﻿﻿ MatPlus.Net

Website founded by
Milan Velimirović
in 2006

5:19 UTC
 ISC 2022

Remember me

 CHESS SOLVINGTournamentsRating lists1-Jan-2022
 B P C F

MatPlus.Net Forum General Our incomplete problem chess world

Page: [Previous] [Next] 1 2 3 4
(61) Posted by Kevin Begley [Thursday, Mar 21, 2013 13:08]; edited by Kevin Begley [13-03-21]

@Georgy (and Nicolas),

>...we may say that we are playing on the same minefield.

I should feel honored, but tap dancing would be too risky. :-)

>BTW, you have drawn the traditional well-known line between "heterodox" and "fairy".

I'm not sure I'm familiar with the "well-known" line to which you refer...
I'd be in your debt if you could enlighten me...

>Dmitri has shown that any stipulation may be reworded as standard #n with a set of conditions (maybe some of them fairy, some of them heterodox).

I do not agree he has shown anything of the sort.
As previously stated, the hypothetical invention he calls a "fairy condition" is entirely invalid; and, what he had intended to show depends entirely upon that condition.

This proves that having no fundamental definition for the term (fairy condition), and no standard by which to determine the validity of any proposal, allows all kinds of unsound claims.

Consider the following hypothetical fairy condition:
"Any stipulated #1 changes to #2."

With this, Dmitri may claim to be first to show many themes previously impossible in a fairy #1.

And, here is a fairy condition to magically render all problems into orthodox #2:
"All fairy conditions (including this one) are ignored, and this problem counts as orthodox #2."

NEWSFLASH: I am proud to announce here, I achieved the impossible orthodox h#5 theme (moments ago)!!
Where do I apply to collect my \$100?
Wait... strike that -- I must settle for only \$50 (crediting Dmitri for the magical, invalid fairy condition).

This is what puts you in a minefield -- only fundamental definitions can safely extract you.
To that end, I'm here to help.

>Where is the obvious line which allows honestly and fundamentally distinguish, when we should apply the conditions and when - change the stipulation?

We have been experiencing some communications challenges, but this question exceeds my limits.

Until we resolve that, I'd like to take the opportunity to address a few issues (some of which may help improve your question)...

WARNING: many issues here -- this will not be brief.

I have explained that fairy elements are fundamentally a rule book which only govern legal move options.
Like an algorithm, which functions only to determine legality a game.
Output:
1) a set of legal move options (can be null, indicating termination of the game), and
2) an error status.
Inputs:
1) a set of retro conventions (if null, retro plays no role),
2) a valid position (note: validity may depend upon its legality, if retro conventions present demand),
3) a set of fairy elements (can be null), and
4) a base set of rules, and

Nicolas is primarily interested in designing a minimal base set of rules (4), and using this instead of the faulty "orthodox" base (which suffers from excess complexity, and constant evolution).
There is merit in this endeavor, because today's "orthodox" is only tomorrow's "fairy condition."
I want to be supportive, but there are three reasons to be skeptical that this will succeed:

1) Neither Nicolas nor I (nor any group of us) can hash out an entirely arbitrary set of decisions, alone.

It is true that we could establish our own base rule set (useful only in our own journals).
But, I'm not interested in creating another new language, I'm interested in improving our universal language.
Thus, the governing authority (I would first recognize PCCC) must be presented each arbitrary option, such that they may be sanctioned independently (by vote).
The best he and I can do would be to provide a proposal (covering a broad spectrum of arbitrary options), and include with it our own set of recommendations (including our analysis for choices, and citing a broad cross section of experts: composers, solvers, programmers, editors, problem database managers, etc).

He and I are unlikely to engage in such a beneficial endeavor, until PCCC delegates acknowledge (by vote) that such an effort would be welcomed!!
Get PCCC delegates to MOVE on this, and you'll see a flood of volunteered improvements.

2) There is a concern that the base rules will need to evolve (much like "orthodox" does).

If you want to establish a constant base, you have only one shot at it.
You have to produce a valid and enduring base, which foresees all future developments and improvements (erosion)... and, you have to produce it on your first try.
Remember: you're depending upon delegates (with their own motivations) to vote on every arbitrary element.

This is like asking politicians (many of whom may be corrupt!) to vote on details in an engineering schematic, for a structure intended to outlast the pyramids.
You are more likely to obtain the blueprint for the Winchester Mystery House (which is not worthy of a single visit).

Inevitably, there will be unforeseen changes necessary.
Therefore, the first step in designing a minimalist base set of rules must be to provide a contingency.
Problems anchored to this base must be seismically protected from the shifting foundation!

Such a design may require considerable planning.
I doubt anyone has given much thought to establishing backward compatibility for shifting base rules.
If they had, some would be attempting to apply this idea to "orthodox."

3) Such an arduous uphill battle requires highly talented individuals, whom will likely discover, in the process, that there is better value in drifting away from chess, entirely.

Why would a talented individual want to burden themselves with improving OUR chess problem world?
The atmosphere is inhospitable to all such efforts.
And, the time it takes a person to realize that this situation will not change, in my experience, depends upon three key factors:
1) the depth of their passion,
2) their threshold for abuse, and
3) how misspent were their talents.

I would be the first to join an alternative federation.
But, I'd want some assurance that it could not spiral toward the same dysfunction.

I would insist on a codex which provided a fundamental definition for any division (and judge jurisdiction).
I would insist on creating a panel of experts, to consult delegates on sanctioning issues.
...that it be a "problem federation" -- not an "ortho chess fed (with a small cellar for retros and fairies)."
...that all problems earn points fairly.
...that proximity to the BASE RULES should be considered ideal, and instruct judges to weight the deviations according to their divergence from BASE RULES (absolute economy).
I would never require a composer to submit a problem to an album (be it a chronicle, or a sanctioned title competition).
I would insist upon no more than ONE competition for titles (not 9 autonomous and independent competitions).

...and I'm sure I'm forgetting a few dozen things that should not be tolerated.
Plus, I bet others could substantially add to this list.

If you want a better problem chess world, the best course would be to collaborate on a large rocket!
The CO2 levels on this chess planet are dangerously unhealthy (perhaps even to the tipping-point).

>I can answer this question myself, but my opinion is subjective and requires understanding beyond beginner's level.
>So, it is not fundamental and may be not even honest, but it is very practical)).)

Please restate -- I would very much like to learn what your intended question was (along with the comments here).

I am surprised by some of the experienced composers questioning the utility/need for a Fairy Codes or more precisely a Chess Problem Codex covering all types of chess problems.

While searching for a precedent for Superguards chess I found that the only sources were the Popeye, Winchloe and many problem websites. I believe that this information (what are the existing fairy conditions and their definitions) should be in our Federation website, not elsewhere.

Well, I fully respect anybody doing fundamental research on chess composition. But please consider these two things:
1) This field is not as easy as for example defining figures in figure skating. The movements there are physically limited while the conditions, pieces etc. in chess composition are only limited by the human mind.
2) There is a classification that is used by almost any chess problem magazine and also Popeye with only few deviations. Anybody working on that issue must hold out questions why there is need for something to be added to it. And he must tolerate the completely legitimate question: Cui bono?
And what is the point in showing that a helpmate can be defined as a direct mate with a condition?

@Raman: I agree that the federation's website should contain as much definitions as possible. Furthermore there should be a process that somebody inventing a new condition or piece can register it there. But after all I have seen here this does not meet the point that Kevin, Georgy, Dmitri and Nicholas are talking about.

@ Kevin,

Yes, this is true that a « chess problem theory » can’t take into consideration each possible feature. But it is also the case of any theory, no?

The “base” (to take your own word) should be enough general to embrace an amount of cases, but not too general to remains useful in practice. So some kind of choice is needed.

Consider the field of possible captures for example. This is a question whether or not the basis should include “stacking” on the same square. Another question relies on castling, where putting the frontier of castling possibilities, only for the couple (K, R), also for couples (Royal Piece, R)? and so on…

Generally speaking, a feature which doesn’t fit the general principles is called “exotic”. Obviously each composer may work in an exotic context, but the one who want to work in a structured context can follow the basis.

When Roberto Silvio and I classified “future proof games”, we add to face this difficulty of constructing a well-chosen basis. For example “tempo move” is not part of our basis of themes, as we were unable to find a correct definition for it. So you are right in claiming that the basis may change. Indeed this is exactly what we claimed too in the article about classical future proof games:

"Let LT be a list of themes. A proof game G is called a classical future proof game RELATIVE to LT if..."

@Nicolas,

I honestly did go down this road, just a few years ago, looking for the most minimal base I could find (which all rules would refer to, and judges would consider optimal).

On the way down, I was considering how to apply backward compatibility models that I have encountered in other disciplines.
What worried me was: most of what I'd encountered had already failed (sometimes several times, and within a very short time span)!

But, when I reached the absolute minimum set of base rules, the entire minimalist pursuit had peeled away with the onion!

Brace yourself...
The most economical set of base rules possible is: Null. Nada. Nothing. Zip. Zero. Zilch.
It was so obvious, that I had missed it completely!
...steering for something, on roads destined to nothing.

Perfection is an empty base class.

When you apply this vacuum, you immediately solve the problem of backward compatibility.
All rules become stand-alone, self-contained, and autonomous ... none are orthodox, none are fairy.
All you have are rules.

When one rule shatters into two, you have two rules (reword, rename, reclassify, and problems are preserved).
The best you can do is to organize these rules according to families.
e.g., Rebirth Family, Idle-Mover Family, etc.

And, you can insist on better naming conventions (according to families).
For example, Take&Make falls under the anti-Circe branch of the rebirth family.
The default version should be named "Anticirce T&M" (and governed according to defaults of that branch, except for the minor rule alteration necessary to determine the rebirth square).
And, realization of the commonly used version only requires the application of some additional fairy conditions (e.g., pawns can not be reborn on the 1st rank) -- which are likely to be useful on a broader scale!
Realizing this, it becomes obvious that a chiral version falls under the Circe branch of the same family: "Circe T&M" (where the captured unit is reborn according to the movement pattern of the capturing unit).

Formally structured fundamental organization is not only valuable unto itself.
It doesn't only serve beginners (in understanding chess problems).
An organized desk drawer doesn't just provide a better view of its contents -- it gives meaning to the drawer, and to the desk; and, this allows us to see new patterns and possibilities, formerly obscured by clutter.

The point is not only to improve our incomplete problem chess world...
Our struggle against chaos and entropy is what brings meaning to our lives.

I think Milan Velimirovic knew this better than anyone!
Read his articles, click his software -- he goes to extraordinary lengths, just to ORGANIZE chess problems, for our benefit...

But, I digress...
Anyway, after passing through the emptiness of the base rule, I turned my attention to unlearning chess, in order to rediscover chess problems through the eyes of a problemist.
I believe this path leads to real improvement.

As each element must have unique coordinates in a given basis, it is worth to notice that if such a basis contains help-self stipulation and reflex conditions, then the help stipulation is not part of this basis! Indeed “help” is nothing but “help-self with black reflex” (modulo inverted colors).

After some conversation with Nicolas, I am now convinced that reflexmate (indeed any "reflex-aim-n" problem) is a compound stipulation, not a fairy condition.

The reason: A fairy condition is a set of rules, which can not be altered by anything in the stipulation (including the aim).

Thus, I believe reflex translates into a complex goal: "h-aim-1 (obviously for black) AND NOT direct-aim-1".
The iteration is tricky, because you want to iterate only on the "s-aim-1" portion, not the rest (iterating on "NOT direct-aim-1" makes no sense).

Therefore, I believe the correct reflex translation is:
"reflex-aim-(n+1)" = "direct-[ [h-aim-1 (for black)] AND NOT (direct-aim-1) ]-n".
"semi-reflex-aim-(n+1)" = "direct-[ h-aim-1 (for black) ]-n".

And, both reflex- and semi-reflex- follow all rules of orthodox chess (read: no fairy condition necessary).
Quite surprising (I had known the semi-reflex case was orthodox, but did not expect this for the full-reflex case)!
Credit Nicolas for this realization -- I really only contributed the idea of a stipulation built upon compounded goals (which, I think, demonstrates the power and stability of a fundamental treatment -- hopefully, this can be further refined and improved by interested problemists).

Also, note: compound aims are quite common (e.g., h-#/=-n = help checkmate or stalemate in n moves, h-not#-2 = help not checkmate in 2, etc).
In the case of reflex, it is GOALs (not AIMs) which are compounded by combinational logic.

By the way, the translation sets a deadline to reach a goal in n moves, where the black pieces oppose the objective (goal + deadline).
Upon successful resolution of the goal, you iterate only on the bracketed section of the goal.
That is, solve a "[h-aim-1 (for black)]", but ignore the "(NOT direct-aim-1)" portion of the goal.
Maybe there is a better way to express this, but for now, this at least facilitates a an unambiguous expression.

ps: Dmitri's last post, along with Georgy's references to it, contain an obvious error.
Both continually insist that there are two players in a chess problem.
That is how a chess player sees a chess problem.
As I have said before, there are not two players in a helpmate, or directmate, or any formal stipulation -- there is only ONE solver, with one objective (goal/aim and deadline), where one color of pieces (generally white, by default) must succeed, and the other color is given some stipulated PLAY (with respect to the goal/aim and deadline: direct-opposition, or help).

Given this view, it becomes obvious why Dmitri has misread the rules of promotion in circe forms.
Not his fault -- the rules are often poorly expressed (as if it were a variant GAME).
Rewrite the rules, ignoring players (only as the legality of movement -- the physics of the chessmen), and Dmitri's entire argument is reduced to nothing.

The trick is to unlearn everything you know about chess, and start thinking like a problemist.
There are not two players.

Open up a game-board on your computer.
Take turns randomly with somebody, moving the mouse.
The rules of movement are defined by the game-board.
Can the game-board regulate movement? No -- it has no way to know who controls the mouse.
Can the game-board adjudicate claims? No -- it can not even hear claims.

The only thing the game-board knows is the condition(s) -- the physics of the chessmen.
All rules (and all fairy conditions) must be described in this manner.
That includes promotions by rebirth, under circe conditions.

They must be defined in terms of pieces, and the motivation of pieces, or moves, and the selection of moves according to motivation.

For example, in PWC, wRa1xbPa7 [+bPa1=?]
A chess player would ask: which player promotes the black pawn?
A problemist knows there are no players -- either the rules provide the white units the opportunity to select a full move (including the promotion on rebirth) according to whatever their motivation w/2 objective (goal/aim and deadline), or the rules provide that the promotion is only determined by the color of the units, according to their motivation.

Remember: the problem solver does not drag an opponent into a solving tourney.
Therefore, a problem can not be expressed in terms of an opponent.

The bigger question: is WFCC a federation of problemists, or is it a federation of chess players?
The name says "chess composition" but how we chose to define that term speaks volumes about whom we are.

I'm unable to answer to your last post, dear Kevin, because I don't know what "reflex" means exactly...

What is clear is that "reflex" is very different when applied to the terminal side or to the other side. When applied to the terminal side, "reflex" is fully clear, this is just allowing the terminal move to be "helped". As an example, white reflex in a helped problem is empty (redundant) as the terminal move is already helped.

Difficulties come from "reflex" applied to the non-terminal side. I remarked that the "Problemesis" definition is not the one used by solving programs! Indeed suppose white is the terminal side, the "black reflex" definition is "if black on-move can reach the goal, then black must reach it". Consider the following scheme:

(= 2+1 )

a) hx2
b) hx2 black reflex

In a) solutions are:

1.Fd4 ç5 2.F×ç5 b×ç5
1.Fé5 ç5 2.Fd6 ç×d6

In b) the solution is (according to WinChloe):

1.Fé5 ç5 2.Fd6 ç×d6

The other solution disappears (also for Popeye) even if it fulfills the black reflex definition... My feeling is that solving programs are using a slightly modified definition, something like "black on move can't have the possibility of reaching the goal".

It thus appears that, before classifying reflex as a stipulation or as a condition, we have to know which is its correct meaning: the one provided by the Problemesis definition or the one provided by solving programs?

>...before classifying reflex as a stipulation or as a condition, we have to know ....

I agree that it is important to nail down the original definition, as provided by Benjamin Glover Laws, in British Chess Magazine, 1893.

However, I don't agree that there need be any debate as to whether this is a fairy condition, or a stipulation.
A fairy condition is a set of rules, which govern the legal moves of the pieces, regardless of any stipulation.

I used to think a reflexmate was a fairy condition, because it alters the rules of movement (constraining white pieces to only play the mate in 1, whenever possible); and, I thought semi-reflex mate need not be expressed by any condition -- since constraint upon black units may be ignored, by iterative analysis of the goal (a sub-stipulation).

After some correspondence with you, I see now that I was wrong: reflex stipulation can NOT be described by a fairy condition!
Why? Because a fairy condition MUST be independent of the stipulation!

Consider problems with a r#2 / r=2 (translated into fairy conditions and properly formed formal stipulation).
The constraint you impose upon the white units depends entirely upon the aim provided (# or =).
This can NOT be a condition -- the condition is only a set of rules, with no information as to the stipulation (or its aim)!

Therefore, reflex must be a stipulation!

Once I realized this, it hit me that reflex has to be described by a complex stipulation.

Consider a semi-r#2: the goal must be for white to achieve a position with a "#1 (for black)" or a "h#1 (for black)... and this must be solved by iteration (by default), down to the ultimate aim (# of white King).
You can describe a semi-r#2 fairly easily, without any fairy condition.
But, a full r#2 is more difficult.

In a r#2 the goal is the same, except white has an additional goal.
That additional goal must be to not allow black a "h#1" (or no "#1 for white").

I don't think you can properly describe GOALs as "move attribute AIMs" (that is, something white achieves upon moving -- like x or +); but, if you view these GOALs as "state AIMs" (that is, white achieves something after having moved -- like # or =), the correct description must be:

semi-r#2 = direct-[#1 for black]-2
r#2 = direct-[ [#1 for black] AND NOT (h#1 for black) ]-2

EDIT: actually, this is more complex than I thought -- it requires an expressed precedent! White can actually satisfy the goal of a r#2 by providing black both a #1 and a h#1 simultaneously, but the #1 takes precedent! This makes it an exceedingly difficult goal to express formally. There is also the matter of the initial state (given by the diagram position) -- what if white begins with a #1. Most, I think, would translate this as unsolvable (as r#2), but the above definition would suggest otherwise.

Perhaps there may be a better description (according to the definition) and/or a better way to express this (e.g., default to ignore NOT goals on iteration)... but, there is no doubt, neither can be considered a fairy condition!
If it must be a fairy condition, it would require dozens of fairy conditions, where the constraint on movement is explicitly instantiated, independently of the aim (in such a way that it always matches the aim).

This has to be expressed using a compounded GOAL, built upon combinational logic -- like we commonly see in "h-[ #/= ]-n" problems (except here, instead of a combination of two aims, we have a combination of two goals).

>...which is its correct meaning: the one provided by the Problemesis definition or the one provided by solving programs?

Neither Problemesis, nor the solving problems should be taken as correct.
The correct definition was likely provided by Benjamin G. Laws, in British Chess Magazine, 1893.
See: http://www.janko.at/Retros/Glossary/Reflex.htm

However, it is always possible (as is the case with Madrasi) that the commonly applied definition is in clear violation with the original intent.
The real question is, what is the commonly applied definition?
I think the solving programs, and virtually every reflex-problem today, is in agreement with the complex stipulation I provided above.

ps:

note that Retro Corner states the following (http://www.janko.at/Retros/Glossary/Reflex.htm):

"The first reflex problem was published by B. G. Laws in a 1893 British Chess Mag. issue. ... There exists a direct reflex stipulation (as opposed to inverse), recently invented by Jean Zeller. Here the goal is checkmate, but whenever it is possible (for any side) to play a so-called helping move that grants the opponent with a mate in one oportunity, one is under the obligation of playing it "by reflex" (unless one can already mate in one)."

Am I reading this correctly, if I translate to:

Laws invented what is now considered the "semi-reflex" stipulation, and Zeller invented what is now considered the full "reflex" stipulation?

As far as I correctly understood, Zeller's reflex is to oblige a side to let the other side using its reflex:

(= 1+2 )

a) x2
b) x2 Zeller reflex

In a) we have the obvious key 1.Bg2 while in b) there is no solution as white must play 1.Bc4 or 1.Bb5 to let black using its reflex. So black has no other option than capturing the Bishop and thus white will never be able to reach the goal by itself...