Geoffrey Nunberg has a great New York Times column, also available from Google Groups, regarding library censorware.
"Computers in Libraries Make Moral Judgments, Selectively"
Excellent points throughout, for example:
Software will never be able to wholly reproduce human linguistic and perceptual capacities, much less distinguish between a Playboy calendar and an Edward Weston nude, or between "Tropic of Cancer" and "Trailer Park Swappers."
Then, too, the architecture of the Internet itself requires filters to block hundreds of thousands of sites that they haven't identified as porn - Google cache sites, for example, and any site that is unlucky enough to be hosted by the same computer that's hosting a porn site.
By the way, as a note of personal credit, that "Google cache sites" aspect was exposed by me:
BESS vs The Google Search Engine (Cache, Groups, Images)
BESS bans cached web pages, passes porn in groups, and considers all
image searching to be pornography.
http://sethf.com/anticensorware/bess/google.php