Random ramblings

Tuesday December 09, 2008

Why the Great-British-Pædowall is a dumb idea

It seems all my HTTP traffic to various sites, including wikipedia, blogger and blogspot, is being transparently rerouted through a Squid proxy by my ISP. This proxy checks the URIs against a block-list provided by the Internet Watch Foundation, and returns 403s for any URIs deemed to contain potentially illegal material - particularly any material of pædohilic interest.

Now, don't get me wrong, I'll stand in line with everyone else to condemn those depraved and sick people who would participate in the abuse of children. However it seems, in all the calls of "Won't someone think of the children?", that we've managed to throw the important principle of proportionality out the window by filtering everyone's internet. This is just a fantastically dumb idea, given the low efficacy of this system relative to the impact and risks imposed on law-abiding society generally.

It's still trivial to share illicit images via HTTP, as there's just no way the IWF can stay ahead of all the new images posted to all the various image and file sharing sites across the internet. Even if they could make a dent on HTTP file-sharing, there are various other protocols - some even designed specifically for encrypted sharing of files.

The impact of this filtering system on generally, law-abiding users:

• Adversely affects latency and throughput of internet traffic, HTTP particularly, but all other traffic too
• Adds complexity and so can not avoid decreasing forwarding reliability of all internet traffic, and even more so for HTTP traffic
• Increases costs, particularly for smaller ISPs:
  • The IWL charge £5000/annum for their block-list
  • More expensive routers may be required, capable of matching and rerouting certain HTTP traffic based on their IP address
  • Additional servers are required, to run the URI matching proxy
  • Requires traffic be backhauled to the virtual-ISP (rather than opt to have BT manage the IP layer)
  • Decreased reliability incurs indirects costs on customers when there are outages.
  • The cost of collateral-damage: E.g. many british wikipedians found themselves unable to edit wikipedia because of bans on their ISPs proxy address[1]. This is just a high-profile example, most will go unreported.
• I know what you're browsing

  The filtering proxies must maintain state on all active HTTP requests, which becomes easily available (at a minimum) to employees of the ISP. Further, poor security practices and/or configuration mistakes can allow this information to be viewed by others (e.g. all customers of the ISP, as was the case for quite a while with at least one UK ISP). Obviously, historical logging of requests is trivial to enable.

• Mission-creep

  Systems deployed today to filter out child-abuse images may tomorrow be appropriated for less universally welcomed purposes. E.g. why stop at child-abuse, why not track copyright infringers (another horseman of the internet apocalypse)? This infrastructure could very quickly be appropriated for more sinister purposes.

Basically:

• Efficacy: 0.
• Impact: high.
• Apparatus ready to be abused by next unscrupulous PM: Installed.

1. Even funnier/tragic is that the IWF blacklisted an article URI, rather than the URI of the offensive image - so the latter is still viewable (e.g. use google cache to view the article). Basically, the UK have made people who seem ignorant of how the internet works the gatekeepers to it.

Update: There's a really interesting thread at UKCrypto about this - posts by Clive Feather and Peter Sommer are particularly interesting (thanks to murb).

There's another aspect to all of this: As the IWF, though governmentally-recognised, are a private organisation they are not covered by the Freedom of Information Act. So this system is completely out of the reach of the powers of oversight available to the general public, despite it having been put in the place at the behest of the government, by the threat of regulation.

Also, its worth noting that one possible argument for the efficacy of this system is that it protects ordinary people from accidentally being exposed to this material. However this argument appears to be struck down by, apparently learned, commentators in the above discussion who point out that is extremely rare (to the point of being almost unheard of) to accidently stumble on child abuse images.

Certainly, in my experience of using the internet for 14 odd years, I don't recall ever seeing anything approaching such. At least, not until the IWF managed to publicise a certain image on Wikipedia...