In September 2002, a joint report published in London by The Independent Television Commission, The British Broadcasting Corporation and the Broadcasting Standards Commission made several references to ICRA. These comments articulate well the substantive criticisms made about the ICRA system. Below is the text of the reply sent to the ITC by ICRA’s CEO, Stephen Balkam. Whilst accepting the points made, this letter sets out the many positive actions ICRA has already taken and is taking to protect children from potentially harmful material whilst preserving free speech.
To see the full report, please click here. (Document is in PDF format)
Background
The [UK Government’s] Communications White Paper (2001) states that [ratings and filtering] mechanisms should be investigated, and makes special mention of the ratings and filtering systems currently being developed for the internet.
We, of course, welcomed the reference in the White Paper to the investigation of rating and filtering systems of the internet. Many other governmental papers and reports have also cited rating and filtering as a means to protect children online. The recent Home Office Task Force Good Practice Models makes specific reference to ICRA in its recommendations. The EU has also recognized and funded the concept of voluntary self-labelling and parental filtering based on these labels. Organizations as diverse as the International Chamber of Commerce, the Global Business Dialogue on E-Commerce, ICSTIS and the interactive Gambling, Gaming and Betting Associations have all pronounced their support for ICRA.
Internet – current scenario
1. Parental control mechanisms such as ICRA were not at all well known among the sample, and while there was a muted undertone of pro-libertarian dissent (‘censorship’), on the whole participants approved of the principle that there was at least one group looking in an organized way at this issue.
Addressing the dissenters for a moment, we have argued the case of choice not censorship successfully before. Prominent US First Amendment lawyer Bob Corn Revere and UK lawyer Mark Matthews have both supported the ICRA approach as being respectful of free speech rights. The Internet Content Rating Association doesn’t rate sites, content providers do that using the ICRA system. Naturally we welcome the support offered to our approach expressed by the majority of participants.
2. … while the principle of putting control of levels of suitability in parents’ hands was appreciated by some, from the outset many participants were concerned about their complexity and have grave doubts about feasibility.
Naturally we welcome the support for ICRA’s principles and what may at first seem surprising is that we agree wholeheartedly that selection of each individual descriptor is way beyond what most parents will want to do. That’s why we developed the concept of “templates.” These are small text files that set all the parameters of an ICRA-compatible filter and do so in much more sophisticated ways than even the most complex user interface could achieve. Furthermore, filters can be made available with templates pre-installed, a real plug-and-play solution – so long as the parent agrees with the values espoused by the template creator.
ICRA continues to encourage organizations with an interest in child welfare and other issues to create such templates.
3. In addition, the credibility of any such mechanism was undermined if it involved sites rating themselves rather than being rated indecently. This was exacerbated if (as with ICRA) there was no obvious channel available to complain about mis-rated sites and ask them to be reviewed.
This issue is one that we are addressing with a new verification system due online within the first quarter of 2003. An Artificial Intelligence agent will analyze labelled sites and compare them with the label. If the AI unit determines that the label is probably inaccurate, human checks will be made. In our experience, deliberate mis-labelling is almost unheard of; nevertheless, it is an issue raised repeatedly by our critics. A range of systems will be put in place to encourage mis-labelled sites to relabel, but as an ultimate sanction, ICRA is likely to publish the URL of any such site, perhaps within a downloadable block list for use in ICRA-compatible filters.
4. … But any remaining appeal generally crumbled when information about the low proportion of rated sites and the implications of either allowing or not allowing unrated sites were discussed with participants. They felt that this seriously undermined the value of the system to the extent that the effort involved in setting it up would not be justified.
We agree 100%. That’s why the opening 2 sentences on the ICRAfilter download page say:
This filter has been developed to demonstrate a transparent, freely available tool to enable parents to control access to material they deem inappropriate or potentially harmful to their children and at the same time protect freedom of expression. It is designed primarily to demonstrate filtering based on PICS and other techniques and is not, in itself, a complete consumer product.
This statement has been present on the filter page since it was launched. ICRA has never claimed that the filter is anything other than a tool to demonstrate how label-based filtering could work. Neither have we claimed that the filter, by itself, is a fully functioning parental control system. Such a system is on the way, however, and it’s called SIFT.
SIFT, the Solution for Combined Internet Filtering, is a joint project between ICRA, Spanish filtering company Optenet and the Greek National Centre for Scientific Research. Co-funded by the EU under its Internet Action Plan, Sift combines filters based on ICRA labels, lists, as developed and maintained by Optenet, and Artificial Intelligence analysis as developed by NCSR. An ICRAfilter-like module acts as chairman, taking “votes” from the various modules as to whether a site should be allowed or blocked. Such a system makes full use of ICRA labelling whilst presenting users with a simple interface, but with further, detailed options available for those that want them. Other filters of any kind may be incorporated and used within the one platform. Beta testing is scheduled to begin in February/March, with public release in June.
ICRA remains committed to the principles of self-labelling as the most open and democratic way of describing digital content. It enjoys the support of governments, policy makers and, crucially, parents. It is the only system that supports freedom of expression as much as it supports parental choice and, as a consequence, enjoys significant support from the adult industry, the online gaming industry and similar sectors.
Technically, self-labelling is the only method by which content providers can communicate with filters directly. Artificial Intelligence agents are becoming more prevalent as the technology improves; however, the number of possible classifications an AI unit can produce is limited. Likewise human classification can only assign content to a limited number of categories, often not distinguishing between one part of a site and another. ICRA labels offer 35 million million potential classifications, are machine-readable and can inform the analysis process, working with AI and lists as shown in Sift, to produce a robust system with high granularity and very few false positives.
In the near future, as much of the totality of human experience and knowledge that can be digitized will be. Parents have a right to decide what types of material their children have access to, and individuals have a right to express themselves through digital means. Only self-labelling, with its unmatched granularity and transparency, verified within a margin of probability by AI, aids both of these dual aims.
Stephen BalkamCEO
January 2003