Shashin theory

[noobpwnftw]

I think it is fair to say that humans can summerize certain chess knowledge and use it to teach another human on the basics.

Up until a certain skill level you'll realize that it is often more of a hunch than a theory. I'm no GM or IM to comment on how the actual "shashin theory" is of any value at that level. Meantime chess engines can be made with the same approach until you need SPRT to see if some idea is really a valid one. People in denial of those things often share a same reason.

BTW I'm fine with let's say someone just yoink Crystal and claim it to be "tactically" suprior amid the elo loss, or insisting on implementing his own ideas regardless. But putting words into someone else's mouth in order to justify some zero effort copy paste is one next level of nonsense.

Most people see clowns as clowns, but when two clowns meet they may see inspiration, it's always a matter of perspective I guess.

[Viz]

The very short and simple answer is that it's a modal way of thinking about chess. I know one of the top trainers in the world uses a similar (but different) appraoch - not at all using Shashin's ideas, but using a modal approach of his own.

1) The whole thing depends on a correct and deep position evaluation.
2) This then throws you into one of several modes (3 in Shashin's case).
3) You then do a search based on this mode.

For humans the modal approach makes a lot of sense because it has been shown that if you tell someone there is a forced win (for example) in a position they are far more likely to find it. However, Shashin's own implimentation simply is not practical in over-board-games because the evaluation function he proposes is far too slow for a human since it requires a lot of counting. It could work at correspondence chess though.

For engines...I'm not sure. It seems a bit false / artificial if you ask me.

Thank you very much, Mr Werewolf - you apparently read the book itself (sth I was not ready to do). I didn't find anything nearly as clear as your post on the internet.
(Also thank you to Mr noobpwnftw for describing what I may find when I looked at the actual implementation in a chessprogram if I'd spent a lot of time and effort to understand).
I have always been fascinated by purely human efforts from and for humans to explain chess in a mostly algorithmic way. When I was a young teenager, there was a German IM named Opfermann ( a name that is kind of funny in itself if you know German) who managed to get his very personal theory on chess published as a cheap paperback in supermarkets, so he definitely wasted the time of tens of thousands of German speakers. He was all about mobility and space control, you were supposed to put little thingies on the board during training and count them to find out who is better. And so I did for some time, until I realized that his theory was nonsense (this one really was, Shashin may be way better).

No human player uses Shashin theory in practice, no engine uses it as well until you count it stockfish clones that use it to lose double digits of elo.
All this vague theories simply don't work, period.

[Guenther]

The very short and simple answer is that it's a modal way of thinking about chess. I know one of the top trainers in the world uses a similar (but different) appraoch - not at all using Shashin's ideas, but using a modal approach of his own.

1) The whole thing depends on a correct and deep position evaluation.
2) This then throws you into one of several modes (3 in Shashin's case).
3) You then do a search based on this mode.

For humans the modal approach makes a lot of sense because it has been shown that if you tell someone there is a forced win (for example) in a position they are far more likely to find it. However, Shashin's own implimentation simply is not practical in over-board-games because the evaluation function he proposes is far too slow for a human since it requires a lot of counting. It could work at correspondence chess though.

For engines...I'm not sure. It seems a bit false / artificial if you ask me.

Thank you very much, Mr Werewolf - you apparently read the book itself (sth I was not ready to do). I didn't find anything nearly as clear as your post on the internet.
(Also thank you to Mr noobpwnftw for describing what I may find when I looked at the actual implementation in a chessprogram if I'd spent a lot of time and effort to understand).
I have always been fascinated by purely human efforts from and for humans to explain chess in a mostly algorithmic way. When I was a young teenager, there was a German IM named Opfermann ( a name that is kind of funny in itself if you know German) who managed to get his very personal theory on chess published as a cheap paperback in supermarkets, so he definitely wasted the time of tens of thousands of German speakers. He was all about mobility and space control, you were supposed to put little thingies on the board during training and count them to find out who is better. And so I did for some time, until I realized that his theory was nonsense (this one really was, Shashin may be way better).

No human player uses Shashin theory in practice, no engine uses it as well until you count it stockfish clones that use it to lose double digits of elo.
All this vague theories simply don't work, period.

Actually it is the first homeopathic chess program.

While there still can be a positive placebo effect inside the body of Human, if he strongly believes in it, it is unknown a user/developer might transfer his belief into a program outside of his body running in a computer. Creationists might disagree.

[Peter Berger]

I think it is fair to say that humans can summerize certain chess knowledge and use it to teach another human on the basics.

Up until a certain skill level you'll realize that it is often more of a hunch than a theory. I'm no GM or IM to comment on how the actual "shashin theory" is of any value at that level. Meantime chess engines can be made with the same approach until you need SPRT to see if some idea is really a valid one. People in denial of those things often share a same reason.

BTW I'm fine with let's say someone just yoink Crystal and claim it to be "tactically" suprior amid the elo loss, or insisting on implementing his own ideas regardless. But putting words into someone else's mouth in order to justify some zero effort copy paste is one next level of nonsense.

Most people see clowns as clowns, but when two clowns meet they may see inspiration, it's always a matter of perspective I guess.

I'll ask the obvious next question.

Given that we have developped some vague general understanding what a program that implemented the Shashin theory may be doing: is this what Shashchess is somehow trying to do indeed? I'd expect like 3 different search modes based on a deep evaluation etc.

If yes - it may be a bad idea or badly done, but it would be a genuine effort. If no: this would be something else.

[noobpwnftw]

Which I already answered, I see not a single set of actual search parameters that is original nor a set of conditions of pruning/reduction logic that is not yoinked from Crystal. While this leaves us what exactly this vague Shashin theory is supposed to be implemented up to interepetation, it doesn't seem to be a genuine enough effort if all that does is diving into some Crystal code snippets which has nothing to do with the said theory.

At least it is to my understanding that Shashin theory is not about taking tweaks of NMP or LMR from Crystal while leaving everything else completely unchanged, i.e. NNUE and all that. One may of course do such things, but claiming for example applying some board symmetry is an implementation of quantum entanglement and therefore it could be called SchrödingerChess is quite a strech.

I find such behavior very typical in scam schemes, they tend to stick as many names and terms to them as possible without having anything to do with those. People fall for that when one or more of those terms pushed a button in their heads and turned their brains off.

[Viz]

guy who "coded" "shashin theory" didn't know what PvNode is, for example. Do you trust on him doing literally anything meaningful?

[Graham Banks]

I think it is fair to say that humans can summerize certain chess knowledge and use it to teach another human on the basics.

Up until a certain skill level you'll realize that it is often more of a hunch than a theory. I'm no GM or IM to comment on how the actual "shashin theory" is of any value at that level. Meantime chess engines can be made with the same approach until you need SPRT to see if some idea is really a valid one. People in denial of those things often share a same reason.

BTW I'm fine with let's say someone just yoink Crystal and claim it to be "tactically" suprior amid the elo loss, or insisting on implementing his own ideas regardless. But putting words into someone else's mouth in order to justify some zero effort copy paste is one next level of nonsense.

Most people see clowns as clowns, but when two clowns meet they may see inspiration, it's always a matter of perspective I guess.

I'll ask the obvious next question.

Given that we have developped some vague general understanding what a program that implemented the Shashin theory may be doing: is this what Shashchess is somehow trying to do indeed? I'd expect like 3 different search modes based on a deep evaluation etc.

If yes - it may be a bad idea or badly done, but it would be a genuine effort. If no: this would be something else.

You could always investigate the differences by ticking each setting in the parameters individually, I guess.

[Peter Berger]

You could always investigate the differences by ticking each setting in the parameters individually, I guess.

I am not sure what you are getting at exactly, Graham - and where these personalities and your posted image of a selection box should be coming from. I don't see them on my computer at all.

I kind of agree if there were exactly three personalities Capablanca, Petrosian and Tal that influenced the search all the time (even if done most clumsily) and that were used and chosen from by some sort of "Shashin theory", this may be seen as an effort of implementing sth like what is claimed.

On my computer I see the following UCI parameters that may be somehow related:

Variety, Concurrent experience, High Tal, Middle Tal, Low Tal, Capablanca, Low Petrosian, Middle Petrosian and High Petrosian.

ALL these parameters are set to off by default in Shashchess 36.

In the README we get:

"Shashin section

Default: no option settled The engine will determine dynamically the position's type starting from a "Capablanca/default positions". If one or more (mixed algorithms/positions types at the boundaries) of the seven following options are settled, it will force the initial position/algorithm understanding If, in the wdl model, we define wdl_w=Win percentage, wdl_d=Drawn percentage and Win probability=(2*wdl_w+wdl_d)/10, we have the following mapping:
Win probability range Shashin position’s type Informator symbols
[0, 4] High Petrosian -+
[5, 9] Middle-High Petrosian -+ \ -/+
[10,12] Middle Petrosian -/+
[13,19] Middle-Low Petrosian -/+ \ =/+
[20,24] Low Petrosian =/+
[25,49] Caos: Capablanca-Low Petrosian =/+ \ =
[50] Capablanca =
[51,75] Caos: Capablanca-Low Tal = \ +/=
[76,80] Low Tal +/=
[81,87] Low-Middle Tal +/=
[88,90] Middle Tal +/-
[91,95] Middle-High Tal +/- \ +-
[96,100] High Tal +-
N.B. The winProbability also take into account the depth at which a move has been calculated. So, it's more effective than the cp."

This decription is not understandable for a mere mortal.

[Graham Banks]

You could always investigate the differences by ticking each setting in the parameters individually, I guess.

I am not sure what you are getting at exactly, Graham - and where these personalities and your posted image of a selection box should be coming from. I don't see them on my computer at all.

I kind of agree if there were exactly three personalities Capablanca, Petrosian and Tal that influenced the search all the time (even if done most clumsily) and that were used and chosen from by some sort of "Shashin theory", this may be seen as an effort of implementing sth like what is claimed.

On my computer I see the following UCI parameters that may be somehow related:

Variety, Concurrent experience, High Tal, Middle Tal, Low Tal, Capablanca, Low Petrosian, Middle Petrosian and High Petrosian.

ALL these parameters are set to off by default in Shashchess 36.

In the README we get:

"Shashin section

Default: no option settled The engine will determine dynamically the position's type starting from a "Capablanca/default positions". If one or more (mixed algorithms/positions types at the boundaries) of the seven following options are settled, it will force the initial position/algorithm understanding If, in the wdl model, we define wdl_w=Win percentage, wdl_d=Drawn percentage and Win probability=(2*wdl_w+wdl_d)/10, we have the following mapping:
Win probability range Shashin position’s type Informator symbols
[0, 4] High Petrosian -+
[5, 9] Middle-High Petrosian -+ \ -/+
[10,12] Middle Petrosian -/+
[13,19] Middle-Low Petrosian -/+ \ =/+
[20,24] Low Petrosian =/+
[25,49] Caos: Capablanca-Low Petrosian =/+ \ =
[50] Capablanca =
[51,75] Caos: Capablanca-Low Tal = \ +/=
[76,80] Low Tal +/=
[81,87] Low-Middle Tal +/=
[88,90] Middle Tal +/-
[91,95] Middle-High Tal +/- \ +-
[96,100] High Tal +-
N.B. The winProbability also take into account the depth at which a move has been calculated. So, it's more effective than the cp."

This decription is not understandable for a mere mortal.

From what I understand, it uses all three settings, plus the intermediary ones, switching between them based on the evaluations at any given time.
I also believe that it is not recommended to use just one of the three settings.

My thinking was that if I started a game with a given opening line and tried playing that line separately with each setting, that might indeed show a different output for each which would prove that they are somehow being used.
I use ChessGUI and it shows that tick box, so I might try this tomorrow and post feedback here.

[Peter Berger]

My thinking was that if I started a game with a given opening line and tried playing that line separately with each setting, that might indeed show a different output for each which would prove that they are somehow being used.

Ah, now I understand your point. Well, I don't think anyone wants to really claim that Shashchess is just Stockfish. (how you get these boxes still amazes me).