Hand-made versus computer-generated puzzles

Everything about Sudoku that doesn't fit in one of the other sections

Hand-made versus computer-generated puzzles

Postby lunababy_moonchild » Wed Oct 05, 2005 9:28 pm

So, in your opinion are the hand-made puzzles superior (as claimed) to the computer generated ones? Insofaras both are created by someone reliable like say, Nikoli for the hand-created and Pappocom for the computer-generated?

I'd just love to know:D

Luna
lunababy_moonchild
 
Posts: 659
Joined: 23 March 2005

Postby PaulIQ164 » Wed Oct 05, 2005 9:54 pm

Can't say as I've noticed a huge difference really. I'd have to side with 'no', though I don't doubt many will disagree.
PaulIQ164
 
Posts: 533
Joined: 16 July 2005

Postby dukuso » Thu Oct 06, 2005 5:41 am

PaulIQ164 wrote:Can't say as I've noticed a huge difference really. I'd have to side with 'no', though I don't doubt many will disagree.


hmm, triple negation. So you assume most will vote 'yes'.

I wonder whether the main reason for hand-made sudokus
is copyright. Some people seem to think, that handmade
sudokus can be copyrighted while computer made sudokus
can't.
dukuso
 
Posts: 479
Joined: 25 June 2005

Postby PaulIQ164 » Thu Oct 06, 2005 10:25 am

Heh, I didn't notice that little grammatical anomaly as I typed it.

I imagine the main reason to hand-make sudokus is that you can sell them as hand-made.
PaulIQ164
 
Posts: 533
Joined: 16 July 2005

Postby dukuso » Thu Oct 06, 2005 10:44 am

what's the average price for one hand-made sudoku
compared with one computer-made sudoku on your local market ?
dukuso
 
Posts: 479
Joined: 25 June 2005

Postby Sue De Coq » Thu Oct 06, 2005 11:13 am

I'm always a little confused by people who claim to have a strong preference for human-generated puzzles (e.g. Nikoli) over computer-generated puzzles because, given a single puzzle, I've no idea how in a blind test anyone would ascertain how the puzzle had been generated. (Any suggestions from the pro-human camp?) Of course, some puzzles are signed by the author/software but in that case the judgement of the puzzler might have been affected by his prior knowledge and the test would not be rigorous.

It would be more interesting to attempt to define the word 'good' with respect to Sudoku puzzles, i.e. to list those characteristics possessed by the 'better' puzzles. Puzzlers who side strongly with one camp ought to be able to explain why their puzzles are so much better.
Sue De Coq
 
Posts: 93
Joined: 01 April 2005

Human composed Sudoku vs Machine made

Postby Pat » Thu Oct 06, 2005 11:37 am

just a reminder that this topic has already been discussed (in the remote past):
Human composed Sudoku vs Machine made
User avatar
Pat
 
Posts: 3441
Joined: 18 July 2005

Postby lunababy_moonchild » Thu Oct 06, 2005 2:46 pm

Sue De Coq wrote:I'm always a little confused by people who claim to have a strong preference for human-generated puzzles (e.g. Nikoli) over computer-generated puzzles because, given a single puzzle, I've no idea how in a blind test anyone would ascertain how the puzzle had been generated. (Any suggestions from the pro-human camp?)

Not sure if I'm convinced one way or another, but I can say the Nikoli ones that I'm doing feel different, somehow, to the Pappocom computer generated ones. I've done the Daily Mail ones and gave up because they left me cold - whether they are hand-created or computer generated I don't actually know. I do know that you can tell if the crossword setter has changed just by doing the crossword, so I wondered if you could do the same with Sudoku.

Sue De Coq wrote:Of course, some puzzles are signed by the author/software but in that case the judgement of the puzzler might have been affected by his prior knowledge and the test would not be rigorous.

There is that. I knew that the hand-crafted ones were Nikoli before I tried them so it's not really a good test (which is why I asked the poll question).

Sue De Coq wrote:It would be more interesting to attempt to define the word 'good' with respect to Sudoku puzzles, i.e. to list those characteristics possessed by the 'better' puzzles. Puzzlers who side strongly with one camp ought to be able to explain why their puzzles are so much better.

This is more what I'm after. It is generally held that balanced puzzles with one solution are better - both first thought up by Nikoli, apparently. And Nikoli do do a fair bit on their website to explain why they think that their hand-created puzzles are better. Not seen anything on this website about it, though. It would be interesting to see.

Pat,
Sorry didn't realise this. I had intended this to be just a poll, though. Thanks for pointing it out and I'll check the thread. Upon checking said thread, it seems that I was in that particular discussion! Jings, crivvens, help ma boab (as we say here in Scotland), must get my brain checked.:D

Luna
lunababy_moonchild
 
Posts: 659
Joined: 23 March 2005

Postby dukuso » Thu Oct 06, 2005 4:05 pm

OK, I followed the links to Nikoli :
http://www.nikoli.co.jp/puzzles/1/hand_made_sudoku-e.htm

I don't understand their argument so well.
They give an example, which they think is a "bad" sudoku,
but they don't give an example of a "good" sudoku.

The bad thing could be, that the first step were too difficult,
while the following steps were too easy ?

That would be pretty easy to avoid by computer too.
Indeed, once you know, what you would consider a good sudoku
you should be able to implement it.

Their statement, that computers will never be able
to make good sudokus like they do reminds me
a lot to the laughing at those early chess-computers,
which had no positional understanding etc.

To me it's clear that the days of handmade sudokus are numbered.
Their might indeed actually be some differences,
the Simonis paper
http://www.icparc.ic.ac.uk/~hs/sudoku.pdf
seems to show that the handmade are
i.e. harder to solve with "forward checking plus shaving".
Although I don't know whether a human can recognize this.
Last edited by dukuso on Thu Oct 06, 2005 1:50 pm, edited 1 time in total.
dukuso
 
Posts: 479
Joined: 25 June 2005

Postby PaulIQ164 » Thu Oct 06, 2005 4:12 pm

Some handmade sudokus have nice little patterns in the numbers - I just solved one where the top "chute" looked like this:

Code: Select all
1 2¦ 3 ¦4 
   ¦   ¦  9
 5 ¦6 7¦ 8


but you could do that to a computer puzzle by doing a simple permutation of the digits on the finished puzzle if you wanted.

(edit: alright, I didn't solve it. I looked at it though.)
PaulIQ164
 
Posts: 533
Joined: 16 July 2005


Return to General