Regulatory Failure: Australia's Internet Censorship Regime

Last Updated: 5 May 2001

"The law argues against Australian claims to being a technologically savvy nation"

-  Peter Coroneos, Executive Director, Australian Internet Industry
Association (IIA) as quoted in Questioning the Oz Net Censors,
Wired, 24 April 2001.

[Note: See also EFA's November 2002 analysis titled Hoodwinking the Public: Australia's Internet Censorship Regime.]

On 19th April 2001, the Australian Broadcasting Authority's Six-Month Report on Co-Regulatory Scheme for Internet Content Regulation July To December 2000 [1] was released by the Minister for Communications, Information Technology and The Arts, Senator Richard Alston.

The report overviews the second six months of the complaints-based Internet censorship regime which commenced on 1st January 2000. A total of 290 complaints were received by the Authority during the six-month period, making a total of 491 for the full year. This paper is a critical commentary on the report.

The ABA has the power to issue takedown notices to Internet Content Hosts in relation to material hosted in Australia that it determines are classified as R (material requiring an adult perspective), X (sexually explicit material) or RC (refused classification, the most common example being child pornography). Such material is referred to as prohibited content under the Broadcasting Services Act. However, R-rated material is permitted if protected by an approved Restricted Access scheme. Under the Act, the ABA, in conjunction with the Office of Film and Literature Classification (OFLC), is required to classify Internet content according to the classification scheme established for films and videos, rather than the less restrictive Publications guidelines.

The Scheme's Shrinking Relevance

Although 290 complaints were received during the six months, only 139 were determined to relate to prohibited content, and of these, only 6 were found to be hosted in Australia. This compares with 16 Australian sites from 201 complaints in the prior 6 months. The situation is illustrated in Figure 1:

There is little that the ABA can do about overseas content except refer it to filtering software manufacturers. In the case of "serious Internet content" (their description), the material may be referred to the Federal Police.

The ABA's direct regulatory powers however, only relate to Australian content, i.e. a total of 6 sites for the entire 6 months. Even this figure may be overstated. EFA is aware that at least one of the complaints related to newsgroup postings "hosted" on an Australian news server. A takedown notice was issued to an Australian ISP running the news server, even though the postings could well have originated overseas and were no doubt replicated on other Australian news servers. In fact, it is likely that almost all of the Australian complaints related to newsgroups, since the report indicates that 13% of complaints (i.e. approx. 37 complaints) related to Usenet newsgroups, and the ABA treats newsgroups as if they were Australian content [2].

The Exploding Statistics Phenomenon

The ABA has gone out of its way to portray the reported figures in a way that overstates the apparent effectiveness of the scheme.

On the ABA's Internet Content Complaint Form, it is clearly stated:

"Note: You can complain about only one item of Internet content in each complaint form. "

However, the ABA reported 67 "items" of prohibited content from the 6 Australian complaints. In contrast, 136 items of prohibited content were reported in respect of the 133 overseas sites. Figure 2 illustrates this phenomenon.

It could be concluded that the ABA is ramping up the "item count" in order to make the figures look respectable in relation to Australian content.

Of these 67 items, the ABA referred 45 items of Australian-hosted "serious Internet content" to the relevant State or Territory police service. However, given the questionable method of reporting large numbers of items from a small number of complaints, it could well be the case that all of these referrals relate to one or two sites (at least 14 of the 45 are likely to be newsgroup posts found in one newsgroup on one ISP's system).

No explanation is offered in the report for the strange discrepancy in the "items per complaint ratio" on Australian sites.

The ABA is similarly evasive when asked direct questions about statistics. At the Senate Estimates hearing in November 2000, the following question was asked [2]:

QUESTION NO. 42: Senator Harradine (Tabled) asked the following Question-on-Notice:
How many complaints in Internet content have been received? How many take down notices have been issued?
Answer: As at the 30 November 2000, 467 complaints had been received. Take-down notices had been issued in relation to 99 items of Australian-hosted content.

This answer can be mistakenly interpreted to mean that 99 of the 467 complaints received in 11 months resulted in takedown notices, when in fact only 22 complaints for the whole year related to Australian-hosted content.

Furthermore, the entire regulatory scheme is shrouded in secrecy. EFA has been attempting for over 12 months to obtain information under the Freedom of Information Act about the kind of sites removed, but without success. This FOI request is now proceeding to a hearing before the Administrative Appeals Tribunal in July 2001.

Discussion

In the Ministerial Media Release A safer Internet for all Australians announcing the release of the ABA report, Senator Alston said "Australian families will welcome the continued removal by the Online Content Regulatory Scheme of illegal and highly offensive material on the Internet, particularly child pornography sites."

In a newspaper article on the report, a spokesman for the Minister was reported as saying "The removal of these illegal and highly offensive sites means families feel comfortable about their children accessing the Internet. That will mean an increase in the uptake of new technology by families and that will help drive the growth of the new economy in Australia."

However, based on the figures reported, such claims must be seen as misleading political rhetoric that grossly overstates the impact of the scheme and lulls Australian parents into a false sense of security. The claims also play on the emotive impact of the child pornography bogieman, without any supporting information about prosecutions of Australians allegedly publishing such material.

The following table (a summarised version of Table 2 in the ABA report), shows the breakup of "items" by classification type:

Of the above, the ABA referred 105 items of "serious Internet content" hosted outside Australia to the Australian Federal Police, and 45 items within Australia to state police. No information is provided in the report about whether any person has been prosecuted in relation to these referrals.

It is a reasonable to assume that the items referred to police concern child pornography, particularly given that the ABA's breakdown of items by category supports this contention, e.g. takedown notices were issued in relation to 45 items that were categorised as either "exploitative/offensive depiction of a child" or "paedophile activity".

While EFA supports the prosecution of those who publish genuine child pornography, the definition used in the film guidelines is confusing. The X category guidelines do not permit the portrayal of "adult persons who look like they are under 18 years", which means that such material would be classified as RC, even though a minor may not have been involved. Furthermore, the definition of child abuse material refers to a "person who is or who looks like a child under 16". Such subjective and anomalous definitions mean that material could be defined as RC according to the guidelines, but may not justify police involvement because it is not actual child pornography.

Cost Effectiveness of the Scheme

The report indicates the ABA is spending over 95% of its effort on complaints about overseas sites that are then referred to filtering companies. This represents a government subsidy to a largely US-based industry that is probably already well ahead of the government anyway. EFA believes that important questions need to be raised about the outcomes and cost effectiveness of the regulatory system, especially as costs of the scheme are not revealed in the report.

For the financial year 1999/00, the ABA spent $915,000 on its Internet regulatory activities, which included only the first 6 months of the scheme's operation [3]. $100,000 of this was spent on classification fees charged by the OFLC. No detailed figures are provided in the ABA budget for 2000/01 [4], but it could reasonably be concluded that expenditure on Internet regulation for the calendar year 2000 was well in excess of $1 million, especially as the second half of the year saw a 40% increase in the number of completed investigations. In addition, NetAlert (the government's community advisory body on Internet content and usage) had a budget of $1m in 1999/00, increasing to $1.5m in 2000/01. This suggests a total cost in the region of $2.5m for the first full year of operation of the regulatory scheme.

The benefits derived from this expenditure are not obvious - a handful of newsgroup posts cancelled from Australian news servers, which is transitory material in any case, and perhaps one or two adult websites forced offshore. Given that the regime uses restrictive film censorship guidelines, it is quite possible that the censored website material is legally available in Australia in other media, e.g. magazines.

It is dubious whether any Australian website would host child pornography material, since Website ownership is too easily determined. This further reinforces the proposition that most of the "Australian" material was found in Usenet newsgroups, most likely originating from outside Australia.

NetAlert's activities are equally questionable in terms of outcome. The report indicates that its main activity has been promoting itself through widespread distribution of fridge magnets and mousemats, with no mention of any specific achievements.

In conclusion, the Australian government is spending about $2.5m per annum for no obvious outcome. Referring ephemeral newsgroup posts to law enforcement agencies is of dubious value. If the police are concerned about child abuse material in newsgroups they would be well aware of the most likely locations.

Furthermore, by encouraging Australian citizens to report child abuse material to the ABA, the system places complainants at risk of committing a criminal offence, since mere awareness of the existence of such material implies that it has been downloaded into the user's machine. There is nothing in the Broadcasting Services Act which grants immunity from liability for users who report illegal content. In fact, there is at least one known Australian case where an Internet user has been prosecuted in exactly these circumstances.

Media Confusion

Media stories about the ABA report suggest that journalists have either not bothered to read the report in detail, or have been misled by the confusing method of presentation of the statistics. Some examples follow.

More than a third of the sites determined to contain 'X' or 'refused classification' content were overseas.
....
The 45 Australian sites had all been removed, a spokesman for Communications Minister Senator Richard Alston said.
           NetAlert bans more porn. Australian IT, 19 April 2001

In fact, 133 such sites were overseas, from a total 139, i.e. 96%, not "a third".

The number of sites removed was 6, on which the ABA found 45 items which they referred to the police. In total 67 items were found to be prohibited content, and were the subject of takedown notices sent to the 6 sites. One would expect that all 67 should have been removed, unless some of them were on sites that relocated overseas on receipt of takedown notices.

The ABA issued 67 "takedown" notices to Internet service providers (ISPs) hosting prohibited sites in Australia.
           ABA blocks some of the offensive sites. Fairfax IT, 20 April 2001

As previously mentioned, takedown notices were sent in relation to 67 "items" found on 6 sites. EFA has evidence that at least one of these takedown notices contained a list of 15 items, so it appears likely that only 6 takedown notices were actually issued in total. Furthermore, takedown notices are issued to content hosts, which are not necessarily ISPs.

... the ABA issued 67 take-down notices for Internet content housed in Australia, two-thirds of which involved child pornography.
           Questioning the Oz Net Censors. Wired, 24 April 2001.

Again, the confusing presentation in the report has been misinterpreted.

... a spokesman for the Minister was reported as saying "The removal of these illegal and highly offensive sites means families feel comfortable about their children accessing the Internet. That will mean an increase in the uptake of new technology by families and that will help drive the growth of the new economy in Australia.
           Net surfers dob in porn sites. The Courier-Mail, 17 April 2001

In fact the Australian Bureau of Statistics figures for the quarter ending December 2000 show a drop of 7000 in the number of households with Internet access. ( Oz home Internet subscribers drop. ZDNet Australia, 1 May 2001).

This suggests one of two situations:
(a) the censorship scheme is having the reverse of the claimed effect,
     or
(b) there is no relationship between the activities of the Net censors and the rate of uptake of Internet access in the home.

The evidence strongly suggests that the latter is the most likely hypothesis. The claims of the Minister, the ABA, Netalert and DCITA about alleged successes and positive outcomes are nothing more than empty and misleading rhetoric.

A further media story about the report was:
     Australian child porn legislation questioned. ZDNet Australia, 20 April 2001

Conclusion

The Key findings from this analysis are:

• the statistical reporting of the ABA's investigations seems to be deliberately designed to confuse and mislead the casual reader.
• unlike classification schemes applied to other media, the government refuses to provide any details on what is being censored.
• The benefits of the regulatory regime are illusory, since almost all of the sites investigated are beyond Australian jurisdiction.
• Costs of the scheme cannot be justified in view of the minimal benefits obtained.
• The ABA seems to be taking on a role that is more appropriately left to law enforcement authorities, i.e. the tracking and prosecution of child pornography.
• users of the complaints regime are at risk of prosecution if they download and report illegal material.

In summary, the scheme is an expensive and unaccountable waste of taxpayers' money that achieves no useful outcomes in the context of the global Internet. EFA calls on the Parliament of Australia to repeal the unworkable legislation that spawned this Internet regulatory regime.

References

[1] Six-Month Report on Co-Regulatory Scheme for Internet Content Regulation July To December 2000 (released 19 April 2001)

[2] Senate Environment, Communications, Information Technology & the Arts Legislation Committee (PDF file) Answers to Questions on Notice, Australian Broadcasting Authority Supplementary Budget Estimates 2000-2001, (23, 30/11/00)
"This included content hosted in Usenet newsgroups, which is treated as content hosted in Australia if the complainant has accessed the content from their ISP's newsgroup server."
http://www.aph.gov.au/senate/committee/ecita_ctte/quest_answers/03aba.pdf

[3] ibid

[4] Australian Broadcasting Authority, Budget 2000/01
The costs of the Internet regulatory scheme are included in the overall amount of $7.4m allocated for "managed regulation of broadcasting".
http://www.aba.gov.au/about/information/

Author: Greg Taylor, vice-chair, Electronic Frontiers Australia
Date: 5th May 2001