General Puzzle Rules & Definitions

[PuzzleScot]

I think we should have general WPC/WSC style puzzle rules/definitions/guidelines that apply to all, or whole genres of puzzles.

I'll kick off with a few...

- All puzzles must be logically solvable from the information given.
- All puzzles must should have a unique solution. The main exception being optimisation puzzles.

In many cases, these facts can help deduce the solving route to the solution.

Definitions:
Cell Any region within a puzzle that is bounded by gridlines. Although usually square, this can be triangular, hexagonal, or any irregular shape!
Blacken To mark a cell completely black. In competition, it is sufficient to clearly indicate which cells are blackened. eg, by marking with an 'X'.
Orthogonally connected Cells connected by a mutual cell edge
optimisation puzzles There is not a unique answer. The aim is to configure an arrangement with the best score
neighbouring cells = orthogonally connected
surrounding cells cells touching the concerned cell, even by a point (orthogonally or diagonally).

Guidelines:
Loop puzzles (eg, Masyu & Every 2nd breakpoint) - There must be a single loop formed (unless otherwise stated), made by joining orthogonally connected cells by a straight line between the cell centre points.

Snake puzzles - Snakes are made from a series of orthogonally connected cells, with a start and end point. The cells forming the snake may not touch any other part of the snake, even diagonally.

Fence puzzles - Some puzzles are formed by drawing a single loop using the puzzle gridlines.

[Mephisto]

"All puzzles must be logically solvable from the information given" --

What 'logically' is depends on the abilities of solver. For example, some of the more complex Sudoku solving methods - e.g. color chains - can be seen as trial & error by an average solver. Any "what if" analysis is some kind of trial & error: A "what if" analysis of depth 1 or 2 is usually seen as "logical" and a "what if" analysis of depth 9 or 10 is usually seen as trial & error. It's a grading problem where "logically" ends and "trial & error" begins.

Moreover, this definition excludes all optimization puzzles. Is this intended?

"All puzzles must have a unique solution. If not, it is considered a poor puzzle if not otherwise stated." --

If a puzzle has two solutions and the author is not willing to fix this he can simple state "two solutions" and all is fine? IMHO not. Let's look at chess problems: Usually, a chess problem has a singe solution. If it has several solutions, then these solutions must be somehow "thematic". [BTW: In chess terminology a poor puzzle is "cooked".]

[detuned]

This is a good idea - get all the technical jargon down in one place!

[Mephisto]

Room: A room is a set of (usually) orthogonally connected cells, usually surrounded by bold lines. In some puzzles the diagramm is divided into rooms (e.g. Heyawake, LITS, Country Road), in other puzzles the division into rooms is the solution of the puzzle (Fillomino, Galaxies, Burokku).

Area: An area is a set of (usually) orthogonally connected cells, not surrounded by bold lines, but defined otherwise. For example, all blackened cells in Nurikabe must form a single area of orthogonally connected cells and the embedded areas of white cells must not be orthogonally adjacent. The black cells must not form an area which includes a square of size 2x2.

Room/Area of orthogonally connected cells: Transitive closure of cells w.r.t. the orthogonally adjacent relation Is there a better definition in plain English?

[PuzzleScot]

It strikes me that this would all be much better in a wiki environment!

[ronaldx]

It strikes me that this would all be much better in a wiki environment!

Don't really agree at all - it's better to have a discussion like this with an active editor (admittedly it requires someone to take charge) rather than anyone being able to change and overwrite good work.

[PuzzleScot]

Actually, wikis have a page discussion area for debatable points, to try and establish a commonly accepted phrasing. Sensible contributors refer to those pages befrore editing the main article.

The main positive is that all the agreed points are in one area without all the discussion interfering.
eg, in a puzzle wiki, there would be puzzle rules, an example, solving tips/guides, maybe some history, and links to similar puzzles or examples.
While all that information *is* in these posts, it is disjointed, and would be clearer to the novice if presented as a concise document, which is what a wiki offers.

[nickdeller]

I think it's best not to try to encapsulate "connected by an edge" and "touching at a point" into single words - it's clearer just to use the extra couple of words and take those two phrases (or some very similar construction) exactly as they stand. Yesno?

[OzerTamer]

"All puzzles must be logically solvable from the information given" --

What 'logically' is depends on the abilities of solver. For example, some of the more complex Sudoku solving methods - e.g. color chains - can be seen as trial & error by an average solver. Any "what if" analysis is some kind of trial & error: A "what if" analysis of depth 1 or 2 is usually seen as "logical" and a "what if" analysis of depth 9 or 10 is usually seen as trial & error. It's a grading problem where "logically" ends and "trial & error" begins.

Moreover, this definition excludes all optimization puzzles. Is this intended?

"All puzzles must have a unique solution. If not, it is considered a poor puzzle if not otherwise stated." --

If a puzzle has two solutions and the author is not willing to fix this he can simple state "two solutions" and all is fine? IMHO not. Let's look at chess problems: Usually, a chess problem has a singe solution. If it has several solutions, then these solutions must be somehow "thematic". [BTW: In chess terminology a poor puzzle is "cooked".]

I COMPLETLY AGREE WITH YOU. TRIAL AND ERROR SHOULD NOT BE IN A GOOD SUDOKU. A SUDOKU SHOULD HAVE A SOLUTION PATH FROM BEGGINING TO END WITH A LOGICAL CONCLUSION.

[OzerTamer]

I think we should have general WPC/WSC style puzzle rules/definitions/guidelines that apply to all, or whole genres of puzzles.

I'll kick off with a few...

- All puzzles must be logically solvable from the information given.
- All puzzles must should have a unique solution. The main exception being optimisation puzzles.

I AGREE WITH ABOVE IDEAS.
TRY AND ERROR CAN NOT BE A WAY TO SOLVE A SUDOKU, MAY BE FOR SOME PUZZLES BUT NOT SUDOKU.

I WANT TO SET UP A COMMON LANGUAGE FOR SHOWING SOLUTION PATH FROM BEGINING TO END USING TEXT AND SYMBOLS LIKE CHESS NOTATION
I ALREADY PROPOSED ONE. YOUR COMMENTS WILL BE HIGLY APPRECIATED.

[Mephisto]

"MUST be logically solvable" and "SHOULD have a unique solution" is a contradiction.

[OzerTamer]

"MUST be logically solvable" and "SHOULD have a unique solution" is a contradiction.

How? I could not see any contradiction on two propositions.
I will be happy if you can explain.

[Mephisto]

If there are two solutions, there is no logical way to decide which part of the solution path the solver should take. The solver must guess an some point. For example, you deduce that R3C5 must be 3 or 5, but since 3 leads to the first solution and 5 to the second one, there is no logical way to decide between 3 and 5.

If a puzzle has two (or more) solution(s), 1. this must be intended by the author; 2. this must be part of the stipulation; 3. this must be "thematic".

Thematic: Consider a proof game in 8.5 moves in chess. The first solution starts with Nf3 and ends with Nc3 and the second solution starts with Nc3 and ends with Nf3. This would be a very nice task.

[OzerTamer]

If there are two solutions, there is no logical way to decide which part of the solution path the solver should take. The solver must guess an some point. For example, you deduce that R3C5 must be 3 or 5, but since 3 leads to the first solution and 5 to the second one, there is no logical way to decide between 3 and 5.

This two solutions can be considered as two different sudoku whose beginning states are same. But, before this I think definition of a Sudoku will stop it as IT IS NOT A SUDOKU but a puzzle looks like Sudoku. In your example, when you come to find value of R3C5 (I addressed it as 3e) if it can be 3 or 5 and each of them take the solution different way, we have to think again if we are solving a sudoku or not. If there is not any other logic to move forward, NO THIS IS NOT SUDOKU.

[Mephisto]

Your answer does not address my complaints. If you deduce R3C5 = 3 or 5, you don't know that both are right -- so you have to guess. (There is no way to deduce that both numbers are right! You most likely assume that your logical skills are insufficient at this point.)

[OzerTamer]

Your answer does not address my complaints. If you deduce R3C5 = 3 or 5, you don't know that both are right -- so you have to guess. (There is no way to deduce that both numbers are right! You most likely assume that your logical skills are insufficient at this point.)

I thought at that point, 3 or 5 can takes you different solutions. If so, I said this can be two different SUDOKU.
But now your compain is not that point. If my understanding is correct, you are saying that, there is no logical conclusion to move forward except try one of them and see if it is ok. My opinion at this point, (which can be discussed), is that A GOOD SUDOKU SHOULD NOT CREATE THIS STATE WHICH CAN BE SOLVED ONLY TRY AND ERROR METHOD. I can not call it Sudoku, but another puzzle. This was my idea to explain on my previous message.