The New Sudoku Players' Forum

[chuckfresno]

Having solved Sudoku's on and off for over ten years, I can tell you that there is a *clear* difference in solving a puzzle hand made by an experienced composer from solving one made by computer. It isn't subtle, it's obvious. I'm not referring just to the look of the puzzle, though the handmade japanese sudoku's often have striking arrangements -- not merely symmetrical, but in an obvious shape or picture. I'm talking about the way a clever composition may lead you down a path which always leads to an "ah HA" moment. (This is especially evident in the larger puzzles of mutliple overlapping grids, but is also apparent in the standard 9x9.)

I have no connection to Nikoli -- or any other puzzle making or selling entity. I'm just a solver. Read these comments from the top people at Nikoli, the creators of the puzzle. I absolutely agree with them:

http://www.nikoli.co.jp/puzzles/1/hand_made_sudoku-e.htm

I'm not some spacey, new-age, artsy type. I'm just a nerdy math geek who likes a good puzzle. If you like your crossword puzzles made by computer, if you like your artwork made by computer, then I guess machine made Sudoku are good enough.

Nikoli has 19 books of Sudoku for sale -- each with about 100 puzzles for about 600 yen plus shipping. As far as I know, all the puzzles in these books are standard 9x9 grids rated with 10 degrees of difficulty. The NIKOLI COMMUNICATION -- their main, quarterly magazine -- typically has only 6 or 7 Sudoku and may have a 16x16 or 25x25 Sudoku as well. Sudoku are only a fraction of the puzzles they create. This magazine includes about 20 types of puzzles on a regular basis and another 20-40 on an irregular basis, the large majority of which can be solved with no knowledge of Japanese -- just logic.

http://www.nikoli.co.jp/storage/howtoget-e.htm

There are about a dozen or so monthly or bi-montly Japanese Number Place magazines. A Japanese bookstore may be of help. They typically contain 150-200 puzzles per issue, with quite a few variations -- larger grids, overlapping grids, non-square sub-areas, disjoint areas, sequential grids (you have to solve the first one in order to get an additional clue to solve the next, etc), various other types of clues such as even or odd cells, greater-than or less-than neighboring cells, and on and on. The vast majority require no knowlege of Japanese.

[simes]

I don't totally agree. But first the disclaimer - I haven't solved many by hand, so perhaps I'm talking rubbish. Feel free to disagree!

I would say it's down to the sophistication of the programmer that wrote the generator. Judging from the puzzles I've seen posted here, I think Pappocom's program does a great job. Several different techniques are required to solve the puzzles it generates.

On the other hand, the "hand-crafted" Guardian puzzles seem to require only the most-obvious, basic techniques. Or have I just been lucky?

[Pappocom]

I propose a "Pepsi challenge". Let's say we take 10 Nikoli puzzles and 10 Pappocom puzzles, give them in random order/s to 10 solvers, and then ask them to say which were hand crafted and which weren't.

I'm only half-serious about the challenge... but if it could be arranged, the results would be illuminating.

A cobbler will always say that leather is best. Audio-freaks will always say that black vinyl beats CDs. It's in the nature of change and progress.

- Wayne

[Roz]

I bought a Suduko magazine published by Puzzler magazine in association with Nikoli about six weeks ago, it was labeled a WH Smith exclusive(UK book shop chain). It was that magazine which got me hooked. However, I then found the program which having a greater range of levels, and difficulty, gives me the means to strive to improve my game. So as things stand I personally prefer the program.

[shakers]

I propose a "Pepsi challenge". Let's say we take 10 Nikoli puzzles and 10 Pappocom puzzles, give them in random order/s to 10 solvers, and then ask them to say which were hand crafted and which weren't.

I'm only half-serious about the challenge... but if it could be arranged, the results would be illuminating.

If this were to be arranged I'd be interested in playing later in summer; having seen the claims in the Guardian, I would be interested in the results.

Not having looked at the Nikoli puzzles, can I assume that they would be solvable using the Sudoku program?

[Pappocom]

Nikoli = Guardian = Nikoli.

Nikoli puzzles feature heavily in my database of over 25,000 Sudoku puzzles published world-wide over the past 5 or 6 years. With one exception, the Nikoli puzzles can all be solved by the Pappocom program.

The one exception requires Trial and Error. This was confirmed to me by Nikoli. It was obviously a momentary lapse by Nikoli, a company which normally has very high standards.

- Wayne

[simes]

Shakers,

Just in case you didn't know, the Nikoli Guardian puzzles can be found at http://www.guardian.co.uk/sudoku

[shakers]

Shakers,

Just in case you didn't know, the Nikoli Guardian puzzles can be found at http://www.guardian.co.uk/sudoku

Simes
http://www.simes.clara.co.uk/programs/sudoku.htm

Thanks - I found them about two minutes ago following the link in your sig, and seeing them listed there!

I'll give them a go at some point, but I still like Wayne's idea of a blind taste test. Note to self: remember to order some more time in each day.

[chuckfresno]

[quote="simes"]I don't totally agree. But first the disclaimer - I haven't solved many by hand, so perhaps I'm talking rubbish. Feel free to disagree!

I would say it's down to the sophistication of the programmer that wrote the generator. Judging from the puzzles I've seen posted here, I think Pappocom's program does a great job. Several different techniques are required to solve the puzzles it generates.

On the other hand, the "hand-crafted" Guardian puzzles seem to require only the most-obvious, basic techniques. Or have I just been lucky?

Simes
[url=http://www.simes.clara.co.uk/programs/sudoku.htm[/quote]]http://www.simes.clara.co.uk/programs/sudoku.htm[/quote][/url]

I'm not refering to the difficulty of the puzzles or the solving techniques required. I wasn't clear. One can easily create a piece of software that will create unlimited, correct, symmetrical crossword puzzles of any degree of difficulty, and they will be just fine to solve for most solvers, most of the time. Each one will be completely unmemorable, indistinguishable from the next. It is more difficult to create a program that would include clever worldplay or other linguistic twists that could make it special. Those of us who drink beer or wine infrequently cannot distinguish between the best in the world and the best in cupboard. That's fine. But I would never dream of telling a wine connoisseur that she's just fooling herself, that they all taste about the same.

I did not mean to imply that I believe that a computer CANNOT eventually create ANYTHING that a human can, and if programmed well enough, that the results will be as-good-as and/or indistinguishable from the human's efforts. I don't see that has been reached in this case. Just as you can only speak from your experience, I can only speak from mine. I have strong doubts that anyone solving puzzles created by this software exclusivley will still find them enjoyable to solve a year from now, that they will find them repetitive, redundant and repetitive. And if I drink enough wine, I may very well develop a palate that can distinguish between good, very good and exceptional wine -- even if I never attempt to aquire this skill.

[jafosei]

Interesting discussion.

I'm not referring just to the look of the puzzle, though the handmade japanese sudoku's often have striking arrangements -- not merely symmetrical, but in an obvious shape or picture.

I went and looked at some of the Guardian puzzles, and I must say that they do have a very nice aesthetic quality to them that isn't apparent in most of the other puzzles I've seen. I think my favorite of the ones available on the Guardian's Sudoku page was this one from June 4th:

Code: Select all

......1..
.5...2.4.
6...3.5.7
.....6.8.
..1...9..
.2.4.....
3.5.7...6
.6.8...2.
..9......

It looks great. The other puzzles I looked at also look wonderful.

On the other hand, the "hand-crafted" Guardian puzzles seem to require only the most-obvious, basic techniques. Or have I just been lucky?

I dubbed several of them into Pappocom's program, and a couple of the were rated 'Hard', so presumably they can be as tough as the ones in the Times.

I've only tried one of them so far (rated easy), and didn't notice anything different about solving it than solving one of the Pappocom puzzles. I'm not a particularly gifted solver, though, so perhaps I lack the talent to appreciate the subtlety involved, or perhaps it's only evident on the harder puzzles. I'll definitely try more of them, as I'm curious if there would be a noticeable difference.

[chuckfresno]

[quote="Pappocom"]I propose a "Pepsi challenge". Let's say we take 10 Nikoli puzzles and 10 Pappocom puzzles, give them in random order/s to 10 solvers, and then ask them to say which were hand crafted and which weren't.

I'm only half-serious about the challenge... but if it could be arranged, the results would be illuminating.

A cobbler will always say that leather is best. Audio-freaks will always say that black vinyl beats CDs. It's in the nature of change and progress.

- Wayne[/quote]

I would take that challenge -- but as jafosei points out, Genuine Nikoli Sudoku often have a more aesthetic quality that I cannot quantify to their symmetry. I'm flipping through a Nikoli book now and I'd say that if you presented me with 20 problems, 10 taken randomly from one of these books and 10 created randomly from your software, I'd be able to pick out about 8 of the Nikolis without trying to solve them. You would probably be able to do this even better than I would.

As far as solving differences, I admit freely that I would not always notice something special in every puzzle. They can't all be gems -- and some gems remain hidden. But I like the special feelling I get when I realize some little trick the composer used. I do not wish to imply that Nikoli's puzzles are perfect and that Wayne's are rubbish, only that there do exist increasingly subtle differences between good, great and exceptional puzzles, and that the more you solve, the more those subtlties become apparant and eventually more important.

And as an audio-freak with many audio-freak friends -- we ALL prefer CDs (and MP3s) to vinyl. The vinyl lovers are a very small group -- I think they might be a myth. I do, however, prefer music and lyrics created by a human over those created by software. It's not merely tonal quality, tempo, meter and rhyme. Some of my favorite music might lose in a contest with a machine if those were the only qualities measured. I want to feel *communicated* with. And I prefer peach over leather.

[chuckfresno]

[quote="Pappocom"]Nikoli = Guardian = Nikoli.

Nikoli puzzles feature heavily in my database of over 25,000 Sudoku puzzles published world-wide over the past 5 or 6 years. With one exception, the Nikoli puzzles can all be solved by the Pappocom program.

The one exception requires Trial and Error. This was confirmed to me by Nikoli. It was obviously a momentary lapse by Nikoli, a company which normally has very high standards.

- Wayne[/quote]

Speaiking of which, why do you usename "Sudoku" for your software and website without giving ANY credit to Nikoli within either? "Sudoku" is their registered trademark -- if there are legal reasons that this doesn't apply to you, shouldn't least give them a nod? (Non-Nikoli Japanese puzzles magazines don't use the word, the call them Number Place. I don't know if they mention Nikoli -- I don't speak Japanese.) Not that you have to have a link to their site or mention that they have sell Genuine Sudoko books over the internet -- but it would be nice if you made it clear that you are neither the inventor of the puzzle nor the creator of the name.

Actually, you imply very stongly, just the opposite, and I quote your banner: "Sudoko one of the Puzzles of Pappocom". Any casual visitor should assume you are the sole creator of Sudoku. And "one of"? What other puzzles do you have in the works? Any that you have created? Or should we expect a dozen of Nikoli's other puzzles (they've created some 200 over the years) to be co-opted? Would you create additional domain names using other of their registered trademarks? Why note take nikoli.com? Even if you can legally -- you are the Judge after all -- can you morally? I doubt you'd lose any sales if you added a classy "thanks" to the creators.

[scrose]

Speaiking of which, why do you usename "Sudoku" for your software and website without giving ANY credit to Nikoli within either? "Sudoku" is their registered trademark -- if there are legal reasons that this doesn't apply to you, shouldn't least give them a nod?

Someone (or Wayne), please correct me if I am wrong, but I have read that Nikoli holds the trademark for "Sudoku" only in Japan itself.

[Pappocom]

I am not going to get involved in a discussion of copyright issues. The subject of Intellectual Property is complex and this is not the proper forum ("forum", in the legal sense) to discuss it.

When I met with Nikoli again in Tokyo at the end of May, they told me that they were very happy that I had done them the honor of using the word "sudoku". They were amused by the thought that there was a new "Japanese" word in the English language.

And why wouldn't they be happy! Look at the market Pappocom has opened up for them. For example, had you heard of Nikoli or Sudoku before November last year?

- Wayne

[lunababy_moonchild]

I'd just like to point out that Pappocom isn't the only person/organisation that's using the word Sudoku over here. Not that that makes it OK of course. And if Nikoli were that upset over it they would do something about it - it seems to me.

Luna