- Blind mode tutorial
lichess.org
Donate

Finding Pawn Breaks automatically

#11

@EmaciatedSpaniard said in #9:

Sure, maybe i didn't understand the purpose of the blog. I thought you'd try to get some insight into how to identify good pawn breaks thru analysis. I can't use stockfish when i'm playing someone and that's when such insight would really help. I think its not so obvious to lower rated players like myself what features make up a good pawn break. That would be quite valuable information.

I totally agree with that, although not the author. I would say this is not a standalone blog. And that there was need to be careful where previous existing methods of communications only using one head point of view, now could use data analysis. But care must be taken for the object used not to be fudged by some implicit floating variable knowledge. So I would agree that this is an open question, and I would think it is also in the mind of the op. I don't think this is a one person question. We just need to construct currently non-existing tools of that scope and chess theory ambition. And now we could be many at it. It may be about us to use such tools and find things.. Why should it be on the shoulder of the op, only. And why should it be one person with all the answers.

I think the blog is not about chess truth delivered to the many, but more about hey open source open reasoning, and audience being chess enthousiast, could use that. ok. not in game (that might have been a problem that assumption of in-game closed book).

There is a lot of study in chess learning, maybe not all need to be ready to wear stuff. This is for studying chess. Each one of us might find our ways in game later. Study first, Win later!

@EmaciatedSpaniard said in #9: > Sure, maybe i didn't understand the purpose of the blog. I thought you'd try to get some insight into how to identify good pawn breaks thru analysis. I can't use stockfish when i'm playing someone and that's when such insight would really help. I think its not so obvious to lower rated players like myself what features make up a good pawn break. That would be quite valuable information. I totally agree with that, although not the author. I would say this is not a standalone blog. And that there was need to be careful where previous existing methods of communications only using one head point of view, now could use data analysis. But care must be taken for the object used not to be fudged by some implicit floating variable knowledge. So I would agree that this is an open question, and I would think it is also in the mind of the op. I don't think this is a one person question. We just need to construct currently non-existing tools of that scope and chess theory ambition. And now we could be many at it. It may be about us to use such tools and find things.. Why should it be on the shoulder of the op, only. And why should it be one person with all the answers. I think the blog is not about chess truth delivered to the many, but more about hey open source open reasoning, and audience being chess enthousiast, could use that. ok. not in game (that might have been a problem that assumption of in-game closed book). There is a lot of study in chess learning, maybe not all need to be ready to wear stuff. This is for studying chess. Each one of us might find our ways in game later. Study first, Win later!
#12

@dboing said in #11:

I think the blog is not about chess truth delivered to the many

I think the blog is about finding pawn breaks automatically.

@dboing said in #11: > I think the blog is not about chess truth delivered to the many I think the blog is about finding pawn breaks automatically.
#13

@Toadofsky said in #12:

I think the blog is about finding pawn breaks automatically.

Are you toying with words? I agree otherwise. But with an automatically definable definition, which means as humans we would have the logic or reasoning ability to also see it without prior experience, by reading the definition.

And that definition, as that is in need of work (for ages now), might want to not be the most restrictive just to please some past individual own restrictions of expertise that would not spell out all assumptions.

Therefore tangling evaluation within the definition or tangling prevalance in some floating variable notion of what is likely to be encountered, might not be the most human generalizable (with accuracy, if one day it could be measure beyond ELO ratings), starting definitoin.

I approve of chosing a wider core definitoin. It might not be a juicy winning recipe when you spot one.. but at least now we can work on it.

It is a question of scope assumptions. I might be reading more that author intended, but that does not matter, that reading exists, and I just tried to explain it more than once.

and now this

What do you mean by pawn breaks?

@Toadofsky said in #12: > I think the blog is about finding pawn breaks automatically. Are you toying with words? I agree otherwise. But with an automatically definable definition, which means as humans we would have the logic or reasoning ability to also see it without prior experience, by reading the definition. And that definition, as that is in need of work (for ages now), might want to not be the most restrictive just to please some past individual own restrictions of expertise that would not spell out all assumptions. Therefore tangling evaluation within the definition or tangling prevalance in some floating variable notion of what is likely to be encountered, might not be the most human generalizable (with accuracy, if one day it could be measure beyond ELO ratings), starting definitoin. I approve of chosing a wider core definitoin. It might not be a juicy winning recipe when you spot one.. but at least now we can work on it. It is a question of scope assumptions. I might be reading more that author intended, but that does not matter, that reading exists, and I just tried to explain it more than once. and now this What do you mean by pawn breaks?
#14

@dboing said in #13:

What do you mean by pawn breaks?

I define pawn breaks as pawn moves that attack an enemy pawn which is blocked by a friendly pawn.

@dboing said in #13: > What do you mean by pawn breaks? I define pawn breaks as pawn moves that attack an enemy pawn which is blocked by a friendly pawn.
#15
<Comment deleted by user>
#16

@Toadofsky said in #14:

I define pawn breaks as pawn moves that attack an enemy pawn which is blocked by a friendly pawn.

Ok. So like the op first take. Then the open reasoning consideration or restrictions as amendments to that.

So there is more than that first step. The open reasoning for me, is that the text is giving building blocks on the restrictions explicitely, and readable by non expert but chess able audience all the way to the SF black box oracling.

I think while the best pipeline control point for the evaluation dimension of the concept (I avoid definition for now, as I prefer the automatic one and also human without other prior than ruleset definition), SF is a breach of open reasoning, as I saw before, in that sense I agree with @EmaciatedSpaniard (with slight anlge difference, but shared the curiosity not satistfied with SF expertise).

But, I am saying also, that one person can't do all of that alone, and that the work before the SF (giving up the open reasoning part), is a fine many heads foundation of language for discussions (and I don't mean formal language, but maybe some pressure toward it, a bit, to fend off the chess theory critique, say, like MFTL would have needed to splash the pond with before. But now is later from splashing that pond. And we might also while using the automatic pressure on our language let some chess theory evovle. And drop the corset of performance one-champion show lurking groove and expectations from one blog. and one author.

So I am saying more thatn one things. Each of which, may just be my deluded point of view. Some of which may be reasonable to others. or none of them. I wanted to disentangle my own ramblings (which is my thoughts as I write). I thank you again, for your concise, chunk crutches to get me the motivation to try again rephrasing myself better than some bulldozer AI (really).

@Toadofsky said in #14: > I define pawn breaks as pawn moves that attack an enemy pawn which is blocked by a friendly pawn. Ok. So like the op first take. Then the open reasoning consideration or restrictions as amendments to that. So there is more than that first step. The open reasoning for me, is that the text is giving building blocks on the restrictions explicitely, and readable by non expert but chess able audience all the way to the SF black box oracling. I think while the best pipeline control point for the evaluation dimension of the concept (I avoid definition for now, as I prefer the automatic one and also human without other prior than ruleset definition), SF is a breach of open reasoning, as I saw before, in that sense I agree with @EmaciatedSpaniard (with slight anlge difference, but shared the curiosity not satistfied with SF expertise). But, I am saying also, that one person can't do all of that alone, and that the work before the SF (giving up the open reasoning part), is a fine many heads foundation of language for discussions (and I don't mean formal language, but maybe some pressure toward it, a bit, to fend off the chess theory critique, say, like MFTL would have needed to splash the pond with before. But now is later from splashing that pond. And we might also while using the automatic pressure on our language let some chess theory evovle. And drop the corset of performance one-champion show lurking groove and expectations from one blog. and one author. So I am saying more thatn one things. Each of which, may just be my deluded point of view. Some of which may be reasonable to others. or none of them. I wanted to disentangle my own ramblings (which is my thoughts as I write). I thank you again, for your concise, chunk crutches to get me the motivation to try again rephrasing myself better than some bulldozer AI (really).
#18

Well, that is to filter stuff, nothing prevent studying the filerred stuff then for what theory is leaving to hunch learning by experience, only it would be clear what would be filters and served to further study. And clear that these would be whatever bias might still be in SF goggle about its embedded function over its notion of what chess position wilderness might be that all the competing engines have been serving it recently, under the ELO (and its dependent pool of games and visited position set variable, rarely on our automatic radar).

So as long as SF is very loud as the projection of chess evaltuion sub-space as en evuation, the obscurity boundaries are where they seem to solidly be stuck, as we do seem to want any oracle, if not human (GMs, or XMs) then machines..

But then, one could still buld eithther the usual single preformance leanring from such experience guide, knowinhg it is the point of view of SF (and all its ELO parters of history, at least last tournments concurrent categories). But only at the evaluation stage.

The rest would be as made explicit conditional filters that the blog author shared with us in terms we can understand from chess land.

While SF entrails might still be hush hush mysterioso... and smoke we don,t really care.. As ELO is all we need.

So then without torturing busy devs of chess engines with explicit puprose to get such above resticted ELO competition measure, doing the same prezenation as the blog author on the rest of the filter reasoning, we can still use our reasoning to figure out more within the smaller sets and I would add constracting low SF value ones...

That could be a more manageable. As long as we keep the SF stamp propagating as the evaluation filter.. It might be recuperable later upon further evolution of chess engine and chess science ( I would not say theory, for fear of stepping on some turh that has precedene, and absolute authority). so since no one really talks about chess science... let me have it.. no harm... (evolution not same as evaluation, sorry for typos).

Well, that is to filter stuff, nothing prevent studying the filerred stuff then for what theory is leaving to hunch learning by experience, only it would be clear what would be filters and served to further study. And clear that these would be whatever bias might still be in SF goggle about its embedded function over its notion of what chess position wilderness might be that all the competing engines have been serving it recently, under the ELO (and its dependent pool of games and visited position set variable, rarely on our automatic radar). So as long as SF is very loud as the projection of chess evaltuion sub-space as en evuation, the obscurity boundaries are where they seem to solidly be stuck, as we do seem to want any oracle, if not human (GMs, or XMs) then machines.. But then, one could still buld eithther the usual single preformance leanring from such experience guide, knowinhg it is the point of view of SF (and all its ELO parters of history, at least last tournments concurrent categories). But only at the evaluation stage. The rest would be as made explicit conditional filters that the blog author shared with us in terms we can understand from chess land. While SF entrails might still be hush hush mysterioso... and smoke we don,t really care.. As ELO is all we need. So then without torturing busy devs of chess engines with explicit puprose to get such above resticted ELO competition measure, doing the same prezenation as the blog author on the rest of the filter reasoning, we can still use our reasoning to figure out more within the smaller sets and I would add constracting low SF value ones... That could be a more manageable. As long as we keep the SF stamp propagating as the evaluation filter.. It might be recuperable later upon further evolution of chess engine and chess science ( I would not say theory, for fear of stepping on some turh that has precedene, and absolute authority). so since no one really talks about chess science... let me have it.. no harm... (evolution not same as evaluation, sorry for typos).
#19

I ran the blog's code (with some modification) on some Games of the Month. There were 2 pawn breaks found in the 4 games. In the December 2023 GOTM, c7c5 is a common move in the French Defence. Move 15 h5 looks like a pawn break to me, but the pawn is blocked by a Knight and so this move was not found by the blog's code. In the October 2023 GOTM, b4b5 is a pawn break.

The 2 found pawn breaks both followed the Pointing Rule. There was one missed pawn break which could be found by modifying the pawn break definition to "blocked by any piece". Some possible definitions:

  1. pawn break: pawn move that attacks an enemy pawn which is blocked by a friendly pawn. (from the blog)
  2. pawn break: pushing a pawn with the intent to force a file open.
  3. pawn break: putting another pawn alongside your most advanced pawn. (from "The Giants of Strategy" by Neil McDonald)
  4. pawn break: pawn move that attacks an enemy pawn which is blocked by any piece.

===================
pawnBreaksRevised2024.py output:

January 2024 GOTM najvazniji_vs_Landeo.pgn
no pawn breaks found
https://lichess.org/@/Lichess/blog/lichess-game-of-the-month-january-24/vpiLewKp

December 2023 GOTM ChessStrawberry2020_vs_DrMikeYeadonIsRight.pgn
c7c5
rnbqkb1r/pppn1ppp/4p3/3pP3/3P1P2/8/PPPN2PP/R1BQKBNR b KQkq - 0 5
https://lichess.org/@/Lichess/blog/lichess-game-of-the-month-december-23/iIK7sqNa

November 2023 GOTM FC7PLAY_vs_vikasvennella.pgn
no pawn breaks found
https://lichess.org/@/Lichess/blog/lichess-game-of-the-month-november-23/dHRdxk5Q

October 2023 GOTM pleaseresignnow_vs_Johnwanph.pgn
b4b5
3r1rk1/1pqnbppp/p1p1pn2/2P2b2/PP1P4/1QN1BN2/4BPPP/R4RK1 w - - 1 15
https://lichess.org/@/Lichess/blog/lichess-game-of-the-month-october-23/yz45dkqO

I ran the blog's code (with some modification) on some Games of the Month. There were 2 pawn breaks found in the 4 games. In the December 2023 GOTM, c7c5 is a common move in the French Defence. Move 15 h5 looks like a pawn break to me, but the pawn is blocked by a Knight and so this move was not found by the blog's code. In the October 2023 GOTM, b4b5 is a pawn break. The 2 found pawn breaks both followed the Pointing Rule. There was one missed pawn break which could be found by modifying the pawn break definition to "blocked by any piece". Some possible definitions: 1. pawn break: pawn move that attacks an enemy pawn which is blocked by a friendly pawn. (from the blog) 2. pawn break: pushing a pawn with the intent to force a file open. 3. pawn break: putting another pawn alongside your most advanced pawn. (from "The Giants of Strategy" by Neil McDonald) 4. pawn break: pawn move that attacks an enemy pawn which is blocked by any piece. =================== pawnBreaksRevised2024.py output: January 2024 GOTM najvazniji_vs_Landeo.pgn no pawn breaks found https://lichess.org/@/Lichess/blog/lichess-game-of-the-month-january-24/vpiLewKp December 2023 GOTM ChessStrawberry2020_vs_DrMikeYeadonIsRight.pgn c7c5 rnbqkb1r/pppn1ppp/4p3/3pP3/3P1P2/8/PPPN2PP/R1BQKBNR b KQkq - 0 5 https://lichess.org/@/Lichess/blog/lichess-game-of-the-month-december-23/iIK7sqNa November 2023 GOTM FC7PLAY_vs_vikasvennella.pgn no pawn breaks found https://lichess.org/@/Lichess/blog/lichess-game-of-the-month-november-23/dHRdxk5Q October 2023 GOTM pleaseresignnow_vs_Johnwanph.pgn b4b5 3r1rk1/1pqnbppp/p1p1pn2/2P2b2/PP1P4/1QN1BN2/4BPPP/R4RK1 w - - 1 15 https://lichess.org/@/Lichess/blog/lichess-game-of-the-month-october-23/yz45dkqO
#20

# top-level code in pawnBreaksRevised2024.py

from os import listdir

if name == 'main':
filenames = listdir(os.path.dirname(file) + '/pgnFiles')
for filename in filenames:
if filename.endswith('.pgn'):
print(filename)
pgnFile = os.path.dirname(file) + '/pgnFiles/' + filename
pb = findPawnBreaks(pgnFile)
for k,v in pb.items():
print(k)
for p in v:
print(p)

# top-level code in pawnBreaksRevised2024.py from os import listdir if __name__ == '__main__': filenames = listdir(os.path.dirname(__file__) + '/pgnFiles') for filename in filenames: if filename.endswith('.pgn'): print(filename) pgnFile = os.path.dirname(__file__) + '/pgnFiles/' + filename pb = findPawnBreaks(pgnFile) for k,v in pb.items(): print(k) for p in v: print(p)