Wednesday, January 16, 2008

The futility of censorship in the Internet age

It was a relief to read that at least one advocate of censorship in the current debate, Bernadette McMenamin, has been prepared to quite firmly state that her call for censorship is strictly limited to depictions of child sexual abuse and further that she seems genuinely interested in understanding the anger of the opposing camp. No doubt this maturity and openness to other points of view is symptomatic of the qualities that have seen her receive many plaudits and gongs in her career.

In the interests of mutual understanding, it is probably worth highlighting the things about which we do agree so that we can tightly focus the remaining debate on the issue of difference - the merits of a technical censorship regime.

I think it is fair to assume that everyone in this debate who is not made of straw would strongly agree with the following statements:

  • child sexual abuse is morally wrong and should be illegal
  • making, viewing or distributing depictions of child sexual abuse is morally wrong and should be illegal (legal censorship).

Where we start to differ is with this statement:

  • if an activity is morally wrong and/or illegal then depictions of this activity must be blocked by technical means (technical censorship), whatever the cost and no matter how futile such attempts will be

It is disagreement about this statement which is the major crux of the debate and the source of most of its heat. There are some in the anti-censorship camp that are opposed to technical censorship on philosophical grounds; there are others who are repulsed by the utter futility of any attempt to impose a technical censorship regime ; there are those who are frustrated by the ignorance of those who insist on imposing such a regime despite its inevitable futility.

For reasons of completeness, I will briefly mention one other source of heat in the debate - whether a mandatory technical censorship regime should apply to depictions of legal sexual activity between consenting adults. However, I will leave discussion of this issue to a later post. Again, it is pleasing to learn that Ms McMenamin has made clear she does not wish to participate in this aspect of the debate.

Without wishing to be patronising and in part because Ms McMenamin has demonstrated a willingness to be informed, I would like to explain one reason why any attempt to censor child pornography with an ISP-level filter is utterly futile, no matter how socially desirable such a filter would be.

It comes down to a technology called Tor that Mark Pesce recently introduced into the public debate.

Like many powerful technologies, and like the Internet itself, Tor is a technology which brings with it powerful potential for achieving a social good while at the same time presenting distressing potential for aiding and abetting a social ill.

I will focus briefly on the good Tor can bring, just to illustrate there is some. Then I will focus on the bad so as to shed some light on why ISP-level filtering will be so utterly ineffective at dealing with Internet-based distribution of illegal content.

Tor works by thoroughly disguising the ultimate origin of traffic that hits an Internet server. In this way it can be hard for an observer to prove that a given client even communicated with a server, much less determine what was communicated. This is a far stronger protection than simply hiding the contents of the connection which is what the familiar SSL-encryption provides.

If you are a whistleblower or dissident, the ability to cover your tracks like this can be quite useful. It allows you to send e-mail with a very much reduced risk that the e-mail can be traced, by technical means, back to its physical source.

Technology, however, is agnostic to its uses. The very capability that can be used to assist dissidents and whistleblowers can be used by consumers of illegal content to cover their own tracks, thereby helping them to evade detection and prosecution by law enforcement authorities, even if law enforcement authorities have tapped the server-end of the connection.

It gets worse, however. Tor has a feature called hidden services in which both the client and server remain hidden from each other yet are still be able to exchange content. In this case, law enforcement authorities that are tapping a client connection can't determine the IP address (much less the physical location of) of the server that a client is communicating with. The reason is - the client simply doesn't know and what you don't know, you can't reveal.

It is worth pointing out that Tor's hidden service feature can also be used for good. For example, by dissidents to organise resistance to a tyrannical regime. The fact that the Internet and Tor can be used for both good and bad purposes should be an unremarkable fact about Tor, the Internet and indeed technology in general. This fact has been true of every technology ever invented and it should be no surprise that this remains true for the Internet and technologies like Tor that it has spawned.

The power inherent within technologies like Tor is ultimately why ISP-level filtering is so futile. Technologies like Tor establish overlay networks which are effectively invisible to the regular internet. Even if there was such a thing as a perfect ISP-level filter that blocked every single bit of adult content on the regular Internet, illegal content flowing on a Tor overlay network would still be unimpeded because such traffic is literally indistinguishable from regular encrypted content of innocent nature.

Presumably some people reading this are thinking. Well that's simple - just block and outlaw Tor. The problem is that although you can frustrate a public Tor network you can't realistically block Tor or technologies like Tor, since Tor traffic is indistinguishable from other Internet traffic that is used to secure Internet banking, amongst other things. To defeat Tor you would have to outlaw all use of encryption and then solve the impossible problem of detecting and blocking covertly encrypted traffic.

The Internet itself was designed to route around brutish physical damage which is what makes ISP-level filtering policies technically challenging to implement. Attempts to surveil and censor the Internet have lead to the creation of technologies such as Tor and these make attempts at ISP-level filtering irrelevant, since ISP-level filters effectively can't see Tor-level traffic.

As John Gilmore said, and it really is no joke, "The Internet treats censorship as damage and routes around it".

If proponents of a technical censorship regime do not believe it is futile, then it incumbent upon them to demonstrate with reasoned argument why this is so. If they are unable to do this they must at least explain why society should commit itself to such an expensive, counter-productive programme in full knowledge of its likely futility.

No comments: