Page 1 of 1

What is a Sudoku? Report general consultation

Posted: Thu 12 Sep, 2019 12:31 am
by detuned
Hi everyone,

See this link for the current state of the report.  Sections 1-4 are complete, but 5 is incomplete (and will hopefully make the report more practical and understandable when complete).

Given the timing, I think now would be the time to open this up to a more general consultation.  There should be enough material there for people to get their teeth into, and to hopefully with this input shape the report into its final form. ... k8sVE/edit#

Re: What is a Sudoku? Report general consultation

Posted: Thu 12 Sep, 2019 8:32 am
by Feadoor
Tom, this is fantastic work - thank you. The pragmatic approach you propose makes a lot of sense to me, and I hope the WPF will take a similar view of it.

On an initial reading, some points jumped out to me as perhaps needing an edit or some extra clarification.

1) "Cells must have a finite number of well-defined edges, with each edge intersecting a maximum of one other cell."

It's not clear to me what "intersecting" means here. If you mean that edges should be shared between at most two cells, then this is not true of, for instance, Parquet Sudoku (unless you consider edges of "big" cells to be composed of smaller degenerate edges).

2) "Each such subdivision must be a one-dimensional group of cells..."

It might be worth making it clear (and perhaps you think the follow-up comments about a "chain of cells" and a "sense of direction" do this) that "bent" rows and columns are still one-dimensional in practice. A well-placed Sudo-Kurve example in Section 5 would reinforce this.

3) The word "substantially" and the phrase "substantially all".

This is used a lot in the "What a classic sudoku looks like" section. I think a degree of ambiguity is required here, and also that the Section 5 examples will go a long way to explaining what these terms mean in practice. But perhaps a short explanation before the first usage wouldn't go amiss.

4) "There must be exactly one way to place the symbols in the set S so that the solution of the puzzle satisfies Principle VI"

I think something is missing from this phrase e.g. "There must be exactly one way to place the symbols in the set S, _subject to any additional constraints_, so that the solution of the puzzle satisfies Principle VI"

Re: What is a Sudoku? Report general consultation

Posted: Fri 13 Sep, 2019 10:33 am
by detuned
Hi Sam, thanks very much for the feedback!

Agree on 1, the wording needs a bit of tightening up (and maybe also to try and use slightly less “technical” language

2 and 3 I’m hoping the final section will help clear up. You have to draw the line somewhere, and I’m certainly not going for full mathematical rigour here. I think it’ll be clear enough where this wording isn’t satisfied by a particular puzzle.

4 that’s an important clarification, thanks!

Re: What is a Sudoku? Report general consultation

Posted: Sun 15 Sep, 2019 9:59 am
by Fred76

I would like to do some comments on this report, perhaps it may help to improve on some points...

First, I would like to say that it is a huge and great work from Tom, I would like to thank him for this ! Depending on how this report is used, the WPF has now a chance to do something serious with sudoku competitions, I think all sudoku fans will be happy if it happens.

I really like the way the report put things in perspective, the structure of it is very clear and gives a clear view of the past and present situation, which should help to give sense to the core of the report which is the 3rd part of it.

I'll post more detailed comments or questions on some part of this report in next posts, but first important thing I would say is a big thank you to Tom Collyer for his serious work !


Re: Part 1: Introduction

Posted: Sun 15 Sep, 2019 12:03 pm
by Fred76
Here are some comments about part 1: Introduction.

I like the way puzzles are spread into the 3 categories (A) classic sudoku, (B) sudoku variant and (C) sudoku hybrid.
You all know I've some concerns about (C) in WPF competitions, but I agree that it is not the task of the report to say (C) is not appropriate in sudoku tournaments. Perhaps what we could say in the report is that the few puzzles of (C) categories that are submitted to a survey on this forum were generally rejected as being appropriate (look for example at the sukaku: viewtopic.php?f=33&t=1962).

I think this is the task of WPF to say something about (A), (B) and (C) in their competitions. Now that we have a report which define clearly enough what (A), (B) and (C) are, we at least are sure that we speak about same things.

I hope that the first positive effect of this report is that we'll no more see puzzles that aren't (A), (B) or (C) in sudoku competition. That will already be an improvement.


Re: Part 2: Summary of Definitions & Recommendations

Posted: Sun 15 Sep, 2019 12:27 pm
by Fred76
I have small concerns with formulation of definition 2 and 3 here. I think it is not a good formulation and that it can even be misleading if you don't read the part 3 of this report.
Definition 2 (Sudoku Variant). A Sudoku Variant is a puzzle which both looks like, and solves like a classic sudoku.

Definition 3 (Sudoku Hybrid). A Sudoku Hybrid is a puzzle which looks like, but does not solve like a classic sudoku.
I don't understand the reference to classic sudoku. If I had to say something about how looks like and solves like a classic sudoku, I would probably be very far from part 3 of the report.
I would say, probably, things like "it is a square grid formed by a 9x9 array of square cells. The grid is further divided into 3x3 regions." (this is a part of your definition of classic sudoku), or "it has given digits", etc... A lot of things that I would say if I take the definition 2 and 3 as true would be far to describe sudoku variants and hybrids.

In another way, if I had to describe to someone what is a sudoku, I would say exactly the contrary of your definition:
I would say a variant is "a puzzle that may look and solve very differently than classic sudoku" but has some principles in common (and then probably I would try to summarize these principles which are basically the one you wrote in part 3 of the report).

I understand that principles defining sudoku hybrid are about how the puzzles look like and principles defining sudoku variant are about how the puzzles look like and how the puzzles solve like and that these principles come from classic sudoku. But I think the formulation with reference to classic sudoku is not good.

That being said, if we read report with part 3, what you mean is clear enough and it's ok for me. My fear is that I hope nobody won't argue based on the rude definitions on part 2 of this report.


Re: Part 2: Summary of Definitions & Recommendations

Posted: Sun 15 Sep, 2019 12:48 pm
by Fred76
Still in part 2 of the report:
I do not consider it a useful exercise to define precisely and absolutely what “looks like” and “solves like” mean. Even if this is something that could be achieved with sufficient precision, I believe that this would come with at least two major disadvantages: namely (1) they would require a level of abstraction which makes the definitions hard to understand and harder to apply to anyone not theoretically minded; and (2) they would unnecessarily stifle and constrain the creativity of puzzle creators.
I think (2) is basically a false belief. I think we already had this discussion on your blog, but if you do some research about the subject, you'll find that constraints generally don't stifle creativity, but in contrary it stimulates the creativity.
Some referencesto this subject: ... aints.html, ... 831a538643, ... 588c8a8619.

This is actually something I noticed in the sudoku world. Very often authors fix themselves lot of constraints for their puzzles, it can be about symmetry or aestetic, minimalism, fit a theme, etc... and this produces in my opinion the most creative puzzles. Of course we don't want these kind of things in this report, but my point is that there is no reason to fear lack of creativity from constraints. The only point about constraint is how large we want the panel type of puzzles to be.

To add an anecdot to this, just few days after I read from a player this argument about "constraint stifle creativity", I was asked to create sudoku for the national competition of this player, and it is the only time the organizer imposed me to create sudoku that fit themes they had chosen.
Of course I had no kind of issue to create sudoku with such constraints, my creativity was not stifled.

In conclusion, I would say I would be happy if the argument (2) doesn't appear in the final version of this report.


Re: Part 3: Principles defining what a Sudoku “looks like” and “solves like”

Posted: Sun 15 Sep, 2019 1:01 pm
by Fred76
About part 3 of this report.

I think it's the core of this report, it was the probably the hardest part to write and it is very good. I have however question concerning:
X. Principle VI gives rise to classic sudoku solving techniques allowing the placement of symbols using the “Rows and Columns” and “Regions” types of subdivision. Where additional constraints introduce further solving techniques, they should:
A. Be complementary to classic sudoku solving techniques.
B. Not be entirely separate from classic sudoku solving techniques.
First, are both A AND B required, or A OR B? it is not clear to me.

To be honest, I don't understant what it means. Perhaps examples in part 5 will give answers, but I don't understand this whole point X.
(I don't understand the meaning of "techniques are complementary or not be entirely separate to classic sudoku solving techniques")


Re: part 4: Views from the WSC Community

Posted: Sun 15 Sep, 2019 1:27 pm
by Fred76
About part 4.

I like the existence of this part and the way it puts historical perspective to the report.

The Simon Prett article is primordial to my eyes, and the questions are still very actual. Even if the question posed was more about classic vs. variant at this time and now it is more about classic+variant vs. classic+variant+hybrid, this article still make think about what WSC should or could be and which place the WPF can give to this competition.
I would add that I'm disappointed about quality of recent WPF Newsletters, which no more includes valuable contents like this article.

Now about Neil Zussman quote:
Within the WPC community, there is a point of view that says that since Sudoku has split from the WPC to have its own competition (i.e. the WSC), Sudoku generally aren’t particularly welcome at the WPC.
I can't see the connection to Zussman post and "Sudoku generally aren’t particularly welcome at the WPC." perhaps I didn't understand well his words, but I don't find where the argument "no sudoku in WPC" is said.
About this argument "no sudoku in WPC", I've to say I heard it, but never found really someone who clearly or publicly stand for it. I would be happy if people that are for this argument manifest themselves (not that I doubt they exist, but...). I read the contrary argument "more sudoku in WPC" several times, mainly from N. Zivanovic (you find it on this forum too).
This is perhaps most succinctly expressed by Zussman who simply says if it ‘looks like’ a sudoku, it’s suitable for a sudoku competition.
I think you forgot to say that Zussman changed his opinion the next day after I created a puzzle that (more than) looked like a sudoku and was surely qualified by him as not appropriate, he then said:
Perhaps 'looks like a sudoku and has at least a semblence of sudoku logic' would be better. But then you'd have to define what a 'semblence of sudoku logic' is. To be fair, I did say I haven't thought out the details thoroughly. However, I am very much still in the camp that says lots of leeway should be given to constructors (except maybe not quite that much leeway).
To be honest, I'm still not sure what is the meaning of Neil's initial post. Would I be wrong if I caricature it by saying "Neil doesn't care about which kind of puzzle types appear in sudoku competitions"? And if it is the case, it is not a very useful opinion to answer the question "What is a sudoku?"


Re: What is a Sudoku? Report general consultation

Posted: Sun 15 Sep, 2019 2:18 pm
by detuned
Hi Fred, thanks for your comments, and I hope the finished report will better answer some of your questions.

Firstly - I think it is absolutely necessary to use classic sudoku as the base of the definitions. I perhaps take your point that the principles I have in mind probably work by themselves without reference to classic sudoku, but without that reference there is a great deal more interpretation to them. I think it's fair to say that many people have at least a subconscious idea of what sudoku looks and solves like, and I think this provides a very useful abbreviation to what are a more detailed set of principles. The way I see it is that hopefully the fully firepower of section 3 only needs to come out in the marginal and controversial cases.

Re your alternative phrasing:
In another way, if I had to describe to someone what is a sudoku, I would say exactly the contrary of your definition:
I would say a variant is "a puzzle that may look and solve very differently than classic sudoku" but has some principles in common (and then probably I would try to summarize these principles which are basically the one you wrote in part 3 of the report).
I don't think this is exactly contrary, I think it is just a change in emphasis. You say variants are different but there will also be some common ground, but I start off with that common ground and basically allow differences through generality. I tend to think it's a bit more intuitive to think about the common ground first rather than immediately concede that things can end up radically different. Cracking the Cryptic I think is the perfect demonstration that difficult variants can be perfectly accessible to people beyond the WSC community by focussing on that common ground.

Second on the side point of creativity. I generally agree with you that just because there are constraints, doesn't mean you can't be creative. On the other hand it is very easy to be very definitive with your constraints, and end up with unintended consequences or fiddly qualifications. I don't think you need to say no to any more than is absolutely necessary.

Re Principle X - this goes hand in hand with Principle XI which definitely does start drawing the line. The wording of X I think could be improved, but the general idea is that I don't want to have to make an exhaustive list of things which don't come under "solves like". It's also a bit stronger than that - I think there are some numerical constraints that start becoming too much about arithmetic and less about symbol placement which could at least now be questioned using this principle. Many of the examples in this forum would be good to include in this discussion, and I'll try to include a few of these for Section 5.