Fairy Classification: preliminary questions
If the project of creating some better database of fairy elements might succeed? If we’re ready to jump into it? Who would be ready to join the team? What is the order and what’s the freedom? And what’s is the main purpose of this database?
Let me quote the words of Chris Tylor’s comment which has just appeared: “I now propose that we go back to basics and make the overall aim of our new database not so much to produce order as to give light – to show users new properties and applications of a range of fairy elements, together with the way in which they relate to other elements (or perhaps stand alone), and maybe in the process point users to ways in which new advances might be made.” – Agree!
Let’s summarize the most important ideas from about 80 comments to Fairy Statistics post:
1. Statistics about fairy elements
We have about 2700 of fairy pieces and conditions ( in Excel file for download | or in PDF (51 page) for watching). And the list of 21 most used fairy pieces/conditions is updated with inventors’ names thanks to your help in the comment (specially by Kjell Widlert and Marjan Kovačević).
| No. |
Name
(WinChloe, French) |
Name
(Popeye, English) |
Number of problems |
Inventor |
year of invention |
| 1 |
Sauterelle |
Grasshopper |
14073 |
T.R.Dawson |
1913 |
| 2 |
Circé |
Circe |
11896 |
Pierre Monréal |
1967 |
| 3 |
Maximum |
Maximummer |
9950 |
T.R.Dawson |
1913 |
| 4 |
Noctambule |
Nightrider |
5053 |
T.R.Dawson |
1925 |
| 5 |
Neutral piece |
Neutral piece |
4963 |
T.R.Dawson |
1912 |
| 6 |
Anticircé |
AntiCirce |
3849 |
Fernand Calvet |
1970 |
| 7 |
Köko |
Köko |
3057 |
Heinz Zander |
1988? |
| 8 |
Madrasi |
Madrasi |
2829 |
Abdul Jabbar Karwatkar |
1979 |
| 9 |
Circé échange |
PWC |
2631 |
Umberto Castellari |
1975 |
| 10 |
Pièce Royale |
Royal piece |
1983 |
T.R.Dawson |
? |
| 11 |
Pao |
PAO |
1908 |
the ancient Chinese piece Cannon |
? |
| 12 |
Lion |
Lion |
1790 |
J. de A. Almay |
1940 |
| 13 |
Maximum blanc |
White Maximummer |
1772 |
|
1926? |
| 14 |
Take & Make |
Take & Make |
1579 |
Hartmut Laue |
2006 |
| 15 |
Alphabétiques |
Alphabetics |
1562 |
Roméo Bédoni |
1985? |
| 16 |
Vao |
VAO |
1557 |
P. Seyfert-Bilterfeld |
1936 |
| 17 |
Locuste |
Locust |
1396 |
|
1926? |
| 18 |
Chameau |
Camel |
1149 |
from the ancient Muslim Chess (the earliest example in WinChloe is by TRD 1913) |
1913? |
| 19 |
Cylindre vertical |
Vertical Cylinder |
1114 |
Teodoro Ciccolini |
1836 |
| 20 |
Léo |
LEO |
1080 |
P. Seyfert-Bilterfeld |
1936 |
| 21 |
Andernach |
Andernach |
1026 |
Hans Peter Rehm &
bernd ellinghoven |
1993 |
Statistics discovered by Shankar Ram from the Auxiliary Tables (as of the latest update) of WinChloe:
- Fairy Pieces: 1574
- Conditions: 1354 (2928 cumulative). This includes special boards.
- Stipulations: 44 (2972)
- Aims: 153 (3125)
Some more statistics about fairy pieces | genres from PDB.
2. INTRO to the classification: questions to the definitions of the classes, by Dmitri Turevski
- What is a fairy? (E.g. is SingleBox, formerly an official chess rule, a fairy? hs#? like helpself, but where sides cooperate to build not s#1, but s#2?)
- What is a stipulation? (Are reflex, series etc really stipulations or conditions? E.g. series-help = help + “White must pass unless can reach the aim”)
- What is a condition and what is a piece (Which is the Imitator?)
3. Preliminary Classification, suggestions by Chris Tylor and Shankar Ram
The Groups:
- Fairy pieces. Some sub-groups would be straightforward, e.g. leapers; others much messier, e.g.hoppers; while some pieces would stand alone or be hard to group, e.g. imitator.
- Piece modifiers OR Piece attributes; i.e. ways of modifiying the properties of a whole range of pieces, e.g. royal pieces, neutrals (a small group). Can be a subgroup of (1) Fairy pieces.
- Fairy variants. Some variants would form fairly clear sub-groups, e.g. Circe variations; others might go in pairs, e.g forms and anti-forms; most would probably resist classification.One obvious sub-group of the conditions main group would be “move restrictors”: maximummer, minimummer, single combat, black must capture, ohneschlag(no captures), black must check, checkless chess…
- Boards, e.g vertical cylinder, boards with holes (a small group).
- Stipulations (a small group). Subgroups: Aims and Ways.
The attributes of each element of the groups:
- Name or names.
- A description or summary of the properties or rules, perhaps defined in terms of the properties or rules of something else, e.g. a rook-lion as a lion moving on rook lines only.
- History, i.e. inventor and date (if known).
- Approximate number of known examples (which would need updating at intervals). The WinChloe database has been mentioned as a source, but there should be others, e.g. the PDB.
- The solving programs (if any) that would support the piece/variant, together with the symbol or name that each program uses for that piece/variant.
- Cross-reference links to other related elements.
4. Sources of information for classification and definitions
If we’re getting into this project, from the moment when the new database appears and has some entries I can start presenting it on JF. So, it might get in use before the whole project might be considered as finished or being up to date 🙂
For now I’ve created a GoogleSheet document Fairy-Classification for online work. The statistics and preliminary classification groups already included for future correction. I’d like to ask Shankar Ram to accept my invitation to be the main moderator of this project. The right to view/edit the document will be granted to all members of the team.
Welcome to join the project!
Your comments and updates are highly appreciated!
Julia,
“A journey of a thousand miles begins with a single step“. And I congratulate you on taking it!
Suggest we call it “The Fairy Chess Taxonomy Project“. (See also)
Of course, I accept your invitation with thanks!
Some additions to the sources:
http://problem64.beda.cz/silo/dawson_caissaswildroses_1935.pdf – TRD’s 1st book in his Caissa series, introducing various elements present at the time
http://problem64.beda.cz/silo/dickins_a_guide_to_fairy_chess_1971.pdf – A Pdf of A.S.M.Dickins’ pioneering work
https://www.dieschwalbe.de/lexikon.htm – Hans Gruber’s great labour of love! Updated till 26-Jul-2020.
http://problem64.beda.cz/silo/dickins_short_history_of_fairy_chess_1975.pdf – Some history
I will attempt to answer Dmitri’s questions:
What is a fairy?
A fairy is any set of aims, ways, pieces, conditions and boards that differ from the current set used in the game.
What is a stipulation?
A stipulation is a combination of an aim and the way of achieving it.
What is a condition and what is a piece?
A condition is the set of rules, restrictions and limitations applicable during the play in a game, end game study or problem. The current set of rules, regulations and limitations used in the current definition of the game is assumed to apply and only those that differ are mentioned in the definition of a condition.
A piece is an entity that occupies a square in a chess board and is able to move to other squares, capture other pieces and also change its nature(e.g: pawn promotion). The way it does all these is clearly defined.
Borderline and other cases:
1a) So selfmates are genuinely fairy by this definition. This is fine, but it radically differs from the current classification used in WCCI, WCCT & FIDE Albums.
1b) What is current rules? FIDE rules of chess? There may be a problem with that: these rules are constantly changing (and problemists have no government over that), does it mean fairy classification is expected to be amended retrospectively?
2) This is so vague that its practical usefulness is really questionable. This vagueness had prevented you from having a definite answer regarding the series/reflex. I would really recommend digging up the old MP threads with the mathematically beautiful approach to stipulations by Kevin and Petko. The core idea is that there are only two ‘ways’ – direct and help (that is against best/worst defense), but any valid substipulation can be an ‘aim’ in its own right, all the rest are conditions. Helpselfmates are obvious, less obvious may be that #2 is in fact direct-(direct-#-in1)-in1 – that is exactly how the solving programs handle it internally leaving no place for ambiguity!
3) A dummy is not a fairy piece by this definition which is somewhat counter-intuitive. On the other hand, if only occupying a square and having a definite behavior are required then holes and magic squares would become pieces.
Let’s see our project as the one describing the current situation! We can try to define what belongs to Fairy group as of today, mentioning some exceptions. Historically we came to some situation and some rules. And in the Introductory words we can give some words about the history, and also about some drawbacks in the current situation or possible proposals for the future.
WFCC rules are not something created once and forever. Perhaps they may change in the future, but before that it is good to understand what we have first. The list of stipulations/aims is not that long. Exceptions and “borderline cases” might be commented and described. And it is good to see them mentioned in different comments, those specific cases we have to pay attention to and to define better. The same time most of other pieces and conditions are easier cases to deal with.
Of course, our project should be something constantly updated. Life goes on. Changes happen everywhere. And we have to upgrade ourselves and everything around 😉
I’ve checked some old discussions on MPF (2012-2013). I won’t like to come back to them, to have any battles. Instead of such debates I’d rather want to have some constructive work together.
I’m not so long time in chess composition and don’t know much about the history. The next generations might know even less. I liked the words by Chris about giving some light. Let’s give it to ourselves and to the next generations!
The wikipedia link above educates us that the taxonomy is also about the study of the underlying principles.
Of course the WFCC rules are not cast in stone, the question is where have they gone wrong? Shankar Ram suggests (if I understood his answers correctly) that selfmates should be classified as fairy (the aim to mate own king differs from that in a game). Vlaicu, on the contrary, suggested that helpselfmates should not be classified as fairies (and he’s not alone in this).
I’ve simply brought up the underlying principle (fairy = legality of moves differs from that in a game) that supports Vlaicu’s suggestion. It was not about battling at all.
Similar arguments can be made as to why the Reflex is a condition (not very different from the Maximummer) and dummy is a piece and neither is a borderline case.
Dmitri,
Will reply in more detail later.
For now:
I remembered a third way of defining a Reflexmate: W & B play in opposition to reach a position where they can h#1. This seems to be sufficient. No additional condition like “should #1 if possible”. The moment a h#1 position is reached, the contract is met. Of course, there’s no need to muddy the waters further by adding this definition here!
Selfmates and Helpmates are in separate sections in FA, WCCT, WCCI, World Cup, YCCC, Olympic as also in most columns due to historical/convenience/familiarity/popularity/volume/judging/etc. reasons. Not because they are inherently “non-fairy”. I think the term “Heterodox” used by Vlaicu in his post may refer to such cases which are not orthodox, but have become so familiar that it seems strange to call them fairies.
That’s very interesting, thanks, I believe I did not know this one!
However, that’s not exactly a “h#1”, it’s a “h#1 Half-duplex” – that is, other side is to fulfill the stipulation from that point. This could be an important difference.
How so? For example, in a r#2, there are 4 plies: W1/B1/W2/B2. After the key, after any B1, the position is a h#1 for W. Similarly, in a genuine reflex try, only 3 plies are required: W1/B1/W2 and the position after W1 is a h#1 for B.
That’s exactly my point.
h#1 stands for helpmate in one. “A helpmate is a type of chess problem in which both sides cooperate in order to achieve the goal of checkmating Black” (this is from Wikipedia, but similar definition can be found in virtually any other source).
Therefore, a type of problem where both sides cooperate to checkmate White is not a helpmate.
It may be called “a h#1 for W”, however, “h#1 half-duplex” sounds more conventional to me, as that’s how it is recognised by Popeye for example (Stipulation h#1 Option HalfDuplex).
It’s a nice and subtle trick though 🙂
“…where they can h#1“. “They” = W or B. So the mated side can be B or W. That’s my interpretation. But I agree it may not stand scrutiny!
It also seems that “W & B play in opposition to reach a position where they can h#1 [for W]” w/o additional conditions only works for semi-r#
In opposition play black defend not by not letting white to reach the aim, reaching the same aim themselves achieves them nothing. In a reflexmate black defends because white is forced to mate and checkmate ends the game. In the above definition nothing forces white to checkmate after 1. W1? B1!
This can be more easily understood if we change the aim from checkmate, to “reach a certain square”, for example.
Black cannot defend by reaching this square themselves first! White is still might be able to reach the aim after that.
Yes, r# is double-face. The semi-r# is a “normal” stipulation/goal and there is a (white) reflex in addition. Note that this additional condition can be defined for other stipulations, even for proof games, as shown in 1257.
Hence it looks like the basic stipulation is semi-reflex, and reflex should be handled as a condition which apply to (almost) different kinds of stipulation. Saying otherwise r# should not be classified as a stipulation, but as a mixed stipulation “semi-r#” and condition “white reflex”.
On the other side r# is not fairy to my mind, just because each solution under this genre is a valid ending of an OTB game (ending because the diagram position is generally not the initial position of the chess game).
The case where the goal is not checkmate/stalemate is interesting. It is always classified fairy, whatever is the stipulation, and I agree with that. Indeed a solution where the goal is not such a terminal move does not correspond to a valid ending of an OTB game. To my eyes this is the right method to classify the fairy/not fairy genres.
“On the other side r# is not fairy to my mind, just because each solution under this genre is a valid ending of an OTB game”
This would also be true for Maximummer.
Yes indeed, mainly because checks are orthodox and parrying a check too in Maximummer (unlike eg Monochromatic). But it doesn’t really matter for fairy/not fairy classification I think, because it is a condition and not a stipulation/aim. Each condition might be considered fairy even if no illegal orthodox moves are allowed (Maximummer is just a particular choice in the set of orthodox legal moves).
A funny thing is that, if one follow my own taste, reflex defined as a stipulation is not fairy, but reflex defined as the stipulation “semi-r” plus condition “white reflex” (my favorite approach), becomes fairy! Btw semi-r is a bad name to my mind because there is by no way something reflex in it.
Is there such a thing as a non-fairy condition? I would assume no, but I’m open for ideas. (What is a condition? Alteration of the rules?)
I believe “reflex” in semi-reflex stands for black acting upon their reflexes: see prey – snatch prey, see checkmate – go for it. That sort of analogy. Can be semi if White refrains 🙂
“Is there such a thing as a non-fairy condition?”
Is there such a thing as a non-fairy piece, except the usual ones?
For example one can define (if not already done) a “maximummer piece” – an orthodox piece with restricted legal moves. Each sequence involving maximummer piece(s) would be a legal orthodox sequence, but probably such a piece should nevertheless be considered fairy.
white Ka1 Pa2
black Kc1 Sb3
how many last moves?
a) orthodox b) semi-reflex
Yes I see your point ha ha! -1 Sd4-b3+ is orthodox legal but semi-reflex illegal as black should have played -1 Sd4-c2# You’re right there is a slight difference between “being able to checkmate” in help-play and “being forced to checkmate” in semi-reflex.
Yep, that was the point. But mustn’t the last move have been a capture?
But in a Reflex-[reach square xy] stipulation, W tries to make B occupy that square, and B does likewise?
Yes, W tries to make B occupy that square, and B is defending except at its last ply. But B has 2 ways to defend. First to go far away from the goaled square (this is the semi-r part) and second to give white the opportunity to reach that square firstly (this is the white reflex part). In such a case white is obliged to reach the goal and the game is over. So the defense is successful because the true goal (B occupy that square) is not fulfilled.
As an example consider :
B: Ka6 W: Ba5 Sd6
r(b8)2
If white is playing eg 1.Sc4 black is escaping 1… Kb5. The key is 1.Bd8! Then 1…Ka7 is forced and black may reach square b8 at its second ply. This is the semi-r face.
The second face (white reflex) makes 1.Bc7? a wrong key. Indeed 1…Ka7 is still forced, but then white is forced to play 2.Bb8 (reflex move) and the game is over, so black is unable to reach the goal at its turn with 2…Kxb8.
The most important point is “and the game is over”. It is missed in the classical definition of an r# because redundant (black is unable to checkmate if it is checkmated by itself). But when the goal is not terminal, it is important to keep this rule in mind (and to mention it clearly once the reflex definition will be included in this workshop!).
“But when the goal is not terminal, it is important to keep this rule in mind”
Exactly! “Reach h#1 position” is not a terminal goal. So, if reflex is defined via h#1, this additional rule needs to be stipulated.
Chris Taylor has unearthed the following problem from the early days of JF:
https://juliasfairies.com/problems/page-17/#No.41. No. 42 by Zoran Janev.
DIRECT-HELPMATE (dh#)
White moves first, and black defendes himself, until the last move, when black and white play h#1.
This is closely related to the definition of a R# being discussed in this thread.
R#: W & B play in opposition to reach position where W can h#1
Dh#: W & B play in opposition to reach position where B can h#1
Of course, in a R# W is obliged to #1 if possible. But in the Dh# analogously, W should h#1 if possible, which is not mentioned in the author’s definition.
But appears to have a cook 1.Ba3 c2 2.Bb2#
And short length duals 1.Bb4 ~ 2.Bxc3#
If we interpret as above. Maybe the author’s intention was a Dh EXACT#3!
Zoran’s intention seems pretty clear from the author’s solution. But in the given definition, “until the last move”, might be puzzling.
dh(n)# means that Black would help ONLY by his last move before/preceding the n-th white move.
White is allowed to mate before n-th move but Black opposes by his first n-2 moves.
So 1.Lb4 indeed threatens 2.Lxc3#. Unfortunately, a short threat disables hypothetical possibilities of defence that would be specific for such stipulation – defending against h#1.5.
The try illustrates this:
1.La3? threatens h#1.5 – 2.Lc1(tempo) c2 3.Lb2# (obviously not the usual h# numbering of the moves) but 1.Lb~ refutes (by check). A defence against that threat is 1…e4 2.Lb4! Lf8 3.Lxc3#
There’s an antidual in the solution, due to a pin: 1.Lb4! e4 2.Ka3(tempo)? Lf8(?) 3.Lxc3???
I agree, the intention clearly was something like Direct-(h#1)-in-exact-2.
“until the last move” should have been stipulated “until the (n-1)th move”.
Amen to that, Julia!
Tamaso ma jyotir gamaya
Shankar, your comments already bring some light! And smile!
Some moves that are perfectly legal in an OTB game are illegal in r#.
Some moves (pass) that are absolutely illegal OTB are legal in seriesmovers.
Nothing of this kind happens in s# (the set of legal moves is always the same as in an OTB game).
There may be a pattern here 🙂
Coincidently(?), the WFCC puts the ser- and r# into fairies, but the s# separately.
One important thing should be considered: Interaction between stipulations, aims and conditions. Some combinations (e.g. Maximummer and reflex-mate: The reflex stipulation is ‘stronger’ than the maximummer condition; or circe+take&make: Which should be done first? The make-move or the rebirth? There are cases where the order is important; Anticirce + locust: For file-based rebirths (pawns, fairy pieces): Is the rebirth on the file of the capture or on the file where the locust arrives?) require a description of how the two elements interact.
Most combinations do not involve interactions, luckily! In the few that do, any ambiguity that exists, like the ones you mentioned, needs to cleared up. These have to be considered as special cases in our classification. We have to avoid further increase of fairy forms like “WinChloe version”, “Popeye version”, “type A”, “type B”, etc.
Unfortunately, when such interaction exists we generally have a branching combinations. Sometimes it is just the order of conditions matter, sometimes it is the issue of interpretation. My favourite example is neutral pieces+Madrasi. The current common interpretation is based on the article by Petko Petkov, but it is not the only possible one. At the same time, it is impossible to propose another interpretation now, as the current one has become kind of “classic”.
Some issues even does not require interaction, for example: is zero-move allowed when it involves real movement (e.g. Rose or cylinders), etc.
This sounds like a very interesting and important project.How can I join in?
Welcome, Dirk! Help is very much needed and appreciated!
My idea was to discuss here some unclear or important questions, to share the steps made, and to involve new collaborators. But to do the work online using Google Sheets.
Everybody can view the current worksheet by this the given link. I’d like to share it with you (and with everybody else willing to join) and to give you EDIT permission, but SHARE/EDIT works for only those who has Google account. Could you, please, create your Google account and send it to me (email)? Then I can add you to the team! Thanks a lot!
I would like to join the team working on the fairy classification
Maryan,
You need to create a Gmail ID and send it to Julia
I have done that.and will communicate about classification on the Google group site only
Here I have used Julia’s site because the question concerning the Imitator appeared in a discussion on that site
Is the piece/condition status of the Imitator a big issue? The subject should not be an impeachment for further progress. In my opinion, it is of course a piece. But it also interferes with the march of the other pieces, so it is also a condition.I think it makes sense to find the word Imitator featured in both groups, Pieces and Conditions (sub-group Movement ). By comparison , the Orphan does not interfere with other pieces,and is sufficiently described in the Pieces group
Well, Maryan. It’s something like the dual nature of light! As composers/solvers/connoisseurs we can just use the Imitator and enjoy the effects. As taxonomists, we can just document both the piece/condition properties and move on.
Hi, I am completely new to most of the concepts on your site, but it has set my mind racing on things that might be interesting. I wonder what this hyper-chess would be like if it were on the Starship Enterprise where chess is played in 3D? I am not sure Mr Spock would like the term “fairies”! But I like the idea of chess evolving in multiple ways. Already, the form of chess that most people play is becoming something that existing computers find too easy, so it is good to prepare for the next generations of computers (like quantum computers) by extending the number of possible moves and perhaps the environment of a piece could have an effect on the rules for how it can move. I will need to see plenty of examples of how the fairy pieces move before I get the hang of how to play.