Wow, nice article. I didn't care about the speed problem because when I wrote my program, computers were fast enough. I only wrote the fast algorithm eight years later in the perl rewrite. Otoh, I think my program generates better (more interesting) mazes.