READ ME FIRST: What is a Sudoku Committee

detuned
Posts: 1744
Joined: Mon 21 Jun, 2010 2:25 pm
Location: London, UK
Contact:

READ ME FIRST: What is a Sudoku Committee

Post by detuned » Fri 28 Dec, 2018 2:56 pm

This forum has kindly been hosted by the UKPA for the purpose of the WPF's "What is a Sudoku?" committee. The committee was first set up in 2017, but failed to report. In 2018 I took on the lead for the committee and my intention is very much not to end in another failure.

Whilst the question "What is a Sudoku?" probably sounds very obvious, the modern day reality of the sorts of puzzles you might solve in a Sudoku competition are for the most part Sudoku variants, most of which are probably unfamiliar away from a relatively small number of enthusiast solvers. Even amongst this more dedicated crowd there is some controversy over some harder or more complicated Sudoku variants.

One point of view says that anything is fair game to the World Sudoku Championship's (WSC) older brother - the World Puzzle Championship (WPC) - however it is a pity to have too much overlap between the two competitions and as such a clear line needs to be drawn with regards to Sudoku variants. Another point of view is that anything that vaguely looks like Sudoku is fair game for the WSC.

The committee's first task is to stimulate discussion and gather as many points of view as possible to see if an initial consensus can be drawn.

Everyone is welcome to post in this forum, but please follow a few guidelines so that we don't end up with total chaos.

1) For each new example to be discussed, please post a new thread.
2) Where possible, please give a source for the puzzle you wish to discuss.
3) For each post, please provide the puzzle itself, together with a short discussion about why you think your example is or isn't a sudoku

DavidC
Posts: 190
Joined: Wed 03 Nov, 2010 9:34 pm

Re: READ ME FIRST: What is a Sudoku Committee

Post by DavidC » Sat 29 Dec, 2018 8:47 am

Hi Tom. How would you like us to approach a situation like this? Without having the actual discussion here, my opinion of HALLOWEEN has two distinct aspects: the use of non-numeric symbols and the use of repeated symbols. Would you like me to simply vote on the specific example you have chosen, or simply add a comment without skewing the voting in an ambiguous way, or vote on the specific and comment on the generic? Cheers, David.

detuned
Posts: 1744
Joined: Mon 21 Jun, 2010 2:25 pm
Location: London, UK
Contact:

Re: READ ME FIRST: What is a Sudoku Committee

Post by detuned » Sat 29 Dec, 2018 9:24 am

A first tip would be to keep comments relevant to their example thread.

The polls are something I noticed I could do with the forum. Like any good referenda, these are purely advisable and are there only to see if there are any particularly strong feelings one way or the other - they are absolutely non binding. Most of the examples are in the grey area so I'm not necessarily expecting to see a clear picture one way or the other. Please vote how you see fit (hopefully in a way that you can vaguely rationalise back to the example) and if you feel that a mere poll doesn't adequately allow you to express yourself, then please do make use of the useful feature of this forum that also allows you to comment!

Mephisto
Posts: 75
Joined: Tue 18 Jan, 2011 12:28 pm

Re: READ ME FIRST: What is a Sudoku Committee

Post by Mephisto » Sun 30 Dec, 2018 10:07 am

First thoughts and basic questions:

A classic Sudoku is not really a number puzzle, because it does not make use of the properties of numbers (Penao axioms). Digits (not numbers!) are simply used as signs. Instead of digits you can also use letters, colors or any symbols (star, circle, square, ...).

On the other hand, there are Sudokus that do actually need numbers, Greater-Than Sudoku or Sodukus with mathematical operations. Are such variants still real Sudokus or not?

A classical Sudoku is based on a latin square. Are Sudoku variants that violate the latin rule still Sudokus or not?

I suggest that we should discuss basic principles in a first step and concrete puzzles in a second step.

detuned
Posts: 1744
Joined: Mon 21 Jun, 2010 2:25 pm
Location: London, UK
Contact:

Re: READ ME FIRST: What is a Sudoku Committee

Post by detuned » Sun 30 Dec, 2018 1:03 pm

Hi Mephisto, thanks for your thoughtful post, there is plenty to unpack here.

Firstly:
I suggest that we should discuss basic principles in a first step and concrete puzzles in a second step.
Does the use of "we" imply that you would like to contribute more meaningfully to the committee's work? My own view on this is that the world of Sudoku variants is so weird and wacky that I think there is real value in exploring a collection of examples, and I'll continue to do so over the next few weeks. My previous academic training has taught me to be entirely sure about the lay of the land before starting to put down formal rules and definitions, and I view view this exploration as necessary groundwork.
A classic Sudoku is not really a number puzzle, because it does not make use of the properties of numbers (Penao axioms). Digits (not numbers!) are simply used as signs. Instead of digits you can also use letters, colors or any symbols (star, circle, square, ...).
I have thought a lot about the label "classic" Sudoku. To me, this is making an implied claim that Sudoku consists of many variants, of which "Classic" is only one amongst many to be respected equally by this committee. This is a valid point of view of course, but it is not the only one. A competing point of view might say that classic Sudoku is the most important variant, and that we might be wary of variants that depart too far from the spirit of the original.
Digits (not numbers!)
I have several reasons to prefer the use of "numbers" rather than "digits"

- The use of digits breaks down for example with 12x12 grids
- There are Sudoku variants where numbers placed form longer strings under certain constraints, and it is clearer and more convenient to reserve the word "digit" exclusively for this context
- Placing numbers is a nice nod to the origins of Sudoku, which was first published as "Number Place"
- Generally, people understand numbers to specifically refer to non-negative integers, rather than their negative, rational, real or complex cousins.

What are the convincing reasons to use the word "digit" over "number"?

Anyhow, to return to the point, for the sake of readability and general application of the report I am supposed to be producing, I would prefer not to directly refer to the Peano axioms! However, I fully agree that the source of many Sudoku variants is the fact that numbers are special kinds of symbols which have many properties including:

- an ordering (which allows us to compare different numbers and say which is bigger or smaller)
- arithmetic properties (allowing us to calculate sums, products, differences and ratios)
- parity (we can say if a number is odd or even)
Are such variants still real Sudokus or not?
This is precisely the question the committee is tasked with answering! My feeling is this will be impossible to do both practically and definitively and some compromises will inevitably have to be made. We are only at the beginning of this work however!
Are Sudoku variants that violate the latin rule still Sudokus or not?
I would be very grateful if you could provide some examples of such variants so we can discuss further in more detail!

Mephisto
Posts: 75
Joined: Tue 18 Jan, 2011 12:28 pm

Re: READ ME FIRST: What is a Sudoku Committee

Post by Mephisto » Sun 30 Dec, 2018 2:39 pm

however it is a pity to have too much overlap between the two competitions and as such a clear line needs to be drawn with regards to Sudoku variants.
A clear line does not exist. This is one of the paradoxes that philosophers have discussed for a long time. The basic thesis is always the proposition of the excluded third.

A caterpillar becomes a butterfly, but there is no moment T at which this animal is a caterpillar and a second later it is a a butterfly.

A poor man becomes a rich man, but when exactly is he rich? One Euro more and the poor man becomes rich?

A Puzzle XYZ is a Sudoku, everyone has agreed that in this forum. Now a rule is changed minimally, and suddenly it is no longer a Sudoku? But the rule change is much more minor than the difference between XYZ and the classic 9x9 Sudoku.

A provocative thesis: If we always change the classic Sudoku rules only minimally, after 1000 steps we come to the conclusion that Masyu is a Sudoku, because we can't overcome ourselves at any step to say "no".

[This posting may not be very helpful but I couldn't resist :-) ]

Mephisto
Posts: 75
Joined: Tue 18 Jan, 2011 12:28 pm

Re: READ ME FIRST: What is a Sudoku Committee

Post by Mephisto » Sun 30 Dec, 2018 2:55 pm

Does the use of "we" imply that you would like to contribute more meaningfully to the committee's work?
I do not know enough about this committee. Primarily, "we" are the members of this forum. But I'm open to be convinced to join.
What are the convincing reasons to use the word "digit" over "number"?
I do not suggest to replace "number" by "digit" if we write down rules for solvers; "numbers" is perfectly right. For an abstract discussion "what is a Sudoku" it may be worth keep in mind the difference.
The use of digits breaks down for example with 12x12 grids
True. The correct wording would be "symbol" ["11" is a symbol, not a number, but now it becomes too philosophically.]
- an ordering (which allows us to compare different numbers and say which is bigger or smaller)
- arithmetic properties (allowing us to calculate sums, products, differences and ratios)
This is exactly the definition of numbers, only in more common words than Peano did it :-)
some examples 
See separate posting.

detuned
Posts: 1744
Joined: Mon 21 Jun, 2010 2:25 pm
Location: London, UK
Contact:

Re: READ ME FIRST: What is a Sudoku Committee

Post by detuned » Sun 30 Dec, 2018 3:26 pm

Mephisto wrote:
Sun 30 Dec, 2018 2:39 pm
A provocative thesis: If we always change the classic Sudoku rules only minimally, after 1000 steps we come to the conclusion that Masyu is a Sudoku, because we can't overcome ourselves at any step to say "no".
I'd like to explore this thesis a little more.

Athletics is a loose collection of different disciplines bound together by some principles, which whilst not being entirely definitive are at least practical. Namely, "Track and Field", and "Running, Throwing, Jumping".

I put it that it is easy to argue that in one sense, the 100m sprint and the 50m freestyle event in swimming have a much smaller difference between them than the difference between the 100m sprint and the shot put. And yet I would argue that very few people seem to have any problem drawing a very clear line in saying that 50m free style swimming is not athletics. So why should it always be the case that we should be unable to say no at each incremental adjustment of the rules?

Mephisto
Posts: 75
Joined: Tue 18 Jan, 2011 12:28 pm

Re: READ ME FIRST: What is a Sudoku Committee

Post by Mephisto » Sun 30 Dec, 2018 4:20 pm

In your example, you make big steps and it is easy to draw a line between.
So why should it always be the case that we should be unable to say no at each incremental adjustment of the rules?
Short answer: Because the number of possible Sudoku variants is infinite, maybe nondenumerable. [think of a puzzle as a function on a set of symbols and thre are more functions than real numbers.]

Consider a Classic Sudoku A and a Sudoku variant B, which we have agreed to be a Sudoku. Now take a Sudoku variant C, which rules are minimal different from B, at least much less different than the rules of A and B. Would you say that C is not a Sudoku? Most likely not. Now consider a Sudoku variant D, which is minimal different from C ...

A concrete example: A is a classic Sudoku, B is a killer Sudoku and C is Kenken/Kendoku. If you agree that Killer is a Sudoku, you must also agree that Kenken is a Sudoku because the difference between Killer and Kenken is much less than the difference between Classic and Killer.

detuned
Posts: 1744
Joined: Mon 21 Jun, 2010 2:25 pm
Location: London, UK
Contact:

Re: READ ME FIRST: What is a Sudoku Committee

Post by detuned » Sun 30 Dec, 2018 5:19 pm

And what exactly is a big step? Consider that:

- 50m freestyle and 100m sprint are both races in which fields of competitors look to get to one end of the course as fast as possible in order to finish in 1st place.

- 50m freestyle is swimming in water, and 100m sprint is running on a track.

In one sense there is a small difference, in another a rather large difference. It strikes me that this provocative thesis has a certain degree of one dimensionality to it - there is an underlying premise that there is some linear scale with two ends labelled "very much a Sudoku" and "not at all a Sudoku". I don't think this is necessarily the only point of view which can be taken.

Taking Sudoku, Killer Sudoku and KenKen (TM) (R) (C) (TMTMR) (C) (T)(M): if you, for example, impose the principle that a Sudoku variant should have some kind of 3rd regional constraint then it is very easy to say that kenken is not a Sudoku variant. You say that this is a small increment in the rules from Killer, but other people might say otherwise according to this principle.

Mephisto
Posts: 75
Joined: Tue 18 Jan, 2011 12:28 pm

Re: READ ME FIRST: What is a Sudoku Committee

Post by Mephisto » Sun 30 Dec, 2018 5:52 pm

impose the principle that a Sudoku variant should have some kind of 3rd regional constraint then it is very easy to say that kenken is not a Sudoku variant.
Ok, then add the uniqueness of the numbers on the two diagonals to KenKen. Now you have a 3rd regional constraint. Is it a Sudoku yet?

You have no problem to accept additional regions (diagonals, windows, etc.). Why do you not accept less regions? Is a latin square with 4 "windows" a Sudoku?

Btw: Is a greek-latin suqare a sudoku? (I'll start a new thread on this.)

Mephisto
Posts: 75
Joined: Tue 18 Jan, 2011 12:28 pm

Re: READ ME FIRST: What is a Sudoku Committee

Post by Mephisto » Sun 30 Dec, 2018 6:15 pm

Thinking again: Now we come to the kernel of the poodle:

If we agree on the rule "A Sudoku must have at least 3 regional constraints" then we can simply exclude many variants without further thinking. For example, we can exclude KenKen (but not KenKen with diaginals).

A second simple rule may be: "All symboles in a region must be unique". This excludes, for exampe, Subset Sudoku and Star Battle Sudoku. [On the other hand, Star Battle Sudoku currently has 1 yes and 2 maybe votes.]

If we add the rule that each region must contain all symbols then we exclude all renban groups and many other variants.

If we find and agree on more such simple rules we are almost done.

detuned
Posts: 1744
Joined: Mon 21 Jun, 2010 2:25 pm
Location: London, UK
Contact:

Re: READ ME FIRST: What is a Sudoku Committee

Post by detuned » Sun 30 Dec, 2018 6:16 pm

I feel this productive discussion is beginning to bear fruit - thank you!

The example of adding a pair of diagonal constraints reminds me of some of some examples responsible for the bad feelings expressed following the 2009 WSC - perhaps this discussion is better placed in this thread:

viewtopic.php?f=33&t=1931

drsudoku
Posts: 1
Joined: Sun 30 Dec, 2018 11:07 pm

Re: READ ME FIRST: What is a Sudoku Committee

Post by drsudoku » Mon 31 Dec, 2018 12:49 am

A sudoku is a symbol placement puzzle where the objective is to place a symbol into each cell so that no symbol repeats in any row, column, or region.

Sub-notes:
  • As with all "proper" logic puzzles, there must be a unique arrangement of symbols from the given information (i.e., one answer).
  • A region is intended to be a multi-dimensional shape, with limited overlap with the existing sets formed by the linear row or column.
Counter to what Mephisto is saying, it is not just that a Sudoku has 3 regional constraints in some places. It is specifically that every single cell in the puzzle is subject to exactly three unique placement constraints tied to the non-repetition of symbols. The most critical addition by Howard Garns that turned Latin Squares into Number Place/Sudoku puzzles was introducing the 3x3 box constraint on top of the row and column constraints that were already there from Latin Squares [which are not Sudoku]. The primary logical mode of solving a Sudoku puzzle is to consider elements that are present (or not yet present) in the grid in order to identify other placements (or eliminate impossible placements) as a result. This is done by scanning rows, columns, and regions, and there are roughly 12-15 known heuristics that apply to solving a sudoku puzzle.

A Sudoku Variation (no longer a sudoku) has 3 unique placement constraints on every cell, as well as some number of additional constraints that may require new heuristics to be used when solving. These additional constraints can be similar to the normal "sudoku-type" constraint (as in Disjoint Sudoku, Extra Region Sudoku, or Diagonal Sudoku), or can involve other kinds of thinking including arithmetic, relational constraints, touching constraints, etc. The majority of the solution should require thinking about the 3 unique placement constraints and no other constraints.

A puzzle hybrid featuring sudoku rules is anything with a set of rules that includes 3 unique placement constraints on some, many, or even all cells, but with other rules that play a major part in deducing the solution.

A (non-sudoku) puzzle is anything that does not have 3 unique placement constraints on any or all cells.

I don't see myself how (1) "What is a Sudoku?" is a question I have anything more to say about than the basic thoughts above. An Arrow Sudoku is not a sudoku. A Diagonal sudoku is not a sudoku. It has some cells with more than 3 unique placement constraints. They are at best Sudoku Variations.

There is a line between Sudoku Variations and puzzle hybrids featuring sudoku that is very tricky, and the rules can seem like a Sudoku Variation while a given instance of the style may solve more like a hybrid (and vice versa).

I do see how (2) "What is a reasonable challenge for a World Sudoku Championship?" is an important question. Even with an answer to "What is a Sudoku?" there are still proper "sudoku" by definition that are not reasonable for a championship due to size or difficulty or the lack of a logical path for its solution.

From the polls it is very unclear if you are going after (1) or (2), but if it is (1) I don't have more to say than what is above. On (2), I wrote an open letter to the WPF board long ago and my basic thoughts there still stand.

Mephisto
Posts: 75
Joined: Tue 18 Jan, 2011 12:28 pm

Re: READ ME FIRST: What is a Sudoku Committee

Post by Mephisto » Mon 31 Dec, 2018 7:59 am

Counter to what Mephisto is saying, it is not just that a Sudoku has 3 regional constraints in some places.
I only quoted detuned and made my comments assuming this definition and questioned it.

However, assuming your Sudoku definition, there are only two kinds of Sudoku:; regular and irregular. Or do you insist on 3x3 boxes?

Secondly, if we accept this strict definition, detuned should modify his question to "What is a Sudoku variant suitable for the WSC" to draw a line between WSC and WPC (this seems to be the primary requirement).

Thirdly, assuming your Sudoku definition the WSC should be renamed to WSVC because almost none of the puzzles there are Sudokus. :-)

Assuming your Sudoku definition, please replace "Sudoku" by "Sudoku Variation" in all my previous comments.

detuned
Posts: 1744
Joined: Mon 21 Jun, 2010 2:25 pm
Location: London, UK
Contact:

Re: READ ME FIRST: What is a Sudoku Committee

Post by detuned » Mon 31 Dec, 2018 11:19 am

drsudoku wrote:
Mon 31 Dec, 2018 12:49 am
I don't see myself how (1) "What is a Sudoku?" is a question I have anything more to say about than the basic thoughts above. An Arrow Sudoku is not a sudoku. A Diagonal sudoku is not a sudoku. It has some cells with more than 3 unique placement constraints. They are at best Sudoku Variations.

There is a line between Sudoku Variations and puzzle hybrids featuring sudoku that is very tricky, and the rules can seem like a Sudoku Variation while a given instance of the style may solve more like a hybrid (and vice versa).

I do see how (2) "What is a reasonable challenge for a World Sudoku Championship?" is an important question. Even with an answer to "What is a Sudoku?" there are still proper "sudoku" by definition that are not reasonable for a championship due to size or difficulty or the lack of a logical path for its solution.

From the polls it is very unclear if you are going after (1) or (2), but if it is (1) I don't have more to say than what is above. On (2), I wrote an open letter to the WPF board long ago and my basic thoughts there still stand.
Thanks a lot for your thoughtful post. I am (still!) familiar with your open letter and have been considering it alongside other schools of thought when it comes to the question. You are right that at this first stage of committee, an exploration of examples, there are effectively the two questions you've identified at play. I believe that the two questions are very closely related however. On the one hand, it's a no brainer that we shouldn't have any puzzle that doesn't come under the category of sudoku (however we end up defining it) at a sudoku competition. On the other hand, in the absence of a canonical definition, the puzzles that have previously appeared at the WSC inevitably have some influence over what this definition should be.

Indeed, one school of thought I've heard says that other sudoku competitions take the WSC as a lead, and as a result include puzzles in their competitions that they might not have otherwise included were it not for the observation "these are the sorts of puzzles you need to be able to solve at a WSC".

As part of the committee's work, I am going through and analysing each and every puzzle that has appeared in the (individual) rounds of the WSC. As you can tell by the examples I have posted in the forum so far, I haven't got that far (I'm starting on 2013 soon, and have posted examples up to 2010), but I'd like to address Mephisto's comment:
Thirdly, assuming your Sudoku definition the WSC should be renamed to WSVC because almost none of the puzzles there are Sudokus.
From my analysis (so far, this is just an initial impression), the proportion of puzzles and points dedicated to classic sudoku at the WSC's seems to be in decline. In the earlier years, this seems to be c. 40-50% of the puzzles and c. 30% of the points - I doubt these numbers are going to be as high when it comes to the more recent years. One hypothesis for this is the emergence of one-off online contests which have placed a much higher emphasis on sudoku variants.

I don't want to jump ahead too fast and move on to the definition without first exploring some more examples, and how things work in practice as well as in theory. Nevertheless the definition proposed by drsudoku is definitely welcome and gives us a lot to think about. An initial question I have is:
- Are we assuming an underlying an underlying square grid with square shaped cells?

---

Looking further ahead, the proposed concepts of sudoku variations and sudoku hybrids seem to me (inevitably!) to be a little vague. It builds upon a school of thought first clearly expressed at the 2017 WSC and WPC, and made concrete in the WSC round "Is it a Sudoku?" and the WPC round "Logidoku", which each featured puzzles with exactly the same set of rules. When I get through my WSC analysis as far as 2017 I will definitely be revisiting this.
The World Sudoku Championships consist of Sudoku and variations of Sudoku. The Classic form of Sudoku is that each row, column and 3x3 outlined box must contain all digits from 1-9. A variant of Sudoku typically has similar constraints for 'row, column and outlined region' along with some additional concept/property/constraint.

The selection of puzzles for the WSC and general organization is at the discretion of the host country. The organizers of WSPC 2017 believe that a Sudoku variant can lean towards a Sudoku or a Puzzle based on the manipulation of the author, regardless of the nature of the rule-set as long as the basic characteristics above are followed. We will attempt to showcase this over different parts of the Championship.
Given some of the examples that have already been posted, as well as those yet to come, it might prove a useful exercise to explore these concepts further. Certainly some kind of classification of sudoku variants is something that I have had in mind as part of this report.

Fred76
Posts: 133
Joined: Mon 09 Aug, 2010 1:53 pm
Contact:

Re: READ ME FIRST: What is a Sudoku Committee

Post by Fred76 » Thu 03 Jan, 2019 9:57 am

drsudoku wrote:
Mon 31 Dec, 2018 12:49 am
I don't see myself how (1) "What is a Sudoku?" is a question I have anything more to say about than the basic thoughts above. An Arrow Sudoku is not a sudoku. A Diagonal sudoku is not a sudoku. It has some cells with more than 3 unique placement constraints. They are at best Sudoku Variations.
Thank you for sharing your definition of sudoku.
I have a few questions about it:

1. You defined 4 categories: sudoku, sudoku variations, puzzle hybrid featuring sudoku rules and non-sudoku puzzles. Puzzles of which category do you recommend to use for sudoku tournaments, especially WSC?

2. In the 1rst category, sudoku, you used the "non-repetition" rule (not the "appear exactly once" rule). It means you place puzzles like, for example, deficit sudoku, blackout sudoku and compressed sudoku (https://logic-masters.de/Raetselportal/ ... ?id=0002DB) in this category and not as sudoku variations. For what reason do you prefer the "non-repetition" rule rather than the "exactly once" rule?

3. sudoku variations vs. puzzle hybrid: Here I think this is the blurring part of your definition. If I understand correctly, there is no puzzle type which belongs definitely to the sudoku variations category, as you can create a puzzle where the extra-rule predominates the solving path and is no longer a sudoku variation, but a puzzle hybrid.
On the other hand, the puzzle hybrid category contains a lot of very different puzzles, I think this category could be divided in more categories. First the usual sudoku constraint can be "on some, many, or even all cells", so I think we could here make 3 distincts categories with this argument (it is possible that someone's opinion is that puzzle hybrids are fine only when all cells are constrained). Then in this category you have the majority of hard "variants" appearing usually in tournaments (which are not in your "sudoku variations" category): it's rare to see a hard variant in a tournament where the sudoku placement dominates the solving, in my opinion. I don't think a lot of people would say these puzzles are not a reasonable challenge for a World Sudoku Championship. But in the same category, you have hybrid between symbol placement puzzle (sudoku) and any kind of other categories of puzzle types (polyomino puzzles, loop puzzles, snake puzzles, etc...), which can potentially be considered as not reasonable challenge for a wsc.
Is there a good reason to place all these puzzles in a unique category (puzzle hybrid featuring sudoku rules), or do you think we can make some more distinctions here?

Fred

Fred76
Posts: 133
Joined: Mon 09 Aug, 2010 1:53 pm
Contact:

Re: READ ME FIRST: What is a Sudoku Committee

Post by Fred76 » Sun 06 Jan, 2019 12:13 pm

Mephisto wrote:
Sun 30 Dec, 2018 2:39 pm
A provocative thesis: If we always change the classic Sudoku rules only minimally, after 1000 steps we come to the conclusion that Masyu is a Sudoku, because we can't overcome ourselves at any step to say "no".
I think even if you divide the path between sudoku and masyu in 1000 steps, there will be some more crucial steps. For example there will be always a step between a puzzle without loop and a puzzle with a loop, and a step between a puzzle containing digits and a puzzle that doesn't contain digits. These kind of steps may be the limits we are searching for.

Fred

Fred76
Posts: 133
Joined: Mon 09 Aug, 2010 1:53 pm
Contact:

Re: READ ME FIRST: What is a Sudoku Committee

Post by Fred76 » Sun 06 Jan, 2019 12:25 pm

Mephisto wrote:
Sun 30 Dec, 2018 4:20 pm
In your example, you make big steps and it is easy to draw a line between.
So why should it always be the case that we should be unable to say no at each incremental adjustment of the rules?
Short answer: Because the number of possible Sudoku variants is infinite, maybe nondenumerable. [think of a puzzle as a function on a set of symbols and thre are more functions than real numbers.]

Consider a Classic Sudoku A and a Sudoku variant B, which we have agreed to be a Sudoku. Now take a Sudoku variant C, which rules are minimal different from B, at least much less different than the rules of A and B. Would you say that C is not a Sudoku? Most likely not. Now consider a Sudoku variant D, which is minimal different from C ...

A concrete example: A is a classic Sudoku, B is a killer Sudoku and C is Kenken/Kendoku. If you agree that Killer is a Sudoku, you must also agree that Kenken is a Sudoku because the difference between Killer and Kenken is much less than the difference between Classic and Killer.
A is Berlin, B is Stuttgart, C is Strasbourg. Do I have to conclude that Strasbourg is in Germany because the distance between Strasbourg and Stuttgart is much less than the distance between Stuttgart and Berlin, which is the capitale city of Germany?
If a border is defined, it means 2 puzzles can be very similar, but each from one side and the other of the defined border, so it's possible that only one of them is a sudoku.

Furthermore, I think Classic sudoku has not to be in the center of the defined field which is sudoku. If D is a latin square, we could say the difference between latin square and classic sudoku is much less than the difference between classic sudoku and killer sudoku. But it's not a problem to say latin square is not a sudoku. Let's say D is Poznan. The distance between D and the capitale city of germany is much less than the ditance between Stuttgart and the capitale city of Germany.

Fred

Mephisto
Posts: 75
Joined: Tue 18 Jan, 2011 12:28 pm

Re: READ ME FIRST: What is a Sudoku Committee

Post by Mephisto » Sun 06 Jan, 2019 6:20 pm

A is Berlin, B is Stuttgart, C is Strasbourg. Do I have to conclude that Strasbourg is in Germany because the distance between Strasbourg and Stuttgart is much less than the distance between Stuttgart and Berlin, which is the capitale city of Germany?
No, because there is a clear border between Germany and France. This border is artificial, not geographical/cultural (and changed often in the past). And what we are looking for – as far as I understand this discussion – is an artificial border between puzzles suitable for the WPC and the WSC (hence that I avoid the term "Sudoku Variant").
If D is a latin square, we could say the difference between latin square and classic sudoku is much less than the difference between classic sudoku and killer sudoku.
Agree.
But it's not a problem to say latin square is not a sudoku.
Why not? Obviously you have some artificial border in mind. But is it agreed? IMHO we need to agree some properties Sudoku variants MUST have and then check these propereties for concrete Puzzles. For example:

1. Cells and Regions

a) In a valid Sudoku variant each cell mst be part of exactly three regions (this excludes e.g. diagonal Sudoku, window Sudoku and latin squares)

b) In a valid Sudoku variant each cell must be part of at least three regions (this does not exclude diagonal Sudoku, but latin squares)

c) In a valid Sudoku variant all cells must be part of at leat two regions and at least some cells must be part of additional regions

d) ...

2. Region Size

a) Each region must contain all possible symbols (i.e. all regions must be of the same size)

b) At least two/threee region classes must contain all symbols, but there may be some smaller regions too (e.g. renban groups)

c) ...

3. Uniqueness of Symbols

...

If we can agree a set of such properties we can simply check a concrete puzzle against it and decide yes/no. And maybe we should discuss each of there properties in a separate thread.

Fred76
Posts: 133
Joined: Mon 09 Aug, 2010 1:53 pm
Contact:

Re: READ ME FIRST: What is a Sudoku Committee

Post by Fred76 » Mon 07 Jan, 2019 10:38 am

Mephisto wrote:
Sun 06 Jan, 2019 6:20 pm
But it's not a problem to say latin square is not a sudoku.
Why not? Obviously you have some artificial border in mind.
Yes, I have some border in mind (I was an sudoku author, so when organizers asked me some puzzles for sudoku competition and let me free about the type of variant I can create, it must have been a concrete meaning to me), but it was not my purpose to say that latin square is not a sudoku (although, yes, I think it's true). My purpose was to say that it's not a valid argument to say "because latin square is close to classic sudoku, latin square is a sudoku".
Mephisto wrote:
Sun 06 Jan, 2019 6:20 pm
IMHO we need to agree some properties Sudoku variants MUST have and then check these propereties for concrete Puzzles.
Yes I agree this is what we should do, and I think this is what Tom will do. If I understand, he wanted to start with discussions concerning examples to have a better idea when it comes to speaking about properties (not sure if it will succeed).
But yes, normally it should be the opposite way: to define what is a sudoku first, then with the definition we can hopefully have an answer to all examples. The problem is that the task which consist of defining what is a sudoku comes after 13 years of world sudoku championship (and 6 years of sudoku gp), and the puzzles of these competitions can sometimes be very far away from the initial thing we called sudoku. So the question is: do we have to write a definition which include all previous wsc and gp puzzles (good luck for the writing), or do we come back to the initial situation, as if we don't know all puzzles appearing in WSC/GP and write a more strict definition; the risk is then, if we look back, to see that some previous championships were far to be a sudoku competition according to the new definition.

Fred

Mephisto
Posts: 75
Joined: Tue 18 Jan, 2011 12:28 pm

Re: READ ME FIRST: What is a Sudoku Committee

Post by Mephisto » Mon 07 Jan, 2019 11:36 am

do we have to write a definition which include all previous wsc and gp puzzles
No! Obviously there is dissatisfaction with the scope of the WSC puzzles and this scope should be restricted. If we would consider all puzzles, only the status quo - with which some are dissatisfied - would freeze.

DavidC
Posts: 190
Joined: Wed 03 Nov, 2010 9:34 pm

Re: READ ME FIRST: What is a Sudoku Committee

Post by DavidC » Mon 07 Jan, 2019 8:13 pm

I wonder whether an Euler diagram would help the discussion? I seriously don't want to trigger a distracting Brexit discussion, but a few years ago I found that Euler diagrams were a great way of drawing out some of the subtle differences between different terms used to describe the supranational European groupings. In my daytime job I often have to compare the positioning of the UK, the Republic of Ireland, the Crown Dependencies, Gibraltar and Switzerland, for example. Maybe France, Italy, (West) Germany and the Benelux countries are the equivalent of Classic Sudoku. Maybe a Diagonal Sudoku is Spain or Portugal. Maybe a Kakuro is Iceland or Turkey. Maybe the equivalent of the Vatican City is so interesting that we should include it regardless. Maybe some of the really obscure unaccptable options are the equivalents of Papua New Guinea, Bhutan and Venezuela. Just a thought. https://en.wikipedia.org/wiki/Template: ... ean_Bodies

detuned
Posts: 1744
Joined: Mon 21 Jun, 2010 2:25 pm
Location: London, UK
Contact:

Re: READ ME FIRST: What is a Sudoku Committee

Post by detuned » Mon 07 Jan, 2019 10:44 pm

Yes I agree this is what we should do, and I think this is what Tom will do. If I understand, he wanted to start with discussions concerning examples to have a better idea when it comes to speaking about properties (not sure if it will succeed).
But yes, normally it should be the opposite way: to define what is a sudoku first, then with the definition we can hopefully have an answer to all examples. The problem is that the task which consist of defining what is a sudoku comes after 13 years of world sudoku championship (and 6 years of sudoku gp), and the puzzles of these competitions can sometimes be very far away from the initial thing we called sudoku. So the question is: do we have to write a definition which include all previous wsc and gp puzzles (good luck for the writing), or do we come back to the initial situation, as if we don't know all puzzles appearing in WSC/GP and write a more strict definition; the risk is then, if we look back, to see that some previous championships were far to be a sudoku competition according to the new definition.
First point: hopefully the collection of examples in this forum is the proof that it would be a bad idea to go straight to the definition first. The reason is we have such a wealth of creativity of different Sudoku variants created over the years that not to consider all this creativity would be akin to burying our heads in the sand.

As an example, I suspect that the Irregular Sudoku example for instance is something that many people would not have thought out, and may well have led people to rethink their own viewpoints and whether extra caveats need to be attached.

I also think that going at this purely dogmatically is not going to be the best way to win a broad consensus. It's one thing to come up with a definition that I or the committee thinks is great, but it will only be worth something if a large part of the community also agrees.

And yes, it would be strange to make a formal definition that retrospectively included everything from previous WSCs and previous Sudoku GPs. If I'm honest I don't think it's even possible to do meaningfully - I challenge anyone to come up with something not completely trivial which would include each of the examples already posted. Even if you could, I'm pretty sure I'd be able to go away and find another example which didn't fit. (Incidentally this is another reason to do as thorough a study of Sudoku variants as we can manage between us)

DavidC: I'm really not sure I understood the point of most of your post, but from what I understand of Euler diagrams, we're going to need a whole bunch of different categories first. Do you have any suggestions?

DavidC
Posts: 190
Joined: Wed 03 Nov, 2010 9:34 pm

Re: READ ME FIRST: What is a Sudoku Committee

Post by DavidC » Tue 08 Jan, 2019 12:07 am

I'm not sure how to insert a diagram image, so maybe I'll refer to plotting a colour-coded diagram with a RAG Status list instead (Red=no, Amber=borderline, Green=OK), with Classic in the centre, other Greens surrounding, Amber slightly further out and Red much further out

This is just an illustrative list, not a proposal.

Classic : Green
Diagonal : Green
Simple Rules (eg Anti-Knight) : Green
Regions (eg Renban, Windoku) : Green
Killer : Amber
Kenken : Amber
Kakuro : Red
Alphabetic : Red
Repeated : Red

Post Reply