Your network blocks the Lichess assets!

lichess.org
Donate

Help us create better chess puzzles!

I actually second the post suggesting linkage via lichess. Also, is there a possibility to delete my account there.
Just for the principle. I intend to contribute as I do like the idea, if it follows open source, open data, and open reasoning eventually, passed the competition obscurity or publication blackout.

I would point to ELOMETER lack of reproducible efforts wasted on the community, if it were interested in more extrinsic ways to learn about the game they love. But then, I hope there are reasoning or argument or some ideas behind the composition of the postion seeds of the puzzles. At the very least, it would be exposed eventually. Expert names, is not as satisfying as some insightful design. Although we take what we can. What should we expect?

Also, I am not attracted by the "better puzzle" title. More by the data analysis experimental design. Please keep us informed as much as the competition secrecy needs permits. And then after not fall into IEEE paywalls for scientific access.

Chess is not an arcane art. In spite of our limitations expanding its finite set of logical rules. I welcome any science attempt.

I actually second the post suggesting linkage via lichess. Also, is there a possibility to delete my account there. Just for the principle. I intend to contribute as I do like the idea, if it follows open source, open data, and open reasoning eventually, passed the competition obscurity or publication blackout. I would point to ELOMETER lack of reproducible efforts wasted on the community, if it were interested in more extrinsic ways to learn about the game they love. But then, I hope there are reasoning or argument or some ideas behind the composition of the postion seeds of the puzzles. At the very least, it would be exposed eventually. Expert names, is not as satisfying as some insightful design. Although we take what we can. What should we expect? Also, I am not attracted by the "better puzzle" title. More by the data analysis experimental design. Please keep us informed as much as the competition secrecy needs permits. And then after not fall into IEEE paywalls for scientific access. Chess is not an arcane art. In spite of our limitations expanding its finite set of logical rules. I welcome any science attempt.

About the ML competition where we have to guess the puzzle's rating, what is stopping me from scraping the lichess puzzles database and submit the real ratings ? I guess if one does that he won't be eligible for prizes

About the ML competition where we have to guess the puzzle's rating, what is stopping me from scraping the lichess puzzles database and submit the real ratings ? I guess if one does that he won't be eligible for prizes

@StudyKeyser said in #13:

About the ML competition where we have to guess the puzzle's rating, what is stopping me from scraping the lichess puzzles database and submit the real ratings ? I guess if one does that he won't be eligible for prizes

Well if this is science based, or has any reproducibility concern, it would not just be about getting some right answer.

It is other puzzles not of the lichess database that are getting ratings, using the user average rating from lichess.

In some way, they may not have the ratings calculated yet, as this is the object of the research. The flow of information is not to predict lichess puzzle ratings. But the lichess user puzzle rating informing their pool of puzzles rating via the user testing theirs.

Why it would be nice to hear about the method. That can't be a secret. What kind of competition is that?

@StudyKeyser said in #13: > About the ML competition where we have to guess the puzzle's rating, what is stopping me from scraping the lichess puzzles database and submit the real ratings ? I guess if one does that he won't be eligible for prizes Well if this is science based, or has any reproducibility concern, it would not just be about getting some right answer. It is other puzzles not of the lichess database that are getting ratings, using the user average rating from lichess. In some way, they may not have the ratings calculated yet, as this is the object of the research. The flow of information is not to predict lichess puzzle ratings. But the lichess user puzzle rating informing their pool of puzzles rating via the user testing theirs. Why it would be nice to hear about the method. That can't be a secret. What kind of competition is that?

how many puzzle tried do you need us to try minimally. And is there some notion of time or duration of puzzle done in a row.

would 30 be enough? How many total puzzle IDs in your set? by the way, curious.

It would be nice to have some feedback about the puzzles we did. so I can stop when you have enough. I want to help, but I am selfish too... On lichess I don't do puzzles just for the rating. There is more stuff to do with them. It is as much about me doing them as it is about the positions themselves, my encountering them.

Maybe the dasborad and history of puzzles is function. I wlil look.. I just did not see the usual underboard limited history, so guessing no history visible to user.

how many puzzle tried do you need us to try minimally. And is there some notion of time or duration of puzzle done in a row. would 30 be enough? How many total puzzle IDs in your set? by the way, curious. It would be nice to have some feedback about the puzzles we did. so I can stop when you have enough. I want to help, but I am selfish too... On lichess I don't do puzzles just for the rating. There is more stuff to do with them. It is as much about me doing them as it is about the positions themselves, my encountering them. Maybe the dasborad and history of puzzles is function. I wlil look.. I just did not see the usual underboard limited history, so guessing no history visible to user.

https://chess.knowledgepit.ai/training/history

so that works. but none of the others. I will assume 30 puzzles is ok for you. I would like in return for my help to know how big the puzzle DB might be. This might not make it a competition disadvantage knowledge leak, would it? (sorry, I am having some bad temper from the notion of secret acting like sand in the pipes of some ideal world I kept fantasizing being part of). And it might not even be that bad in your endeavor. Just a reminder of many things that have been getting on my nerves. I can't even blame the heat today. Full disclosue, I am in a ranting mood about the game of knowledge that real chess (OTB time pressure and big book competition, increasing big book) have become. And that a lot of the bandwight of communicatoin is overwhelmed it seems about that. As if chess was only about the knowledge game. And how fast knoweldge advantage can be used as part of the game ruleset, almost. I admit. This might be spam. off-topic. etc... delete at will.

https://chess.knowledgepit.ai/training/history so that works. but none of the others. I will assume 30 puzzles is ok for you. I would like in return for my help to know how big the puzzle DB might be. This might not make it a competition disadvantage knowledge leak, would it? (sorry, I am having some bad temper from the notion of secret acting like sand in the pipes of some ideal world I kept fantasizing being part of). And it might not even be that bad in your endeavor. Just a reminder of many things that have been getting on my nerves. I can't even blame the heat today. Full disclosue, I am in a ranting mood about the game of knowledge that real chess (OTB time pressure and big book competition, increasing big book) have become. And that a lot of the bandwight of communicatoin is overwhelmed it seems about that. As if chess was only about the knowledge game. And how fast knoweldge advantage can be used as part of the game ruleset, almost. I admit. This might be spam. off-topic. etc... delete at will.

@dleli said in #5:

Good points. It was my understanding that the research front here was more methodological about the puzzle construction method, and then perhaps the very pool glicko rating system on one hand, and how one could use that different pool of puzzles but common sub-pool of players (having gotten their ratings over at least 30 puzzles from the whole pool of human-puzzle IDs in Lichess) to perhaps transmit some seeding of rating estimate for the new competition events with a different half of pool, the puzzle IDs. Sorry, it would be clearer with a potato diagram and arrows. I stumble in the stream version. of something rather simple otherwise. I guess calibration problem in few words...?

My wet brain might be hallucinating. (I guess I might be natural intelligence, then). That is the imagined idea proposed that, I thought, was in the words of the blog. I still do think that it is a methodological question, but it might not be the actual goal.
Thinking aloud as usual: But it would not be framed as a data analysis competition. So I guess the title is not what sparked my interest, and it might have been a slippery choice. But I would not blame the authors. This might be the norm, is it not? Limited window of visibility, do what you can to attract relevant attention.

At least "puzzle" is in the title. And people might have not computed another title, about calibration experiment.

And it might still be true that a side effect of the project might be more informative puzzles. I guess that is why I compare with ELOMETER. And fuss about transparency of reasoning and method. For this to actually be a better offer, it might not just be about rating, but about the specifics of its database content, and possibly coaching theory or something we might find improving our understanding of our experience.

@dleli said in #5: > Good points. It was my understanding that the research front here was more methodological about the puzzle construction method, and then perhaps the very pool glicko rating system on one hand, and how one could use that different pool of puzzles but common sub-pool of players (having gotten their ratings over at least 30 puzzles from the whole pool of human-puzzle IDs in Lichess) to perhaps transmit some seeding of rating estimate for the new competition events with a different half of pool, the puzzle IDs. Sorry, it would be clearer with a potato diagram and arrows. I stumble in the stream version. of something rather simple otherwise. I guess calibration problem in few words...? My wet brain might be hallucinating. (I guess I might be natural intelligence, then). That is the imagined idea proposed that, I thought, was in the words of the blog. I still do think that it is a methodological question, but it might not be the actual goal. Thinking aloud as usual: But it would not be framed as a data analysis competition. So I guess the title is not what sparked my interest, and it might have been a slippery choice. But I would not blame the authors. This might be the norm, is it not? Limited window of visibility, do what you can to attract relevant attention. At least "puzzle" is in the title. And people might have not computed another title, about calibration experiment. And it might still be true that a side effect of the project might be more informative puzzles. I guess that is why I compare with ELOMETER. And fuss about transparency of reasoning and method. For this to actually be a better offer, it might not just be about rating, but about the specifics of its database content, and possibly coaching theory or something we might find improving our understanding of our experience.

been clicking on many links on the front page. a few are not 404. I am glad that the "source code" link leads somewhere

https://github.com/FrugoFruit90/lilaPuzzles

I looked at the repository, but I don't think the readme file is having any changes about possible modificatinos and purpose of that fork. Maybe authors could give us some hints, about the extent of open access to what they can spare for us given the competition informatino black out (???). Or we can be patient. I think open source code is limited when it comes to such data analysis research projects. So the curious might be left hungry. That would be me.

been clicking on many links on the front page. a few are not 404. I am glad that the "source code" link leads somewhere https://github.com/FrugoFruit90/lilaPuzzles I looked at the repository, but I don't think the readme file is having any changes about possible modificatinos and purpose of that fork. Maybe authors could give us some hints, about the extent of open access to what they can spare for us given the competition informatino black out (???). Or we can be patient. I think open source code is limited when it comes to such data analysis research projects. So the curious might be left hungry. That would be me.

could we download at some points all the problems we tried for further chess work on them for our trouble.

There are some things in the solutions on one problem I can't understand why the longer solution is better than the shorter I found (did fail anyway before, but my thing with puzzle is not at all about win or loss, it is about understanding their point).

I promise I will keep that in private studies and so. For now, it is copy past the list of moves, fortunately I think the character table for my OS or browser for figurine coincide as text with English SAN. or idk. but it works. modulo some reordering. for the variations of my errors.. inclusing that which puzzles me.. :). post-puzzle. the most fun part.. I get to understanding my own misunderstanding of chess in that position...

But a dump button to PGN file of the puzzles. (I guess we can't even do that in Lichess can we, other than scraping the history, and scrolling to the bottom, and maybe hoping the inert database of puzzles has all the puzzles, and then extract all those with the hashes from the scrape? However I guess not for this puzzle dataset. Not yet. I understand. That might be methodological content under competition. Engineering is not science. I confuse the 2 sometimes. Although, data analysis is supposed to be more sciency.

I apologize for being a noisy participant. A small price maybe... :) or not. It comes from my enthusiasm for databases of positions in general, and their potential for characterization, a blind spot of big data machine learning. But now we deal with not clearly well-posed problems like how "far" does a given chess position's set might "range", and then the meta problem of learning about it, in relation to the bigger sets one individual remaining lifetime experience might encounter.

So database properties beyond their number might become a research topic. If not already. But not in chess. I did not see any sign of this kind of questions, other than in my own ramblings. I wish this project will be informative for the community eventually.

It ought to be, like my old physics teachers long ago would pepper the reasoning. As they were not waiting for deduction, only all the time to go ahead. Just saying this is a humble, partially educated hunch of mine. And curiosity, for sure. So dangling it in the blog got me a bit too excited perhaps. I live of small hopes from little clues days at a time on lichess.. between chess boards.

could we download at some points all the problems we tried for further chess work on them for our trouble. There are some things in the solutions on one problem I can't understand why the longer solution is better than the shorter I found (did fail anyway before, but my thing with puzzle is not at all about win or loss, it is about understanding their point). I promise I will keep that in private studies and so. For now, it is copy past the list of moves, fortunately I think the character table for my OS or browser for figurine coincide as text with English SAN. or idk. but it works. modulo some reordering. for the variations of my errors.. inclusing that which puzzles me.. :). post-puzzle. the most fun part.. I get to understanding my own misunderstanding of chess in that position... But a dump button to PGN file of the puzzles. (I guess we can't even do that in Lichess can we, other than scraping the history, and scrolling to the bottom, and maybe hoping the inert database of puzzles has all the puzzles, and then extract all those with the hashes from the scrape? However I guess not for this puzzle dataset. Not yet. I understand. That might be methodological content under competition. Engineering is not science. I confuse the 2 sometimes. Although, data analysis is supposed to be more sciency. I apologize for being a noisy participant. A small price maybe... :) or not. It comes from my enthusiasm for databases of positions in general, and their potential for characterization, a blind spot of big data machine learning. But now we deal with not clearly well-posed problems like how "far" does a given chess position's set might "range", and then the meta problem of learning about it, in relation to the bigger sets one individual remaining lifetime experience might encounter. So database properties beyond their number might become a research topic. If not already. But not in chess. I did not see any sign of this kind of questions, other than in my own ramblings. I wish this project will be informative for the community eventually. It ought to be, like my old physics teachers long ago would pepper the reasoning. As they were not waiting for deduction, only all the time to go ahead. Just saying this is a humble, partially educated hunch of mine. And curiosity, for sure. So dangling it in the blog got me a bit too excited perhaps. I live of small hopes from little clues days at a time on lichess.. between chess boards.