Communications Minister Stephen Conroy announced last week that government's pilot of ISP-level internet filtering is about to go ahead,  with six ISPs named in the first phase of the trial. The Senator has for months refused to answer any questions or criticisms about the plan, citing this pilot as evidence that all concerns are being listened to and worked on. However, fundamental policy issues remain unaddressed, and some in the community, including Electronic Frontiers Australia, worry that the pilot will serve as little more than a smoke screen.

From the start, the Federal government has treated the internet like a new-fangled television, to which the old censorship regime can simply be extended by legislative fiat. They want to create a PG-rated internet channel that people in homes with kids can tune in to, and ban anything illegal or "unwanted" from the internet airwaves entirely. This fantasy demonstrates a fundamental misunderstanding of the role of internet service providers and the reality of how the internet actually works.

The internet isn't a broadcast medium, where a few media companies control a limited amount of content. Billions upon billions of web pages exist, millions are created every day, and ordinary users are responsible for much of the content. Text, audio, images and video are mixed together. And it's global, largely beyond the reach of Australian lawmakers. Once you understand this - as the government clearly does not - the idea of a "Department of the Internet" regulating all the content seems absurd, and technological solutions become impractical at best. 

The filtering policy has started to resemble the proverbial dog's breakfast. Among the questions yet to be answered: Where is the evidence of a cyber-safety emergency that requires government intervention instead of parental education? Why is an expensive national ISP filter preferable to cheap filters installed on home computers? What age level is the kid-friendly filter to be targeted at? Who will decide what material is to be banned? What software must ISPs use, and how is it to be deployed?

The upcoming pilot is supposed to address the issues, but what exactly is it a pilot of? Conroy’s statement says that "the live pilot will provide evidence on the real-world impacts of ISP content filtering, including for providers and internet users. It will provide evidence to assist the Government in the implementation of its policy." But the document released by the government states that it only requires participating ISPs to implement the mandatory tier, wherein the ISP is required to filter a few thousand web pages from a government blacklist. It does not specify how this is to be accomplished technically, leaving it up to the ISPs to sort out the details. The original filtering policy, providing a “clean feed” for families, is barely mentioned in the pilot specification. The legitimacy of the trial is further called into question by the ISPs selected to participate. Although Telstra, the nation's largest ISP, declined to take part, the second- and third-largest ISPs did apply. Yet neither Optus nor iiNet, both publicly critical of the proposal in the past, were selected for this round.

The initial round of ISPs are Primus Telecommunications, Tech 2U, Webshield, OMNIconnect, Netforce and Highway 1. Those chosen are by and large obscure and too small to credibly test the ability of the filtering scheme to scale to a national level. Estimates for the combined market share of all 6 ISPs are only a few percent at best. The largest of the six, Primus, has the most significant customer base, but they are only testing the mandatory, blacklist-based filter. 

The statement released last week states that testing “with each ISP will take place for a minimum six weeks once filtering equipment has been obtained and installed.” As ISPs are on their own in procuring and configuring the necessary servers and software, it could take some weeks before things get underway.

We are told this will make Australia's children safer. By focusing mainly on blocking access to illegal and other material only adults would try to access, the trial undermines this rationale. We already know, from the government's own tests that no software exists that can reliably filter the internet down to a level appropriate for a child. To even pursue filtering on a national level implies the existence of a generic Australian child for which a definition of 'appropriate' could be agreed upon by all parties. Even if there was such software (and it didn't have a drastic impact on network performance, as tests show) it would be embarrassingly easy to bypass.

As for blocking that illegal material, the conventional wisdom amongst experts is that very little child abuse material is published on the web, instead being traded by secretive communities using peer-to-peer technology that cannot be filtered. Were a web site to make it onto the blacklist, thus thwarting an attempt to access it, the perpetrator would simply have to use one of many free tools on the internet to bypass the ISP's filter. There is no suggestion even by Government that this filter would aid law enforcement, and nobody, including the ISPs themselves, has suggested there is any possibility that the pilot will tell a different story.

There are many considered and well-documented objections to the filtering policy. The government is taking a broad new censorship power for itself, without a mandate from the public or a cogent accounting of why we should allow it them. The trial will not put these questions and concerns to rest. Australia still demands answers. The government should first make a case to the electorate that the policy is even necessary, instead of leaping into a premature trial of unspecified technology. To date, rather than participate in the debate and engage with the plan's critics, the Minister has either stonewalled or accused his opponents of insufficient care for the welfare of children. If the policy was really a sound one, the government would not have to resort to such tactics.


  1. Although I agree with all the technical objections and the censorship angle. I am also worried about PRIVACY. Someone will be looking at all my traffic to see if I access anything bad. Even at the http level with web mail all my private email could be monitored for inappropriate material.
    In the trial, I have asked and not been answered, that I be assured, web mail be excluded. ISP users may give permission but if I send an email to that person and they reply with the usual copy of my email, why should not I also be asked for permission. Also whats stopping someone also capturing commercial information. Theres a lot of trust here that all players will do the right thing.

    This is not like banning a book from the shelf, its more like leaving it on the shelf and having someone watch me all the time to see if I choose it. It makes the "Australia card" trivial in comparison.

    Comment by Greg on 25 February 2009 at 00:04
  2. I wish this would just all go away. There were recently demonstrations in NZ over that blog law, yet I don't think it would happen here. How do they decide what we read/watch. A lot of these "Christian/Catholic" authorities are trying to force their morals on us. One does not have to look very far to read about all the sexual abuse cases. The corrupt governments etc.
    [ Joh Bjelke-Petersen, George Bush ]

    I am older both than Conroy & Rudd. So why should people who have little experience of the internet telling me how I should use it.
    Can Senator Conroy do and pass any of the brain bench tests? His technical knowledge seems to be non-existent.
    If one can get past the firewalls of China, this one will be easy to crack.
    Most child abuse seems to come from private schools..

    Comment by Nigel Colhoun on 25 February 2009 at 01:14
  3. The trial is supposed to evaluate the performance of the filtering technology, both with regard to accuracy and the latency imposed on the end user's web request.

    However, by selecting ISPs with such low numbers of subscribers, and only trialling subscribers that opt in to be filtered, the results will never reflect real-world scenarios and be inconclusive at best.

    To perform an accurate test of the filtering performance, the government needs to test the filtering environment they are proposing: i.e. mandatory filtering for all users on all requests. No opting in or out, let's test exactly what the government is proposing.

    Secondly, the test must be performed using example ISPs from all ranges in the market, including top-tier ISPs with millions of users.

    It is only by simulating the mandatory proposed filtered environment for a significant number of users that the trial can determine the accuracy and latency of the intended system.

    George Bray
    Australian National University

    Comment by George Bray on 27 February 2009 at 00:45