so I was wanting to go on a theme self diagnostic spree where I would try to not calculate much, to test my theme awarenss (under random subjective/pattern based themes). But I wanted to also test the idea that the shorter the more a failed puzzle would have fewer theme to fail upon, and hence while lichess is not providing enough feedback that that way (sample sizes for the performance ratings, and population data about all things themes, beyond the automatic taggers, i.e. the human puzzle data that is not in the shared database, puzzle events and thematic human data).
so I just tried the one move puzzles as entry constraint. Duh! I am asking to get frustrated. . The first puzzle I got, what 1 move deep only because it is one of those premature (human chess) stopping, from a perfect-looking flat profile in multi PV at that move.
So. Useless. I can only guess that my success was based on the right feature I thought to leverage. And vote as present, although I can't see any reward any time soon on the board, even using the main PV.
so. It might serve when I miss, but when I succeed, this is boring as hell. Too much oracle faith expected. Not my chess.
Why even play chess. And not let the engine also solve the first move, while at it....
so I just tried the one move puzzles as entry constraint. Duh! I am asking to get frustrated. . The first puzzle I got, what 1 move deep only because it is one of those premature (human chess) stopping, from a perfect-looking flat profile in multi PV at that move.
So. Useless. I can only guess that my success was based on the right feature I thought to leverage. And vote as present, although I can't see any reward any time soon on the board, even using the main PV.
so. It might serve when I miss, but when I succeed, this is boring as hell. Too much oracle faith expected. Not my chess.
Why even play chess. And not let the engine also solve the first move, while at it....
But now wondering. If I fail from mis-evaluating deeper surrogate goals/theme, than the theme most immediate being used? For example, maybe, if the flat profile involves many themes, and I might have foreseen some of those, but misjudged it (as might be my current state of learner internal model, to be improved upon). This would not help at all.. as the theme that anyone might have put, and the Lichess tagger as well, would be only about the immediate move theme association.
so maybe some trade off.. Deeper puzzle sol. Might have less of flat stoppers, in proportions (and other such artifacts of games sourced puzzlers and solution definitions). sure I can vote down.. but that won't make the experience more interesting.
what is the porportion of really short puzzles that are human short, and not such bloops.... no way to tell. I guess I would need to scan the database with engine mulitPV analyeses, and filter for solved position next candidate falt profiles. (hoping). not my rabbit hole priority now. if going back to such toolery, it would have to be more about human chess stuff than working around this construction. I can't understand why they don't filter that themselves. They are aware of those flat profile ones.... I know post puzzle opportunity to see that.. but is it worth the waste of the puzzle learning value for its short solution..
well.. moving on. anyway. maybe a bunch of themes can't be deployed on actual human chess one move puzzles.
so maybe some trade off.. Deeper puzzle sol. Might have less of flat stoppers, in proportions (and other such artifacts of games sourced puzzlers and solution definitions). sure I can vote down.. but that won't make the experience more interesting.
what is the porportion of really short puzzles that are human short, and not such bloops.... no way to tell. I guess I would need to scan the database with engine mulitPV analyeses, and filter for solved position next candidate falt profiles. (hoping). not my rabbit hole priority now. if going back to such toolery, it would have to be more about human chess stuff than working around this construction. I can't understand why they don't filter that themselves. They are aware of those flat profile ones.... I know post puzzle opportunity to see that.. but is it worth the waste of the puzzle learning value for its short solution..
well.. moving on. anyway. maybe a bunch of themes can't be deployed on actual human chess one move puzzles.
so, in the engine tagged puzzle depth section (on theme 2 level flat page), 4 categories.. 1, 2, 3, and 4+
I did 1. only one puzzle.. and got demoralized by remembrance of things past. The non-human calibrated puzzles.
I think it is very simple. I seek a full human puzzle experience. I don't want the lolipop rating increase, I demand to see rewarding things on the boart at the point of solution being declared so. If the skill set (and their levels) needed on average to succeed those few moves in the current non-satistfying puzzles sub-set, is not allowing the last solving position to be compared with the seed position of the puzzle for some that-human-that-just-solved-till-that-last-position ablities to assess the board, then the experience is not calibrated for human population.
it might be for the human developper of the automatic solver segmentatoin criteria... but is that not not the goal or target.
I find compositions, not from our games a lot more rewarding.. So, now that machines can make way more of those, would it kill someone to filter out those puzzles. I would be willing to put my imagination to some level of common lingo discussion about how to automatze such new theme (assessment fog thick in this one, might be the name of that theme. or its meaning, i leave it to better lingo compliant thinkers to find the better one word).
there is already the 3 PV score flatness criteria itself that comes to mind.. it might mean a theme... called solved position flat profile. . we could filter that out of our experience if we had such study need.
but I guess as long as there is not even a simple AND, OR, NOT syntax for the filters.. I might be just musing in my own cavity.
I did 1. only one puzzle.. and got demoralized by remembrance of things past. The non-human calibrated puzzles.
I think it is very simple. I seek a full human puzzle experience. I don't want the lolipop rating increase, I demand to see rewarding things on the boart at the point of solution being declared so. If the skill set (and their levels) needed on average to succeed those few moves in the current non-satistfying puzzles sub-set, is not allowing the last solving position to be compared with the seed position of the puzzle for some that-human-that-just-solved-till-that-last-position ablities to assess the board, then the experience is not calibrated for human population.
it might be for the human developper of the automatic solver segmentatoin criteria... but is that not not the goal or target.
I find compositions, not from our games a lot more rewarding.. So, now that machines can make way more of those, would it kill someone to filter out those puzzles. I would be willing to put my imagination to some level of common lingo discussion about how to automatze such new theme (assessment fog thick in this one, might be the name of that theme. or its meaning, i leave it to better lingo compliant thinkers to find the better one word).
there is already the 3 PV score flatness criteria itself that comes to mind.. it might mean a theme... called solved position flat profile. . we could filter that out of our experience if we had such study need.
but I guess as long as there is not even a simple AND, OR, NOT syntax for the filters.. I might be just musing in my own cavity.
lichess.org/training/themes
Lengths
> One-move puzzle 447,478
A puzzle that is only one move long.
> Short puzzle 2,179,316
Two moves to win.
> Long puzzle 970,888
Three moves to win.
> Very long puzzle 311,330
Four moves or more to win.
Lengths
> One-move puzzle 447,478
A puzzle that is only one move long.
> Short puzzle 2,179,316
Two moves to win.
> Long puzzle 970,888
Three moves to win.
> Very long puzzle 311,330
Four moves or more to win.
I had one hypothesis of puzzle study strategy before. Mrpushwood, is a proponent of such practice to alleviate the artifacts of puzzles with respect to whole game experience. If one a taking puzzles as training for game experience, I guess.
mate puzzles, take the surrogate goal uncertainty out. (kind of same philosophy at long learner time scale of using endgame type of problem seeds in study mode, I have to specify that, even if I am always in that mode, most fo the time when playing, or it has priority, per my instincts and needs for intellectual rewards, really, it is fun for me, within my range (evolving bar), i guess, I might be wired a bit weird that way, compared to visible chess population distribution of such traits, in the modern culture I have been perceiving so far.
But then guess what comes next. Since I still want coverage of all the modern chess culture "consensus" (or at Lichess scale at least, tested by algorithm blindness, and frozen concepts from their programmers, which is objective, so good within).
Can all subjective (not bad, as there are definition theories for us, on top of the tagger definitions which are the programmers subjective attemtps at maknig this objective for us, they also make a human more useful version, which might allow human generalization, that machine code might not, or that programmer, in chess wisdom, would not want to impose as the only possible truth for the whole lichess user population Hence acknowledging the state of the language in tactical chess theory (which is our most consistent so far, but still debatable, if only, we would talk to each other cooperatively and acknowledge that there is fog to "conquer" not alone.. and that boring work (for some) of language efficacy construction or efficaciou construction of languaeg. of efficacious language construction (the union of that, or the meaning I might have, which I am not sure any of those is accurately transmitting... I would use //// but then some people might think I have being precise or accurate about those per their own habits of delimiters, in the world of strings communitcation.. I guess I mightbe meaning ||||. 2 key things peripheral touch typing map.
/// also 2 keys. I know without blinking. cloud of words. while not yet having a common language... I am just "wasting" time looking at many angles too often....
slow most of the time. but then... once in a while.... tada. (I guess this might be long term gambling psychology, I faill most of the time, but I remember better the random times when I was using my memory limitations and the emotion take over the fog left from that. things that happen then get imprinted deeper... ? so there is not only repeated association (as that might get the emotion fading out of picture).. more than one type of associative learning.. one is of high amplitude emotoin short time event, the other seems more of the habituation or fatigue type. Could it just pruning under the radar for global spurious associations, a "substacting" force or pressure... But our intuition is using both. We are just more conscious, given our raw memory time scales of retention (raw sensory memory, maybe).
so. conclusion. there might be fog.. still....
mate puzzles, take the surrogate goal uncertainty out. (kind of same philosophy at long learner time scale of using endgame type of problem seeds in study mode, I have to specify that, even if I am always in that mode, most fo the time when playing, or it has priority, per my instincts and needs for intellectual rewards, really, it is fun for me, within my range (evolving bar), i guess, I might be wired a bit weird that way, compared to visible chess population distribution of such traits, in the modern culture I have been perceiving so far.
But then guess what comes next. Since I still want coverage of all the modern chess culture "consensus" (or at Lichess scale at least, tested by algorithm blindness, and frozen concepts from their programmers, which is objective, so good within).
Can all subjective (not bad, as there are definition theories for us, on top of the tagger definitions which are the programmers subjective attemtps at maknig this objective for us, they also make a human more useful version, which might allow human generalization, that machine code might not, or that programmer, in chess wisdom, would not want to impose as the only possible truth for the whole lichess user population Hence acknowledging the state of the language in tactical chess theory (which is our most consistent so far, but still debatable, if only, we would talk to each other cooperatively and acknowledge that there is fog to "conquer" not alone.. and that boring work (for some) of language efficacy construction or efficaciou construction of languaeg. of efficacious language construction (the union of that, or the meaning I might have, which I am not sure any of those is accurately transmitting... I would use //// but then some people might think I have being precise or accurate about those per their own habits of delimiters, in the world of strings communitcation.. I guess I mightbe meaning ||||. 2 key things peripheral touch typing map.
/// also 2 keys. I know without blinking. cloud of words. while not yet having a common language... I am just "wasting" time looking at many angles too often....
slow most of the time. but then... once in a while.... tada. (I guess this might be long term gambling psychology, I faill most of the time, but I remember better the random times when I was using my memory limitations and the emotion take over the fog left from that. things that happen then get imprinted deeper... ? so there is not only repeated association (as that might get the emotion fading out of picture).. more than one type of associative learning.. one is of high amplitude emotoin short time event, the other seems more of the habituation or fatigue type. Could it just pruning under the radar for global spurious associations, a "substacting" force or pressure... But our intuition is using both. We are just more conscious, given our raw memory time scales of retention (raw sensory memory, maybe).
so. conclusion. there might be fog.. still....
ok back on track.. Mate is that a solving strategy given my buried goals above... (don't hurt yourself too much, I did digress, dear putative, in abstract, reader, I don't re-read myself, past the typing task, I might have AFKs echeos later, prompting me to make "amends" through more ramblings though).
Mate enough covering of all learnable tactical themes? I can get the depth control to objective satisfaction within my assessment fog on my current state of chess skills (I guess one might want "pulsating" object for that, once the ambient space working hypothesis can be agreed for long enough to show results or interesting stuff).
other problem: the more one puts constraints on the population of puzzles, the more risk of difficulty coverage being sparse.
try the one move mates for examples.. Well, it might be because that is very close to the logic of the ruleset.. and our spatial human skills being not that dependent on experience pruning as much.
There is also the problem of resignation in human games... I wonder if lichess would have more mate puzzle wealth if it continued automatically all the resigned games... (it might still stall a lot, since a mate is a mate is a mate, and flat profiles might abound, unless putting the administrative ROT of shorter mate is better on top of the core ruleset.
I already made an argument for long puzzles to not be that good for thematic weakness specificity, for the measure based only on the win loss variables. (the AND for win, and the OR for loss, makes a loss less thematic specific.. so one needs a lot more samples, and then the human biology fluctuations might come into play... and lichess cutoffs by time the thematic history (dashboard).
I guess I might just go blind and "have fun"...
Mate enough covering of all learnable tactical themes? I can get the depth control to objective satisfaction within my assessment fog on my current state of chess skills (I guess one might want "pulsating" object for that, once the ambient space working hypothesis can be agreed for long enough to show results or interesting stuff).
other problem: the more one puts constraints on the population of puzzles, the more risk of difficulty coverage being sparse.
try the one move mates for examples.. Well, it might be because that is very close to the logic of the ruleset.. and our spatial human skills being not that dependent on experience pruning as much.
There is also the problem of resignation in human games... I wonder if lichess would have more mate puzzle wealth if it continued automatically all the resigned games... (it might still stall a lot, since a mate is a mate is a mate, and flat profiles might abound, unless putting the administrative ROT of shorter mate is better on top of the core ruleset.
I already made an argument for long puzzles to not be that good for thematic weakness specificity, for the measure based only on the win loss variables. (the AND for win, and the OR for loss, makes a loss less thematic specific.. so one needs a lot more samples, and then the human biology fluctuations might come into play... and lichess cutoffs by time the thematic history (dashboard).
I guess I might just go blind and "have fun"...
there might be some soft spot amidst all the factors I just mentioned (or thought I did).
ok. combining. I might not have finnished process my factors yet.
puzzle premise (and automatic detection) is SF blunder detection. a mate puzzle is therefore something that lichess either knows from the games having been concluded itself using the mate threat and either got resigned post-blunder by the blundering side, or was mated.
but it might be a bunch of non-human (the 2 of the game being processed) mates that SF saw, that were not in the game.. ok.. thought I had some closure and could dismiss the concern. now still fog. inconclusive.
I guess the first questions, is what is the impact (data wise) of resignation onto the questions of mate puzzles extraction from our games.
i thought the premise of puzzle selection might make my concern moot.. but now.. either out of stamina to keep computing myself (fallibility acceptation here, wink, wink), or there is still room for that question. this was an afterdoubt... tired. for a while.
puzzle premise (and automatic detection) is SF blunder detection. a mate puzzle is therefore something that lichess either knows from the games having been concluded itself using the mate threat and either got resigned post-blunder by the blundering side, or was mated.
but it might be a bunch of non-human (the 2 of the game being processed) mates that SF saw, that were not in the game.. ok.. thought I had some closure and could dismiss the concern. now still fog. inconclusive.
I guess the first questions, is what is the impact (data wise) of resignation onto the questions of mate puzzles extraction from our games.
i thought the premise of puzzle selection might make my concern moot.. but now.. either out of stamina to keep computing myself (fallibility acceptation here, wink, wink), or there is still room for that question. this was an afterdoubt... tired. for a while.
I think I get swallowed into a tunnel vision mode when having to use language for my thinking to the point of making apparently readable strings or one stream of such.... It seems I do my better thinking when I am not trying to explain through this torture. i.e. AFK.
It seems my mind eye gets liberated too. (or is it my need to mask the increasing signals from sustained posture?, or is it that there is no single reason... ahhhh, how could it be, I need a single factor, or I will go drooling vegetative state).
Well... Small brains need what they can perceive as clues, and if the small brain is not completely inadequate for its niche, and comes from a social animal evolution, and made it to the point where it can pull on that social thing called communication, and the individual instance is not too damaged or stuck in own cavity, in a hermetical self-consistent shrunken universe, well, then, 2 or more small brains can make for a better externally persistent brain produced culture. (no medals required to call it that).
So duh! ?
It is quite possible that the puzzler scanner, in our games, is only looking for blunders and own PV with mates. It could care for nothing about whether the game continuation from blunder did go to final core rule set consumption. Or not. It has its own world of chess. A world which we dare not ask about the specifics off, well, I did try, but now, how long can one act the Don Quixote, or the loopy clown? Or the single hero resisting the blob thing (I am thinking of the island in the TV series. The prisoner. The truth comes in the form of a white balloon, whenever the character tries to escape. First version of the series).
But that is one single such balloon here. The belonging to the perceived dominant group culture. At some point, the own reasoning ability, becomes a defect (question, always). I might prefer to think I am deluded. Or that we are a few deformed perspective ones like that. The reason of the group. The tension of the individual and the group (projected but also having the final tangible consequences, that we are bound to). The isolated mind.
It might be a fluctuating thing.
It seems my mind eye gets liberated too. (or is it my need to mask the increasing signals from sustained posture?, or is it that there is no single reason... ahhhh, how could it be, I need a single factor, or I will go drooling vegetative state).
Well... Small brains need what they can perceive as clues, and if the small brain is not completely inadequate for its niche, and comes from a social animal evolution, and made it to the point where it can pull on that social thing called communication, and the individual instance is not too damaged or stuck in own cavity, in a hermetical self-consistent shrunken universe, well, then, 2 or more small brains can make for a better externally persistent brain produced culture. (no medals required to call it that).
So duh! ?
It is quite possible that the puzzler scanner, in our games, is only looking for blunders and own PV with mates. It could care for nothing about whether the game continuation from blunder did go to final core rule set consumption. Or not. It has its own world of chess. A world which we dare not ask about the specifics off, well, I did try, but now, how long can one act the Don Quixote, or the loopy clown? Or the single hero resisting the blob thing (I am thinking of the island in the TV series. The prisoner. The truth comes in the form of a white balloon, whenever the character tries to escape. First version of the series).
But that is one single such balloon here. The belonging to the perceived dominant group culture. At some point, the own reasoning ability, becomes a defect (question, always). I might prefer to think I am deluded. Or that we are a few deformed perspective ones like that. The reason of the group. The tension of the individual and the group (projected but also having the final tangible consequences, that we are bound to). The isolated mind.
It might be a fluctuating thing.
Join the Dboing's Musings team, to post in this forum