> Hmmm, sounds like genetic algorithms would be better in this case.
Not if we want *prove* that there aren't any better solutions to the puzzle.
Setting up a small script that will find a *very good* solution isn't a big
deal, but getting the set of best solutions is something else.
