Hello guys!Today I was analyzing a game I played and found out that one of the moves I played which changed the entire course of the game was a blunder.So i moved Ne5 attacking c6 square.If you were black and it was your move,how would you respond?.Here is the game lichess.org/Axi7LOYM/white#23
You win more material with the computer's move and your position looks good too. With your move, if you weren't winning a piece, its clear your position would be a total turd since you lose the right to castle and his pieces become very active. Compare his pieces with that bishop of yours. Maybe its a little surprising that the computer thinks a piece up position is that close, but sometimes that happens when your position is really crappy except for the material.
I would have played the same or more likely something worse.
I'm really surprised by the engine evaluation after Ke3+ in the suggested line, but (a posteriori) I agree that Black position looks fine and this Knight safely parked at e3 is going to be a nightmare.
Very interesting game, thanks for sharing!
P. S.: note that this "blunder" means only you could get a better outcome, but still you have an advantage anyway.
I'm really surprised by the engine evaluation after Ke3+ in the suggested line, but (a posteriori) I agree that Black position looks fine and this Knight safely parked at e3 is going to be a nightmare.
Very interesting game, thanks for sharing!
P. S.: note that this "blunder" means only you could get a better outcome, but still you have an advantage anyway.
with the queen move white ends up with a knight up.
with the white king move, but the more punishing black night move
it is a queen for 2 knights, but an annoying stable knight deep in white territory.
as previous post noted.
I would not focus so much on one having to see that deep as the oracle would grade it a blunder, it is our lot, humans to play constantly in suboptimal chess.. (well most of us).
I think seeing the long term consequence though might raise the interesting positional evaluation that a spatially deep relatively unchallenge knight pressure before castling might do, the king is exposed without it being an endgame. white can't do much but address that. At the very least it makes it a question to further explore.. probably in other games where the motivation background is hight than in postgame alone.
but back to op concern about the word "blunder".. It is verbal inflation, consequence of a certain lack of realism, in established but implicit theories of learning.
It does get numbing to have such verbal inflation based on such depth of turn by turn divination.. Maybe there ought to be other ways of grading ones play, while not forgetting the excellency of the oracle. something more adjusted to the players level band around current rating estimate maybe? So that one can learn over many games not just perfection distance in one game... i find skimming on errors to see the most accessible to own level of vision as a layered feedback approach..
not very popular stance.. There is some much bling non-theories of learning deeply entrenched in many of the chess teaching tool, it is exhausting trying to suggest otherwise.
lichess trying to adjust from the game outcome backward, is still stuck to the ceiling as the progressive measure over the whole game for "learning from your mistakes"..
with the white king move, but the more punishing black night move
it is a queen for 2 knights, but an annoying stable knight deep in white territory.
as previous post noted.
I would not focus so much on one having to see that deep as the oracle would grade it a blunder, it is our lot, humans to play constantly in suboptimal chess.. (well most of us).
I think seeing the long term consequence though might raise the interesting positional evaluation that a spatially deep relatively unchallenge knight pressure before castling might do, the king is exposed without it being an endgame. white can't do much but address that. At the very least it makes it a question to further explore.. probably in other games where the motivation background is hight than in postgame alone.
but back to op concern about the word "blunder".. It is verbal inflation, consequence of a certain lack of realism, in established but implicit theories of learning.
It does get numbing to have such verbal inflation based on such depth of turn by turn divination.. Maybe there ought to be other ways of grading ones play, while not forgetting the excellency of the oracle. something more adjusted to the players level band around current rating estimate maybe? So that one can learn over many games not just perfection distance in one game... i find skimming on errors to see the most accessible to own level of vision as a layered feedback approach..
not very popular stance.. There is some much bling non-theories of learning deeply entrenched in many of the chess teaching tool, it is exhausting trying to suggest otherwise.
lichess trying to adjust from the game outcome backward, is still stuck to the ceiling as the progressive measure over the whole game for "learning from your mistakes"..
@dboing said in #4:
> Maybe there ought to be other ways of grading ones play
Blunders/mistakes/inaccuracies don't grade playing, but accuracy. Contrary to other sports (a mistake by Tiger Woods would be a fantastic performance for me) here we have just a centipawn loss count: Carlsen blunders have the same "blunderness" than mine, we'd need a measure of difficulty to discriminate (I assume my blunders are easier to spot).
> Maybe there ought to be other ways of grading ones play
Blunders/mistakes/inaccuracies don't grade playing, but accuracy. Contrary to other sports (a mistake by Tiger Woods would be a fantastic performance for me) here we have just a centipawn loss count: Carlsen blunders have the same "blunderness" than mine, we'd need a measure of difficulty to discriminate (I assume my blunders are easier to spot).
I meant something that take into account the current level of play as a relative baseline. Odds conversion curves also exist and behave very well (apparently, there average conditional distribution does, did not see any error bars anywhere, not lichess not SF, and i forgot about lc0) from the looks of one of the first Maia papers figure (11).
They had done the same work that lichess does to relativise accuracy for high ratings, but for all sorts of ratings, or the lichess population.
So, not to dismiss the SF dependent accuracy measure we have been using, or any notion that perfect or best chess is not best chess, but something not using only expertise as the manageable objective chunks given where one is at, might allow more adjusted chunks of "learn from your mistake" projects. Sometimes perfect is the enemy of good.
Not the first time I have such thoughts.. Lichess page on accuracy has evolved quite well over the past 5 years.
Also, I may not have meant what you understood by grading ones plays... I meat the learn from your mistake one direction point of view. SF dependent accuracy (SF accuracy itself is without any external referential itself, it depends on the engine pool ecosystem diversity, which has never been really studied AFAIK, they have some human generated test suites, but I don't think they are used to test how diverse all the engine pool behaviors on the boards are). It is fine as long as we know what it is, that accuracy referential.
I just find it not very realistic or adaptive.. to the long process of improvement over many games. I have the same problem with studying from best games, irrespective of the learner status.. as if it was accessible to all, other than as imitation learning, likely of the move type rather than from the position features. How is one going to learn about all the hidden games not played within only one game.. How to spread the perfection goal over a smoother adaptive gradient of difficulty.
It that what you were hinting at. by difficulty.?
www.semanticscholar.org/paper/Approaches-to-Measuring-the-Difficulty-of-Games-in-Dziedzic-Wlodarczyk/d28235adc2c8c6fdfaa474bc2bab931129149fd6
They had done the same work that lichess does to relativise accuracy for high ratings, but for all sorts of ratings, or the lichess population.
So, not to dismiss the SF dependent accuracy measure we have been using, or any notion that perfect or best chess is not best chess, but something not using only expertise as the manageable objective chunks given where one is at, might allow more adjusted chunks of "learn from your mistake" projects. Sometimes perfect is the enemy of good.
Not the first time I have such thoughts.. Lichess page on accuracy has evolved quite well over the past 5 years.
Also, I may not have meant what you understood by grading ones plays... I meat the learn from your mistake one direction point of view. SF dependent accuracy (SF accuracy itself is without any external referential itself, it depends on the engine pool ecosystem diversity, which has never been really studied AFAIK, they have some human generated test suites, but I don't think they are used to test how diverse all the engine pool behaviors on the boards are). It is fine as long as we know what it is, that accuracy referential.
I just find it not very realistic or adaptive.. to the long process of improvement over many games. I have the same problem with studying from best games, irrespective of the learner status.. as if it was accessible to all, other than as imitation learning, likely of the move type rather than from the position features. How is one going to learn about all the hidden games not played within only one game.. How to spread the perfection goal over a smoother adaptive gradient of difficulty.
It that what you were hinting at. by difficulty.?
www.semanticscholar.org/paper/Approaches-to-Measuring-the-Difficulty-of-Games-in-Dziedzic-Wlodarczyk/d28235adc2c8c6fdfaa474bc2bab931129149fd6
@dboing said in #6:
> It that what you were hinting at by difficulty?
>
> www.semanticscholar.org/paper/Approaches-to-Measuring-the-Difficulty-of-Games-in-Dziedzic-Wlodarczyk/d28235adc2c8c6fdfaa474bc2bab931129149fd6
I would go back to define "difficulty", currently we have the rating but it is not easy to grade the difficulty for a given puzzle/position unless it enters the system of many people trying to solve it.
What is more difficult, a puzzle with 6 moves not very challenging one by one, or a one-move puzzle with two similar alternatives? Maybe the mistake is trying to fit a multi-dimensional idea in just one quantity.
Even previous knowledge by the player (that we can assume random) can transform the difficulty. This one is easy, if you know the box rule:
lichess.org/export/fen.gif?fen=8/8/8/k7/r4P2/8/R3p3/4K3_w_-_-_0_2&color=white
This one is not that easy, but it becomes trivial if you know the stuff about key squares (Black to move):
lichess.org/export/fen.gif?fen=8/8/8/8/5P2/1k6/3K4/8_b_-_-_0_1&color=black
etc...
> It that what you were hinting at by difficulty?
>
> www.semanticscholar.org/paper/Approaches-to-Measuring-the-Difficulty-of-Games-in-Dziedzic-Wlodarczyk/d28235adc2c8c6fdfaa474bc2bab931129149fd6
I would go back to define "difficulty", currently we have the rating but it is not easy to grade the difficulty for a given puzzle/position unless it enters the system of many people trying to solve it.
What is more difficult, a puzzle with 6 moves not very challenging one by one, or a one-move puzzle with two similar alternatives? Maybe the mistake is trying to fit a multi-dimensional idea in just one quantity.
Even previous knowledge by the player (that we can assume random) can transform the difficulty. This one is easy, if you know the box rule:
lichess.org/export/fen.gif?fen=8/8/8/k7/r4P2/8/R3p3/4K3_w_-_-_0_2&color=white
This one is not that easy, but it becomes trivial if you know the stuff about key squares (Black to move):
lichess.org/export/fen.gif?fen=8/8/8/8/5P2/1k6/3K4/8_b_-_-_0_1&color=black
etc...
@OctoPinky said in #7:
> Maybe the mistake is trying to fit a multi-dimensional idea in just one quantity
> Maybe the mistake is trying to fit a multi-dimensional idea in just one quantity
This topic has been archived and can no longer be replied to.