Learn from your mistakes
Everybody who fails to grow despite investing considerable effort fails to learn from his mistakes.
What is a mistake?
That is highly personal. If you solve problems at CT, for instance, then there are problems which you solve a tempo, problems that cost you an unreasonable amount of time, and problems that you fail to solve by coming up with the wrong move. If you compare yourself with somebody of the same rating, you will see that he has difficulty with quite different problems.
When you are clueless for a long time or even reach the wrong conclusion, we have a mistake at hand. Take for instance the following diagram:
I deviate a little from my own definition, since I found the solution just in time, yet I have the feeling I should see what is going on a tempo, and I do not. But that doesn't matter for the idea I try to explain. I just feel that these kind of positions take me way too much time. The reason for that is I have to calculate this position move by move, while I don't have an overall picture of what is going on.
The combinations in the position are pretty straightforward. If I want to learn from my mistakes, I have to find a way to defy step by step calculations, in favour for seeing what is going on in one go.
Verbal thinking can be of help here, but that is definitely not the solution. How can I simplify the position in my mind so that I'm no longer confused by my own thoughts about the position? How can I prevent that such such position takes me so much time everytime? How do I learn from my mistakes?
What is a mistake?
That is highly personal. If you solve problems at CT, for instance, then there are problems which you solve a tempo, problems that cost you an unreasonable amount of time, and problems that you fail to solve by coming up with the wrong move. If you compare yourself with somebody of the same rating, you will see that he has difficulty with quite different problems.
When you are clueless for a long time or even reach the wrong conclusion, we have a mistake at hand. Take for instance the following diagram:
I deviate a little from my own definition, since I found the solution just in time, yet I have the feeling I should see what is going on a tempo, and I do not. But that doesn't matter for the idea I try to explain. I just feel that these kind of positions take me way too much time. The reason for that is I have to calculate this position move by move, while I don't have an overall picture of what is going on.
The combinations in the position are pretty straightforward. If I want to learn from my mistakes, I have to find a way to defy step by step calculations, in favour for seeing what is going on in one go.
Verbal thinking can be of help here, but that is definitely not the solution. How can I simplify the position in my mind so that I'm no longer confused by my own thoughts about the position? How can I prevent that such such position takes me so much time everytime? How do I learn from my mistakes?
Hurray Tempo is back :)
ReplyDeleteAnd for your problem .. maybe you calculate wrong?
-Material balance
-Find all weaknesses ( in tactics find all HE's ( = tactical weaknesses ) see http://www.neoneuro.com/downloads/chuzhakinssystem.pdf !!!!! skip the first 7 pages )
-Consequences of last move of opponent
-Look for Candidate moves ( especially if you are still clueless: look at every! possible move starting with queen forward move ...rook ... knight backward.. King move)
...
Chessability is not only about chess itself, its about methods how to crack problems too
Actually I had no intention to come back. But sometimes (or maybe always) life lives itself. Thx for the link to the chuzhakins system, I'm looking into it.
ReplyDeleteI had a look for about an hour at the position above, asking myself, "what obscures my mind in this position, what is needed to see what is going on in the position faster?" The themes are welknown and straightforward. After looking at each piece, and asking myself "what is it doing?", the position became clearer and clearer. Hopefully that will help me to avoid the same tedious work in the future.
I make a database in ANKI (hattip to Aox) with 50 diagrams with mistakes/clueless/timeconsuming positions. I will investigate them for an hour or how long is needed to clarify them. Then I will repeat them with ANKI, so I will not forget.
Well..
ReplyDeletethere are say 100? things to see in a position, piece A is protected by piece B, piece C pinned to the King, protected by a rook which is protected by an other rook....
Now you need to make the right combination of the involved pieces and the right combination of possible moves.. voila
This link above can help ( in middlegame-positions ) to reduce the amount of pieces which need to be considered. But at the end.. you need a large STM ( 64Bit CPU vs 8 Bit CPU)and/or lots of Chunks and/or memory ( BUS/Memory speed) and a good thinking process( Algorithm )
But at the end.. you need a large STM ( 64Bit CPU vs 8 Bit CPU)and/or lots of Chunks and/or memory ( BUS/Memory speed) and a good thinking process( Algorithm )
ReplyDeleteWell.. yes and no. There are problems with a high rating that I see as simple. And there are problems with a low rating that confuse me. After an hour of study, I see the confusing problem as simple too. For a confusing problem, even the highest bitrates etc. are not often enough, while anyone can solve a problem that is totally clear in seconds. No matter the rating of the problem.
The things you mention like high memory speed etc. are needed when you meet a confusing problem which you try to clarify in a little amount of time. Which is the hard way. Doing the same in the study room is a different matter.
I made a blunder set, containing puzzles I failed. Well, the training outcome wasnt that good for me. I guess it is o.k. to train your blunders. The really important thing, however, is: you need to become faster. Not only in puzzles you once failed, but in general - all puzzles are worth to be trained and to reduce the solving time needed. Maybe blunders are especially worth to be trained, but I could not notice any boost in particular from my blunder-puzzles.
ReplyDeleteSpeed is what is needed.
Aox tried to reduce solving time in boardvisions or easy puzzles. Not an easy thing to achieve. If we can not become faster - I dont think we can beome better.
Aox tried for instance 6 sets of easy-Mate-in-1-puzzles, each 300 puzzles in size.
He could not become faster, isnt that right, Aox?
I tried, too, and I could not become faster. I trained just 1 set of 300 puzzles.
The only critique I have here is: my solving time was already quite fast: 35 Mate-in-1-puzzles. So maybe there is simply a limit on such an easy task. This is what I believe (but didnt try so far): I can not become significiantly faster in those M1-puzzles. However, I could get my solving time down in 5x300 M2-puzzles, in the range not too easy 1200-1400 CT Blitz. Still easy, but not ultra-easy. Significiantly would be: getting my solving time down from 6 sec on average to 3 sec on average. This would need to be tested, though. Who is doing the work?
You are trying to learn every possible manifestation of a rabbit as a different thing. That is both tedious and useless. You need to recognize the archetype of all rabbits. That way you wil recognize all different kinds of rabbits in the sky. No matter how much the cloud they are made of differs from the archetype. You never had to learn the googol amount of different clouds that form a rabbit in the sky.
ReplyDeleteYou don't have to speed up, you need to see clairity in a position stead of confusion. Speed is the automatic result you get for free.
Some of the problems where you think they are easy are in real not that easy, you did not see some motives the stronger player may see, maybe there are some more candidates, maybe some dangers for you or some extra chances of the "opponent" you did not see...
ReplyDeleteSometimes we find the right solution quick "by chance", sometimes we dont make a typical blunder of players of our strength "by chance".
The good thing about chesstempo, you may look at the errors of the othere tacticians. This may give you a deeper understanding of the problem and the typical errors of others.
But in general, the better player will find a solution quicker than a weaker player.
This paper http://www.cogsys.org/papers/ACS2015/article7.pdf is about the calculation of the "difficulty of tactical chessproblems". IF this paper makes any sense then the difficulty of a problem is ( to some extend ) given by the amount of calculation needed to solve it. The used/estimated calculation algorithm is a traditional candidate move algorithm.
So if we are weaker=slower then we are not propper with your calculations, we get lost in the tree of calculations, we calculate to many candiates or our visualisation/evaluation is week.
I quess its both
"The combinations in the position are pretty straightforward. If I want to learn from my mistakes, I have to find a way to defy step by step calculations, in favour for seeing what is going on in one go."
ReplyDeleteI guess that if this takes somebody long then it is because he is not used to see that a "check" is very forcing. It is not that 1.BxBe5+ is just a capture. It is also a check giving move.
After 1...Qxe5 the recapture 2.Qxe5+ is again a check. This makes solving this puzzle so much simpler, but you really need to be aware that it is a check.
If this puzzle takes long to solve, I could imagine it is this detail which is missed and makes you think so much longer.
How to improve here in speed? I feel that this is an example of lack of boardvision - here: Fritz boardvision "check".
Maybe also the "defense" training, because if black answers: 1...RxBe5 (instead of 1...Qxe5) then one could miss for quite some time that the Rf8 is hanging. But if you have trained with Fritz boardvision "defense" training and see 50 undefended pieces per minute, then you dont even conciously think about 1...RxBe5 because it would hang the rook on f8, and thus does not even reach your concious thinking but filters out the respond 1.RxBe5 and only continues to look at 1...QxBe5 --> after which 2.Qxe5+ is a check, and for me it is already solved here. No need to continue to calculate 2...RxQe5 because - again - it hangs the rook on f8. However, I feel I kept calculating and calculated here 2...RxQe5 and 3.Rxf8, but this is rather triggered by the well learned behaviour to keep capturing until the positioned calmed down and no captures are left in the position.
For even more detail, see what I have written at aoxomoxoa wondering blog about this puzzle. It is interesting how many thoughts aox put into this puzzle, whereas I thought almost nothing, but the main variation.
The better tactician can do 2 things better : calculate better ( less errors, deeper, faster ) and spot more patterns imedialtly ( = amount of puzzles sloved in hyperspeed ).
ReplyDeleteI think both skills are related. If you can spot more pattern => its easier to calculate and if you can calculate better => you learn how to spot more pattern ( i hope ).
We did try to improve the spot-rate by spaced repetition of tactical pattern asf, with some result. Board vision can help too, its (part of ) the first layer of our tactical neuronal network, but we did not find a real breakthrough until now.
I will continue with the thoughtprocess-calculation idea for a while but i dont think this will be decisiv either, remiem was working that way and had no improvement. But i want to test all different possible methods to get better in tactics.
If I remember well Munich put a lot of effort in labeling the combination types in a problem. Did that coïncide with his biggest progress?
ReplyDeleteI would add the most important factor to Aox's comment.
ReplyDeleteAox said: "The better tactician can do 2 things better: 1)calculate better and 2) spot more patterns immediately".
BETTER play (tactician) knows WHAT to focus on. He finds stronger moves earlier - without looking at "strange" moves (weaker player focus on).
It is (may) not be strong related. For example: I know many patterns, but I calculate rather weak. There are players who calculate "naturally" and others who does it "by force". I am the second type one (even If I want to be the 'natural one').
I think that being better at tactics requires UNDERSTANDING the tactical motives VERY well and develop a good board vision. If you add to this the patterns database and the ability to calculate well (the most important things first) - you will be a master (at tactics).
What do you think about this idea?
The calculation methods like Chuzakins etc. are too slow to be of much use OTB.
ReplyDeleteThese systems are meant for the study room. Patterns are the main way to speed up.
The idea that you need to have a very big database with an extreme amount of patterns isn't very logical. My next post will be about that.
The method of Chuzakin is fast, because the HE's don't change at every move. Once recognised there is only an small update from time to time. This way tactics becomes a static thing like a positional weakness (double-pawn, backward-pawn, hole...).
ReplyDeleteI use this system now for some months and there was no change in my mixed tactic puzzle rating, so the whole overhead should pay of at OTB
This feels very ctart ish. Does anyone know if this is a composition or if it occured in a game ? I actually care about those sort of things...... saw the bad ass diagonal and the back rank weakness losing the rook but need to sharpen up.
ReplyDeleteIt is a problem from Chess Tempo. All problems there stem from real games.
Delete