EARLIER FEATURES

FEATURES CONTENTS

LATER FEATURES

READERS WRITE
CREATING AND SOLVING NONOGRAMS

Jan Wolter

I read your April 2005 article at http://www.icpug.org.uk/national/features/050424fe.htm with interest.

I have a free Nonogram web site similar to the playtsunami site mentioned in your article. It's at http://www.unixpapa.com/pbn/ . There are quite a few similar sites around.

Using the tools publicly available on that site, I've created some 125 puzzles, though many not of professional quality - I'm a better programmer than artist. I wouldn't describe them as "horribly difficult to create" at all. About half the time my first drawing proves solvable. If it doesn't solve there is usually only a little fiddling needed to make it work. Of course, that's partly because I've gained some experience and know what kinds of things to avoid to make the puzzle solvable.

Naturally, my site includes an automatic puzzle solver. Without that creating puzzles WOULD be horribly difficult. Though I get many puzzles to work on the first try, it isn't unusual for me to try many dozens of variations before I get an image that is solvable and reasonably attractive. If I had to solve each one manually, especially for a large or difficult puzzle, I'm sure I'd go quite mad. The more you want to refine the artwork, the more you'll have to do repeated solving.

The FAQ part of my web site gives a lot of information on how the solver works and how it is implemented in Javascript. You can use it on the web site, but only on puzzles you create yourself. It's an incredibly simple-minded program and not actually a full solver. It will solve 90% of all Nonograms. Others it will get stuck on, but usually the user can fill in a few spots that it was too stupid to notice and restart it to get a solution. It wouldn't be at all hard to extend it to be a full solver, but I have no use for that. It would solve puzzles that humans could only solve by making a guess by working out the consequences to see if it reaches a contradiction. When a contradiction happens it would have to erase back to the point where they made the guess. I wouldn't consider that a good puzzle. My half-witted solver is actually a bit stupider than a human and that makes it just right as a tool for a puzzle designer.

On my site I applied some devious programming tricks to make the thing run fast, but the basic program isn't difficult. I'd expect any second year programming student to be able to implement the stupid solver. Anyone who has had an intro AI course should be able to build a full solver.

Ed: Jan's site requires (free) registration to access many of the features, so I have not looked at it extensively. Nevertheless, it sound interesting.

TOP