The idea is this, a lot of websites have dynamic pages.  Take these forums for example, the "Forum" link at the top leads to a certain page, with the same "Forum" link at the top - so if a spider kept going through the link, it would be stuck in an infinite loop.  Thus I guess the idea is that the spider won't continually re-index pages it has indexed before.  Another possibility is that it uses a priority system, so, although it may be stuck in an "infinite loop", it won't really have any effect.

Other argument is that I'm sure people have tried it with Google and Yahoo, and I doubt those multi-billion dollar companies would be brought down by such a simple trick.  Might work on lesser intelligent spiders I guess.