FEN (Forsyth Edwards Notation) • page 6/6 • General Chess Discussion • lichess.org

Basically I was putting on some axis, an ordering of all visible supports about chess games and positions as encoding methods to share information, in other words communication about chess and its world of ideas (not to get bogged down into specifcs).

That ordering was from compressed to exploded encodings. From the least space taking code to the most explicit one.
bitboard, SAN, FEN, jpg, svg, html +CSS as in lichess, and the "basic input vector".

I think that last part may have been causing problems. mea culpa. I was trying not to get bogged into my own terminology or any specific one because i did not need to, given the other items in my order. But that is exactly what some blog about NNue has been using to describe the Neural Net input encoding for chess positions.

All these support can be made to be bijective --- mapped, to each other with enough flexibility, and flags. However, they differ in how much a sensory system is able to process them depending on how it digests the environmental information, is it verbally through a sequence of symbols, or that but also with limited redundancy (compression), or is it visual, like the human visual cortex?

Neural nets architecture are essentially capturing that aspect of the human visual cortex, that all pixel, or retina cells, or optical nerve fibers through the visual cortex layers, are being perceived at the same time, before the conscious mind starts making even the first candidate move thought that can be made verbal (as most conscious thought is, and as logic connectors between statements works). However, even when using graphics card, in training, the parallelism is still compromised by the sequential nature of silicon logic circuits (processor, bus). Hence the speed in perception and sensory-to-conscious first objectification opportunities of the wet brain, is not taken advantage of in execution during play. However, the fact that the encoding in even more exploded than other items (some people call it high level encoding or representation) in my ordered list, is testament to that visual parallelism importance in chess thinking, "d'après moi".

Finally, i made a joke about being spoiled by lichess, because a chess naive ignoramus like me could still do chess and only use SAN for variation navigation without being fluent at all in it, because i chess-grew up without it (like a musician using its ears only), real boards, and now the perfect world, of 2D chess-board, and perfect vertical overview (i use 50 zoom, to have it near best retina field of view, where we have more density).

Finally, FEN, I only last year became aware of its existence, (as of most anything chess online), and was happy to see that only the empty squares were compressed, and that while it could be read (scanned from left to right) as a sequence of characters, it could also be chunk-viewed with a glance, when aligned in some 2D layout.

I hope this non-tweet size reply, to your tweet size request for elaboration (which i misunderstood as bad faith or tweet laziness), is helpful. Sorry for the op or other commenters, if I took space away.

@selfbrain I may have been too quick in my assumptions about your stance w.r.t. my posts. So assuming this thread is into some relative quiescent state, I will try to dissect and elaborate beyond tweet size norm and expose myself to rambling appearance (or deed, as in indeed). Basically I was putting on some axis, an ordering of all visible supports about chess games and positions as encoding methods to share information, in other words communication about chess and its world of ideas (not to get bogged down into specifcs). That ordering was from compressed to exploded encodings. From the least space taking code to the most explicit one. bitboard, SAN, FEN, jpg, svg, html +CSS as in lichess, and the "basic input vector". I think that last part may have been causing problems. mea culpa. I was trying not to get bogged into my own terminology or any specific one because i did not need to, given the other items in my order. But that is exactly what some blog about NNue has been using to describe the Neural Net input encoding for chess positions. All these support can be made to be bijective --- mapped, to each other with enough flexibility, and flags. However, they differ in how much a sensory system is able to process them depending on how it digests the environmental information, is it verbally through a sequence of symbols, or that but also with limited redundancy (compression), or is it visual, like the human visual cortex? Neural nets architecture are essentially capturing that aspect of the human visual cortex, that all pixel, or retina cells, or optical nerve fibers through the visual cortex layers, are being perceived at the same time, before the conscious mind starts making even the first candidate move thought that can be made verbal (as most conscious thought is, and as logic connectors between statements works). However, even when using graphics card, in training, the parallelism is still compromised by the sequential nature of silicon logic circuits (processor, bus). Hence the speed in perception and sensory-to-conscious first objectification opportunities of the wet brain, is not taken advantage of in execution during play. However, the fact that the encoding in even more exploded than other items (some people call it high level encoding or representation) in my ordered list, is testament to that visual parallelism importance in chess thinking, "d'après moi". Finally, i made a joke about being spoiled by lichess, because a chess naive ignoramus like me could still do chess and only use SAN for variation navigation without being fluent at all in it, because i chess-grew up without it (like a musician using its ears only), real boards, and now the perfect world, of 2D chess-board, and perfect vertical overview (i use 50 zoom, to have it near best retina field of view, where we have more density). Finally, FEN, I only last year became aware of its existence, (as of most anything chess online), and was happy to see that only the empty squares were compressed, and that while it could be read (scanned from left to right) as a sequence of characters, it could also be chunk-viewed with a glance, when aligned in some 2D layout. I hope this non-tweet size reply, to your tweet size request for elaboration (which i misunderstood as bad faith or tweet laziness), is helpful. Sorry for the op or other commenters, if I took space away.

dboing

edited

#51

bitboard, SAN, FEN, jpg, svg, html +CSS as in lichess, and the "basic input vector"

actually, I should have put jpg, or other non-vector drawing last. as the least compressed.

which makes be remark, that another criterion besides completeness or memory size or bandwidth constraint (i.e. compression is good), the fidelity of information transmission between sender and receiver in a noisy environment (and sometimes our own words or ideas may be both signal and noise present at the time of reception). Which encoding or representation is more robust to a bit error, at position reception, that which is minimal or most compact, or that which has redundancy? Would a sequence of FENs be more robust to transcription error or the SAN version of the same game? think of databases....

Basically, if that were the constraint the order would end up being the same, from fragile to robust....

bitboard, SAN, FEN, jpg, svg, html +CSS as in lichess, and the "basic input vector" actually, I should have put jpg, or other non-vector drawing last. as the least compressed. which makes be remark, that another criterion besides completeness or memory size or bandwidth constraint (i.e. compression is good), the fidelity of information transmission between sender and receiver in a noisy environment (and sometimes our own words or ideas may be both signal and noise present at the time of reception). Which encoding or representation is more robust to a bit error, at position reception, that which is minimal or most compact, or that which has redundancy? Would a sequence of FENs be more robust to transcription error or the SAN version of the same game? think of databases.... Basically, if that were the constraint the order would end up being the same, from fragile to robust....