Censorware Filters Cause College To Change Name

lee writes: "Beaver College voted to change their name because censorware filters prevent people from browsing to beaver.edu. Full story at http://web.philly.com/content/inquirer/2000/06/13/city/BEAVER13.htm. To add insult to injury, the same censorware filters would probably not let anyone view the story, either..." We mentioned this story previously; just following up with the actual decision.

Comments

.edu in censorware?

[Twon]

I find it surprising that the .edu TLD is included in censorware at all, given that it's the only one actively restricted to its original purpose. Thoughts?

Common Thread

[LoonXTall]

There seems to be a common thread running through censorware: it doesn't work.

Look at the recent /. story on image filtering. Now this. There have probably been plenty in the past, too.

If any of these companies had been watching the fall of others, they would know they need to add the following things to their algorithms:

• Better heuristcs, perhaps including context. The average 5-year-old isn't going to think of censored material when you say "beaver", so the word on a page or in a domain shouldn't be auto-censored. Since "wood" is likely to be part of a legit description, beaver+wood shouldn't be auto-censored either. But if the page also contains lots of (he || she || his || her || him || other censorwords), then it's safe to stop. This system does necessitate the censorware sitting between the network and the browser, but it's far more accurate. Since some caching software sits there already, this isn't impossible.
• Phrase recognition. "You must be 18 years old to enter this site" for example. Might already be there.
• Force access of "/" before subdirectories. As a consequence of the above; then you can make sure the 18-to-enter message isn't bypassed. A minor inconvenience to get into a geocities site, but worthwhile.
• You cannot censor a URL. They're too inaccurate (cinderella.com).

-- LoonXTall