In the end, it took Senator Stephen Conroy and the Department of Broadband, Communications and the Digital Economy a whole year to manage a six-week trial of ISP-level filtering.
Coincident with the 215th anniversary of the US Bill of Rights, swamped in news coverage of Copenhagen and office Christmas parties, Conroy finally released the report he’s been sitting on since September.
The timetable blowout wasn’t a good sign and anyway, I’d been telling people that the results were a given before the trial even started. It seemed to me inevitable that the trial would find that net censorship is made out of unicorns and rainbows. And, wouldn’t you know it, the document claims that everything is sweetness and light and that every internet user will get a free pony.
Or does it?
The trial, part of the Government’s Cyber Safety Plan, didn’t get off to a flying start. Not only did it face staunch criticism from pretty much everyone in the internet business in Australia, its terms of reference were so loosely defined that it seemed unlikely to deliver conclusive evidence one way or the other. Officially, the trial sought to assess the technical feasibility of ISP-based web filtering — would censoring content have any impact on performance and productivity?
Unfortunately, the authors of the report, which is based on trials managed by independent testing laboratory Enex Testlab, have withheld some of the data about participating users needed to make sense of the results.
For example, we know from comments made by one of the ISPs involved in the trial, Nelson Bay Online, that only around 15 of their customers chose to participate. Their performance testing results were likely to be close to perfect because their Marshal8e6 R3000 system was never placed under any significant load.
But what of the other ISPs who took part in the trial? Primus has a much bigger customer base than Nelson Bay Online — did their testing involve just a handful of users? And if so, were they the ones that suffered a 27.2 per cent performance degradation — or was Primus one of the ISPs using the imaginary magical censorbox that apparently made the internet 17.32 per cent faster?
We don’t know the answers to these questions and that makes it hard to assess the technical findings of the report. Nonetheless, it’s clear that claims that the clean feed will have a "negligible" impact on performance are preposterous nonsense. The loose terms of reference for the trial are matched by Enex’s loose and fearless definition of the term "negligible"; it’s one with which I’d wager most internet users in Australia would take issue.
Another performance question that’s unanswered by the report is so simple even Senator Conroy should be able to understand it.
Conroy has spent two years telling Australia that the only important factor that would justify or inhibit a censorship regime was network performance. Why, then, did his trial fail to test network performance at any speed faster than 8 megabits per second? Since April, Conroy has been touting a National Broadband Network which runs at 100 megabits per second — over 12 times faster than the fastest speed tested by Enex Testlab. None of the trial’s technical specs, published in Appendix 1 of the report, approach that speed. Most of them actually show notable "flat-spots" between 7 and 8 megabits per second that are indicative of bottlenecks.
What is the nature of those bottlenecks? We don’t know because the report doesn’t tell us. So we’re now confronted with the prospect of a Government marching into a 100 megabit future — with absolutely no idea whether their mandatory censorship system will be able to keep up.
So — at best — the jury is still out on whether the clean feed inhibits performance and the explosive responses online yesterday suggest that the smart money isn’t with Conroy.
Let’s move to cyber-safety. Does this trial suggest that it’s possible to successfully filter internet content? Can the clean feed clean up the net or can the filter be circumvented? Again, conclusions are stymied by the lack of data in the report.
We can’t tell whether WebShield, the ISP with arguably the most practice at running censorship systems, was the one that could only block 8.1 per cent of users trying to worm their way through the filter — or whether they were at the other end of the spectrum and able to block 94.5 per cent of circumvention attempts. It is clear that the filter is far from impenetrable, doubly so if even the very best censoring ISP in the marketplace can be thwarted by anyone who’s clever enough to search for the method on Google.
There’s a bigger question still lingering unanswered, however, and it’s one that the Government has been disinclined to answer for the two years I’ve been asking it: What, exactly, are they hoping to achieve with the clean feed?
Sometimes they tell us that it’s to protect children from net nasties but the fact that the notorious ACMA blacklist only contains 1300-odd nasties — of which only half are nasties that have been refused classification (RC) — suggests that this is a marginal endeavour. If their position is that the trillion-URL-internet only hosts 600–700 blockable URLs, perhaps there’s a better solution than mandatory filtering? If the list of blockable URLs is so small, is there any point to the whole exercise? Could the Government use its law enforcement clout to achieve the same result without spending $44.5 million on censorboxes?
Sometimes the Department claims that the clean feed will prevent "accidental exposure" to RC content, whatever that means. But I’ve been using the internet virtually every day since 1989 and I’ve not only never stumbled upon RC content, I’ve never met anyone else who has done so.
And sometimes Conroy et al assert that we need net censorship to stop the distribution of child porn. If that is so, this report suggests that stopping the distribution of child porn is an unlikely outcome of the clean feed, given that every single one of the tested systems was able to be circumvented, and, more importantly, that none of them work for any protocol except HTTP.
While Senator Conroy has failed to explain in coherent terms why the Government wants the clean feed, there are plenty of reasons why it’s a terrible idea:
There’s no real problem to solve online.
Even if there was, I can’t see any serious public demand to solve it.
Even if there was public demand to solve the problem, Conroy’s report, incomplete as it is, indicates that a clean feed won’t do the job.
Even if the clean feed was a workable solution to a problem that doesn’t seem to exist, it will be both expensive and unreliable.
Even if the clean feed was a magically technically perfect system, it will be administered by public servants who have trouble spelling HTTP — let alone controlling the content it carries.
And finally, even if a functional net filter was scrupulously administered by experts, it’s highly likely the blacklist will leak again.
All of which gets us to the crux of the matter: the Government’s own report demonstrates that easily circumvented censorware is ineffective, and their own experience shows that they’re, well, completely hopeless at maintaining blacklist confidentiality.
After two years of back and forth on net censorship, it’s astonishing to find this is the best that Stephen Conroy has got to offer. Every time you hear someone utter the words "child pornography" in defence of the net filter, remember this: if Conroy seriously wanted to tackle the distribution of child pornography, there’s no way in hell he’d propose this policy.