I recall a mention by Jeff and Joel, perhaps on the podcast, of forwarding requests from web-crawlers to a dedicated environment.
The idea was to serve the same content as the real site, but to have more control over the format for SEO optimization as well as to keep the crawlers from overwhelming the "live" site.
I'd like to know if there is any additional descriptions of this approach and whether it was just a passing thought or an idea that was implemented.