Primarily scraping of comet/ajax updating sites of course, but also some interesting uses for testing - crawling a site taking screenshots and then comparing screenshots across releases looking for layout bugs was quite a fun thing. The same principle as https://github.com/BBC-News/wraith