The Porn Block Fiasco goes mainstream


The knowledge that attempting to block porn on the internet is bound to backfire has now gone mainstream. (BBC News, Telegraph) Well, there’s a temptation to say “we told you so”, because we did. Repeatedly.

So far, sites we know that are subject to overblocking on either TalkTalk and BT include BishUK (a sexual education site for teenagers), LGBTfriend, Edinburgh Women’s Rape and Sexual Abuse Centre, Sexual Abuse Scotland, Doncaster Domestic Abuse Helpline and Reducing The Risk (Another domestic abuse help site).

At the same time, the filters failed to do what they were supposed to – when BBC Newsnight tested on TalkTalk, 7% of porn sites were still accessible.

The trouble is one of resource – the same problem that makes “Report Abuse” buttons problematic. As of the end of 2012, there were over 600 million web sites on the internet. We’re probably over a billion by now.

If all 2,500 TalkTalk employees spent their entire time checking web sites, averaging one minute per site to classify it, it would still take over three years to check a billion web pages – by which time there would be another billion sites to check. You can filter this some as only around a third to a quarter of sites have unique content, but even with 2,500 staff you’ll never be able to keep up with new content.

The solution will inevitably involve technology, perhaps with some human input for the top 0.01% of sites. (One person can probably get through that much in about a year) But it’s only in the last three or four years that the so-called Scunthorpe Problem has been mostly solved, with notable recent relapses including Virgin Media censoring TV programme descriptions in 2011 if they included “anal” (As in “Arsenal”) and “dick”. (As in Dickens)

Given these problems, is it any wonder that automated filters are going to get it wrong spectacularly often? It doesn’t help that the whole system is shrouded in secrecy, with no notification of blocking, no way of checking what sites are being blocked and no clear appeals process.

The problems with such systems have been well known within the industry for years, which will have been a large reason why ISPs resisted implementing these filters and only did so under pressure from government ministers.

Oh, and as to the most serious problems facing children on the internet – Grooming, Cyberbullying, etc? Filters don’t help there at all.

  1. #1 by sigma on 19 December 2013 - 22:02

    Remember that TalkTalk have repurposed what was originally called their “Virus alerts scanning engine” to provide content filtering. The technology comes from their partner Huawei and was originally marketed by Huawei as GreenNet.

    Huawei brochure:
    http://www.huawei.com/ucmf/groups/public/documents/attachments/hw_111690.pdf

    Multiple value-added service capabilities:
    Provides carriers with the Green Net solution on the network side based on flexible control over subscribers and applications.
    URL filtering:
    Based on the URL category library, the URL access
    control capability (time + parental control) helps carrier filter out pornographic and violent information on the Internet, to construct a healthy Internet environment and exalt carriers’ social images.
    Application control:
    Based on the DPI technology, the network application access control capability (time + parental control) helps carrier control applications such as online
    games and online chatting, to prevent Internet addiction and safeguard juvenile‚Äôs growth”

    Basically,
    1. URLs are collected from network traffic and passed to a filter. TalkTalk claim that parameters that contain personal data are stripped out, but there is no evidence of this.
    2. The filter checks the URL against a black/white list. If its not known on either list, it passes it to a work queue for onward analysis.
    3. The profiling system picks the URLs from the work queue, access (or attempt to access) the URL in order to profile the site contents, profile the content, then updates the black/white list. In other words, it replays your browser request. In addidion to local TalkTalk back and white lists, the URLs are passed back to Huawei to update their master lists.

    Here’s some more liks you might find useful.
    http://broabandtrafficmanagement.blogspot.co.uk/2010/07/dpi-deployment-15-talktalk-use-huawei.html

    http://www.theregister.co.uk/2010/07/26/talktalk_stalkstalk/

    http://www.ispreview.co.uk/story/2010/08/23/uk-isp-talktalk-defends-website-url-tracking-system-from-privacy-concerns.html

    http://www.v13.gr/blog/?p=161

    http://blogs.fluidinfo.com/terry/2012/12/05/destructive-invasive-and-dangerous-behavior-by-uk-isp-talktalk-aka-stalkstalk/

  2. #2 by Ian Eiloart on 20 December 2013 - 12:13

    Also blocked is the Liberal Democrats party’s LGBT web site at lgbt.libdems.co.uk

  3. #3 by The Shonko Kid on 20 December 2013 - 12:39

    Blacklists or content filtering will ALWAYS fail miserably with both false positives and negatives. Anyone relying on this to protect anyone from anything will be disappointed, and I hope, open to litigation.

    The only sure fire _technical_ solution is a whitelist; only allow access to pre-verified sites and content. The internet used to have such a thing, AOL, and has already passed judgement on what it thinks of walled gardens and the like.

    Otherwise, I’m afraid it’s down to parental supervision and guidance, if you want to protect the kids from what’s on the internet.

  4. #4 by Steve on 20 December 2013 - 12:57

    They don’t want to ‘protect the kids from what’s on the internet’ though do they. That’s not the objective.
    The objective is to protect the powerful from what’s on the internet – ie: us.

(will not be published)


%d bloggers like this: