.

    This is a simple position though. Other positions are not so clearly imbalanced, like if you are giving up material for activity (a sacrifice), and the variations can become mind-boggling and it is much harder to develop a concrete evaluation. Even in those positions at some point you have to ask yourself (1) what exactly do I get/lose in that forced variation, and (2) do I like it? Tal and other attacking players search for forced draws as an "out" when looking at piece sacrifices, and in many games by top-level attacking players it can be extremely difficult to evaluate what is going on. But most positions, especially in beginner games, are not that complicated and the evaluations are in fact very black and white. The only way to improve your ability to know how deep to look in a line is to practice doing it!! And the general calculation principles I am proposing here will help you clearly decide what to do in simpler positions, and can be used to help navigate (but maybe not solve) the more complicated positions that arise in games of Tal, Shirov, etc.

    The general idea with looking for "refutations first" is that it is more efficient to try to "prove" that your candidate is wrong. That holds true in any kind of calculation you do. You do not want to fall in the trap of calculating bad moves by your opponent, like this logic: "if black moves here, I have a mate in 1, so I should go ahead an play my move and hope he moves his king there so that I can win." If you were black, would you make that move?

  3. Evaluation (q7). The basic elements are king position, material, activity, and pawn structure. I recommend beginners start out with material evaluations. Mentally take off the board all material that is the same, and compare what is left. So in the above position we can eliminate one queen for each side, one rook for each side, one bishop for each side, and five pawns for each side. All that is left is one white pawn. You should always do this as the very first step when you look at any position. Eventually it will become natural and will only take a split-second.

    Threats and Force (q4 and q9). A threat is anything that could change the material/activity situation on the board like a check or a capture, or a threat to do one of those things. But threat can also include "positional" things like preventing castling, creating a weak pawn, controlling a file/diagonal, etc. As you improve, you will learn how to see more threats, but at first you should focus on mates and captures. "Force" is directly related to threat. The bigger the threat, the more you are "forced" to address the threat AND the fewer choices you actually have. For example a double-check is the most forcing move possible because it only allows one type of defensive idea, which is to move the king (other than checkmate, which allows zero replies). A threat to win a queen is less forcing than a double-check, but it is more forcing than a threat to capture a pawn. I sometimes use the term "forced" move and "only" move in the variations. With "only" I mean it is the player's only legal move on the board. By "forced" I mean the player might have other moves, but they are all worse than this particular move (and you prove that by calculating the other moves).

    Identifying candidates (q3 and q5). I propose that there are two types of candidate moves -- offensive, and defensive. This blog series has focused exclusively on defensive candidate ideas, and I propose that there are only ever five defensive ideas in response to any threat (but only three for a check, of course). "Offensive" moves are harder to find and require study of tactics. Some moves don't threaten anything (like say castling), and aren't considered "offensive" or "defensive". In terms of what NOT to consider...when looking at defensive candidate moves, I propose that you ONLY need to consider the five ideas. Any other move is a waste of time since it does not address the threat...unless of course you have decided to sacrifice material, which is fine as long as it is done intentionally! When looking at offensive ideas, it is much harder to say what NOT to look at. I have a blog post on ways to find moves that creat threats that could give you some suggestions, in addition to studying tactics.

    Prioritizing calculations (q6, q8, q10). Yes, you should always start by calculating forcing moves first. Not all positions have forcing moves, but when there are, you need to calculate them a bit individually to decide what you think (evaluate) the forcing move a threat? If so, what exactly is the threat (or is it not a threat)? If it is a threat, how big of a threat is it? Be specific and detailed. In general you should look at the most forcing offensive moves first (those that allow the fewest replies), because (1) it could lead to a forced mate or forced win of material, and (2) if not, you learn a lot about the position (key defenders, key weaknesses, etc). You should also calculate defensive replies based on force as well, because that is often the fastest path to finding a refutation.

    I hope that answers your questions!!

  4. Thank you very much for taking your time for answering my questions. Now my understanding has much less gaps :). Good work! Well done!