The consequences of filtering

The Government is proposing a system to filter the Internet. The proposal is a complaints-based system that would block specific URLs only after they have been brought to the government's attention and also deemed to be Refused Classification. It would operate using a secret list and make it mandatory for all ISPs to block sites on the list for all Internet users in Australia.

“Refused Classification” and “illegal” are not the same thing, and the problems of having a mandatory, secret blacklist should be readily apparent. The consequences to the wider Australian society if this filter proposal succeeds do deserve discussing. This proposal doesn't have a single redeeming feature that will make Australia any better than it currently is.

There are also obvious known drawbacks to the proposal. The first is the cost of the system. Filtering products are expensive to purchase and run. Smaller ISPs will not be able to easily carry the cost, and will be forced to immediately pass them on to customers. Larger ISPs, with more capacity to absorb the cost, will be able to strongly compete on price for a short time. This will force competition out of the market. It will mean even more expensive Internet for Australians, especially as the government intends for ISPs to carry the entire cost of the proposal. Access to the Internet in Australia is already among the most expensive and inadequate in the developed world.

While the technological issues are minor compared to the political and civil liberties issues, they should be enough themselves to scuttle this proposal. The Internet is currently running on a backbone of IPv4. Every computer connected to the Internet has a unique number assigned to it. IPv4 only allows for 4 billion computers to be connected simultaneously, and the pool of available numbers is rapidly diminishing. IPv6 is already starting to take over, and within two years almost every ISP will be forced to support IPv6. The trouble is that most current filtering systems are incapable of supporting IPv6. The proposal will saddle ISPs with expensive, inadequate equipment.

The next technological issue is latency. The government is making big claims about speed, but it is confusing itself on the difference between speed and latency. While it may be true that speed decrease for individual website loads is negligible, the compounded effects have not been measured. This is particularly important in high bandwidth applications, such as video games, film distribution and teleconferencing. Minor reductions in speed have a cumulative effect, and this could dramatically affect performance in such applications.

This issue is particularly important given the government's proposed National Broadband Network. The maximum speed tested in the Enex trial was 8 Mbps. The NBN's speed is expected to be 100 Mbps. There is little reason to think that the trial results are applicable to a higher speed network. Additionally, at least one participant in the Enex trial reported they had less than 15 households in the trial. Telstra didn't even bother to test using any real people. There is no reason to think that such a small sample is representative of the result of filtering the millions of connections across Australia.

High traffic sites cannot be blacklisted as there will not be the capability to deal with the huge number of requests for sites like YouTube. There is a lot of material on YouTube that would be considered Refused Classification because the rest of the world doesn't attempt to ban such material. As recent events have shown, Google is not keen on the idea of removing such material. This means that only sites that have softer voices will be blacklisted. This is inherently discriminatory.

There is also the Streisand Effect to consider. On the Internet, attempting to censor something inherently focuses attention on it. A site that is considered low traffic at the time it is added to the blacklist may quickly become a high traffic site when its presence there is discovered. Slashdotting is a similar effect. This is when an extremely popular site links to a far less popular site. The large number of people who then visit the low traffic site often bring it down by their volume. Both effects could significantly degrade Australian Internet performance as low traffic sites become high traffic to the point the filtering equipment may not be able to cope.

There is no real possibility that the blacklist will not eventually be leaked again. Mandatory censorship of this scale will motivate large numbers of the tech-savvy to reverse engineer the list. It is relatively easy to do. If the list truly is the “worst of the worst”, and its security can't be assured, its publicly availability is unconscionable. If the list is as poor quality as the current ACMA blacklist, then leaks will do little more than be a continuous source of embarrassment to the government as the edge cases are brought to light by the media. A more disturbing thought is that the list is reconstructed but not made public because those who reverse engineered it were seeking out the illegal material which is also encompassed by Refused Classification.

The government has not provided any figures on how much of the total Refused Classification material on the Internet it expects to block. This is an important consideration. If the government only ever expects to cover a small percentage, then there seems little point in even having the policy. The web contains more than 1 trillion unique URLs. The current ACMA blacklist contains less than 1,500 URLs. And the government has admitted that performance will degrade if the list becomes too long.

The unintended social issues this proposal will create are deeply concerning. The filter is being sold as a way to protect people, particularly children, from accidental exposure to offensive material. There is no evidence of impending societal collapse from the decades at the current level of exposure. The reality is that a list of a few thousand will have a negligible preventative effect. It will, however, mean that parents will become complacent and oblivious to online dangers. If this proposal succeeds, the government will claim that the Internet is now “safer” in its advertising. That will be the sole message mums and dads will take away from the entire debate. This will mean less parental supervision, and will actually magnify our children's exposure to age inappropriate material. It will cause the very problem it attempts to solve.

The ACMA has conducted numerous studies over the years that show that Australian parents are well aware of online dangers and are happily managing their children's access. There are currently ISPs in Australia that specialise in providing filtered access, like WebShield. These companies have a filtered Internet service as a core component of their business model, rather than the cheap bolt-on after thoughts that reluctant ISPs will implement. Such companies do have low take up, despite the massive amount of free advertising the government has just provided. The previous government's NetAlert scheme similarly had low take up. It is clear that Australians understand the risks and deliberately choose to have unfiltered Internet. It is laudable to try and make their job easier, but this proposal will not do that.

Then there is the matter of privacy. Authorities require extenuating circumstances to eavesdrop on the telephone conversations of Australians. Intercepting private mail also requires a good reason vouched for by the judiciary. What the government is proposing, however, is a system where every packet of information sent by Australians is opened and inspected, before deciding if it should be allowed to reach its destination. It is not enough to say that those with nothing to hide have nothing to fear. Most people close the door when they go to toilet. While they aren't doing anything illegal there, that does not mean they particularly want to be observed doing so. Privacy is, and must be a fundamental right in a civil society.

There is also the future reaction of users when they encounter material which is potentially Refused Classification. If it isn't already blocked, are they to assume that the material is sanctioned by the government? Sexual explicit material, which isn't, and never could be covered by any blacklist is not Refused Classification. This is clearly material that is age inappropriate for children. Also, as a complaints-based system, the majority of complaints are going to be registered by our nation's biggest wowsers. Are we content to have what we see decided by the most easily offended? Such people will also be the loudest voices to keep increasing the scope of what is blocked.

Even more bizarrely, the government is proposing outsourcing censorship decisions to foreign countries. Incorporating foreign lists into the Australian blacklist without reviewing the content is handing another country the power to decide what Australians can and cannot see. Foreign lists have had their own embarrassments. The British Internet Watch Foundation, for example, has blacklisted material our Prime Minister might consider “absolutely revolting”, but wouldn't bat an eye at the Classification Board. Properly incorporating foreign lists will turn into an expensive exercise. It already costs over $500 per URL for review by the Classification Board.

If the filter proposal passes, Australia would be the first and only Western democracy to introduce such broad and far reaching censorship on the Internet. Our actions thus far have already given comfort to more authoritarian regimes. Our filter model will end up justifying limiting freedom of speech among the most despicable governments. It would blacken our reputation, weaken trade, discourage discerning would be migrants and constrain our moral authority. Respected international human rights organisations are already expressing concern about Australia over this proposal.

There is a saying: “The Internet treats censorship as damage and routes around it.” The whole concept of the Internet was to have redundant means of communication should any number of nodes be destroyed. Circumvention is thus not only possible but also exceptionally easy. Both the Enex and Telstra reports made this very clear. Even WebShield was unable to prevent circumvention and this is their core business. Thus, the filter proposal will do nothing to stop those who want illegal material from seeking it out.

The unavoidable conclusion is that the government's Internet filter proposal will be no more than an expensive waste of resources, delivering no benefits and having little or no value to show for it. It will magnify the social problems it claims to be attempting to solve, and destroy Australia's reputation into the bargain. It is well past time this idea was killed off once and for all.