Yesterday I was stricken with a fascination for the methods for generating a word search puzzle. What's the best way to make a compact puzzle? What's the fastest way to get a certain percentage of words to overlap? How do we define "optimal" here? If it's coverage percentage, then an easy way to make the optimal puzzle is to put all the words in a line. If we put expected area, the count of letters, ever the actual area taken up then we can get numbers above one when we have many overlaps. I think that is a good quality to have. Even though we have a discrete grid of letters, if all the words are of sufficient length, then we know the maximum number of intersection points. Is it better to have four words intersect in six places or just one? So many questions and so many things to experiment with :D

You probably know where this is going. So, which repository have I created this time? Well, none yet. There will be one up in a little bit ;)

Edit: I've dicked around and finally just uploaded a repository, this one called wsearch . I procrastinated horribly, so there isn't really any working code there yet. Right now there is just a basic cellular automata that traverses a grid in a spiral outward starting at the center. Whoop-dee-ding. :)