Fiddling with Silman
After my previous tournament in December, I wrote a post about in which areas I still suck . Endgames, preparation of the assault and the opening. I pretty much solved my openings. About the preparation of the assault I had a revelation a few weeks before my tournament in July.
Yesterday, I decided that endgames must be my next goal (finally). I lost 2.5 points in the endgame in just 7 rounds. I adopt the suggestion of Robert and will use Silman's endgame course as the base.
When I see an endgame position, usually my mind goes blank. Somehow the triggers don't fire. I had a good experience while exploring an endgame a few months ago. After just a few hours, I discovered already a few concepts that are more generally applicable. Even better, the discovery of the invasion in the endgame was the indirect cause of my revelation of invasion in the preparation of the assault a month later.
So Silman must guarantee that I focus on the absolute minimum I need, and fiddling around with simple positions will hopefully get me some new concepts or even a logical framework. Let me give it a first try.
 |
| Diagram 1. White to move |
Usually, my mind is pretty blank with this type of positions. No triggers are firing, so to speak. I have some vague notion that I must drive the king to the rim and then separate the knight from the king and conquer it.
Silman gives a bit more information: I must mate the king while avoiding to be forked.
So there seems to be two possible goals:
- conquer the knight
- mate the king
 |
| Diagram 2. White to move |
This gives a clue about what is going on. When the white king is in contact with the knight, white can conquer the knight with a simple tactic. For instance:
 |
| Diagram 3. White to move |
The following tactics work:
- 1. Qe5+ chasing the king away from the defense of the knight
- 1. Qa8 pinning the knight
- 1. Qh1 skewering the knight
- 1. Qa4+ chasing the king away
 |
| Diagram 4. Black to move |
The blue squares are forbidden for the black king while the blue circles are forbidden for the knight, because white can decide matters with a duplo attack. Being it a double attack, a pin or a skewer.
Update in blue
ReplyDeleteWe usually think of strategy as “what to do” and tactics as “how to do” that strategy. Which comes first, the chicken or the egg?
ReplyDeleteThinking about your discussion above, I think there is an inversion that takes place between strategy and tactics. You start with an overall “general” strategy: two possible goals at the highest level (too high to actually be useful for determining a move sequence)—(1) conquer the knight and (2) mate the king. You then drop down to the tactics level and determine HOW to go about accomplishing either of those two goals. You assess the potential alternatives, determining what works and what doesn’t work. The strategy for what to actually do emerges from the concrete tactical considerations. You now have an idea (a “plan”) for what to do for the immediate sequence of moves, without necessarily having prior direct knowledge of how to handle this “type” of endgame.
In short, a concrete strategy is an emergent property of available tactics.
That’s not at all what is expected from studying endgames in the “proper” way.
BUMMER!
I just realized something important. The two overall (high level) goals are NOT at the same distance from the ultimate end goal: checkmate. Conquer the knight is a preliminary goal to mate the king. Given that idea, perhaps it is best to first work toward the most immediate overall goal (conquer the knight), while keeping the eyes open for possibilities to skip that intermediate goal in favor of the ultimate goal. If our opponent primarily focuses on retaining his knight, he may overlook the possibility of being checkmated with the knight still on the board.
DeleteNever lose sight of the ultimate goal. Eternal vigilance is required at all times!
The idea is similar to the idea of having two goals/threats and combining them into an (eventual) double attack. Keep the opponent busy with immediate (smaller) threats while aiming for a longer term large threat. Eventually, the sequence of immediate threats will coincide with the long term threat as a direct double attack. At that point, the opponent will have to choose the lesser of two evils because he cannot cover both attacks at the same time.
DeleteWhen training, it is important to keep in mind that we are trying to CREATE patterns in LTM, not RECOGNIZE patterns. We may recognize cues that remind us of some particular (yet vague) pattern that most likely is unnamed. The focus should be on the cues, not on naming the pattern. It really doesn't matter if important patterns are named or not; our brains are quite adept at dealing with unnamed patterns.
Delete"I don't know what the name of this pattern is, but I know it when I SEE it." Paraphrasing Justice Potter Stewart
Bent Larsen showed me how it works. The patterns are already there. The context that lead you to the patterns has to be provided. He showed me a mate with the wrong context: this is a positional problem.
DeleteFor this problem above in the post, I had no context at all. After a few hours fiddling with tablebases, I have created my own context. The next step is to absorb it.
The endgame patterns are mainly tactical. So they are already there. There are a few patterns that are endgame specific. Like promotion, breakthrough, fortress, opposition et cetera. But besides that, context is needed. Creating context will take time.
The rules that Silman provides are horrible. About pawns on the second rank and whether or not the kings are in opposition et cetera. Every rule must be converted into context by fiddling with it. Until it becomes simple and natural and needs no longer words.
DeleteSomehow I was biased that the usual tactics wouldn't apply here, intimidated ("blanked") by the empty board. Why on earth did I think that? No context, no triggers.
ReplyDeleteAfter Silman I want to look at this
ReplyDeleteYou see how inadequate "rules" are for communicating emergent concepts.
ReplyDeleteIt looks somewhat like what MDLM tried to achieve with his microdrills.
ReplyDeleteFrom a trial and error point of view, it is much more attractive to move the queen than to move the king. But limiting the options of the opponent should be the guideline. That's where tactics start to emerge.
ReplyDeleteFiddling around with endgames has brought me two ideas:
ReplyDelete- Invasion
- limit space
Both are applicable to the middlegame too.
Concepts take care for the infamous "transfer" between positions. It is easy to see what to do when the knight is replaced by a bishop. You can simply approach your king by keeping him on the other color.
ReplyDeleteFor a rook it is more difficult, because the rook covers more squares. Thus limiting the possibilities for your king.
Roughly, the evaluation algorithm of Stockfish emerges: increase your own possibilities while decreasing your opponents' and test for tactics.
ReplyDeleteThe third idea: start with the slow movers.
ReplyDeleteThe tablebase indicates that 1.Kg8 and 1.Kg7 are equally fast while 1.Kh7 is 1 ply slower.
ReplyDeleteHypothesis:
Move the white king to the cutting edge of the black knight and the black king. A bit like the Reti manoeuvre.
Question: how would that be when the knight is replaced by a bishop? Maybe the same because the bishop needs to stay close to the king in order to prevent duplo attacks.
It is a bit murky.
Delete8/8/8/4k3/8/3K4/3P4/8 w - - 1 2
ReplyDeleteLook how awkward the description in the book is:
Stepping forward is the most important thing, with Opposition taking a back seat to this. Why? Because White can only win if he gains control of the queening square, while also making sure that his pawn is safe. This takes us to the winning formula for such positions.
RULE: Move as far forward with your King as possible, while always making sure that your pawn is safe!
I don't say it is wrong, it is just an awkward way to communicate. It scares people or make them think it is boring. That's the effect of word salads in general.
It is no criticism on the book. I'm only saying that you need a better way to communicate with your system 1 and system 2. I suggest fiddling around.
Delete