Combatting child porn or just another censorware outfit?



The Guardian reports that British Telecom, “Britain’s largest high-speed internet provider” will begin censoring web sites to combat child pornography. The project is known as Cleanfeed and the article states that “who attempt to access illegal sites will receive an error message as if the page was unavailable.” The article implies that that list of illegal sites was compiled by the Internet Watch Foundation (IWF) but all of this remains unclear to me. While IWF does monitor sites that host child abuse images and report them to police it does not appear to be involved in the production of URL black lists, public lists anyway.

[NOTE: I have since been informed that this Cleanfeed is NOT the same Cleanfeed as mentioned in the article. So take the following merely as my thoughts on the Cleanfeed service and has been slightly edited as a result :]

Cleanfeed, however, is a “pre-filtered broadband Internet” service that maintains a database of more that 6 million (6280873 total domains to date) categorized domain names that are further classified as “banned” or “allowed”. The following is a quick, and less than thorough, assessment of Cleanfeed using its own site testing tool. In short, Cleanfeed only seems to categorize domain names, not full URL’s and is plagued with all sorts of (mis) classification issues.

Cleanfeed seems to only index domain — not full URL’s. This is significant for only the domain is categorized — any other content hosted on that domain also falls into this category. This is an obvious problem that will lead to major misclassifications and thus collateral blocking in several ways. One involves a situation in which a domain has been categorized in some “allowable” category but content hosted deeper within the website contains “banned” content. Because the root domain is “allowed” the “banned” content will also be accessible. For example, a check using Cleanfeed’s lookup tool shows that http://au.geocities.com/misssuziward/nudepics.html, which it reduces to au.geocities.com, and is allowed as part of the ‘Hosting Providers’ category — despite the fact that the URL is full of pornography. In turn, if such a domain does contain “banned” content, perhaps on one user account or somewhere deep with in the site, should the entire domain be blocked? What if an entire domain were blocked because one user or one page contained “banned” content?

In some cases such as playboy.com the IP address, 209.247.228.201, is also classified as ‘Sexually Explicit’ but the IP address form the domain www.photonudes.com, the domain itself is also classified as ‘Sexually Explicit’ but the IP address, 216.131.92.81, through which the same web site is “allowed” as part of the ‘General Allowed’ category. The same is true of www.anonymizer.com (the Anonymous web surfing service that can be used to circumvent filtering), which is banned as part of the category ‘Anonymous Proxies’ but accessible through it IP address 168.143.113.10 which is “allowed” as part of the ‘General Allowed’ category.

Another concern, which I have only touched on as I’ve yet to find an example, is the possibility that the blocking on an IP address would also effectively block all domain hosted (virtual hosting) on that same server.

As with other URL filtering databases there are numerous classification problems and some outright weirdness (Is www.viagra.com ‘Sex Education’? Maybe.)

Porn Allowed?

The URL redirection service http://www.kickme.to/ www.kickme.to is banned as ‘Sexually Explicit’ but all these sites are accessible because they are classified in the category ‘General Allowed’:

#[‘General Allowed’]
(‘Allowed:’, ‘http://www.giant-tits.info/’)
(‘Allowed:’, ‘http://www.cum-on-tits.biz/’)
(‘Allowed:’, ‘http://www.captainstabbin.org/’)
(‘Allowed:’, ‘http://www.citysnatch.com/’)
(‘Allowed:’, ‘http://www.teenbondageclub.com/’)
(‘Allowed:’, ‘http://www.der-ficker.de/’)
(‘Allowed:’, ‘http://www.phonesexyellowpages.com/’)
(‘Allowed:’, ‘http://www.free-anal-porn.com/’)
(‘Allowed:’, ‘http://www.sexnetzwerk.de/’)
(‘Allowed:’, ‘http://www.free-black-pussy-sex-pictures.com/’)
(‘Allowed:’, ‘http://www.adult-paysite-reviews.com/’)
(‘Allowed:’, ‘http://www.younghooters.com/’)
(‘Allowed:’, ‘http://www.milfbangvideos.com/’)
(‘Allowed:’, ‘http://www.free-gay-chat-rooms.net/’)
(‘Allowed:’, ‘http://www.formatureaudiencesonly.com/’)
(‘Allowed:’, ‘http://www.freebrunetteezine.com/’)
(‘Allowed:’, ‘http://www.youngnudecelebrities.com/’)

Different domain, same site, different category?

For example http://members.aol.com/ (which redirects to http://hometown.aol.com/ which is classified as ‘Hosting Providers’ is classified as ‘Vehicles & Transport’.

Another odd case involves classifying http://world.altavista.com/ as ‘Schools & Education’ and http://babel.altavista.com/ (same site, different domain) as ‘Anonymous Proxies’.

Another example is http://www.moqawama.org/ banned as ‘Hate Speech’ but the same site at http://www.moqawama.tv/ is allowed as category ‘General Allowed’.