Report of the Advisory Board

January 2000

Below is the report of Internet Content Rating Association’s Advisory Board report to the ICRA Board of Directors. It is the culmination of several months’ intense work by thirteen international experts. This report was recently presented by Nigel Williams, Chairman of the Advisory Board, to the ICRA Board and formed a central part of our discussions and decisions on the future of the labelling and filtering system. While the report was well received and praised for its breadth and scope, the ICRA Board of Directors are not bound by its findings, nor do they endorse all of its recommendations. The final version of the revised ICRA labelling system is planned for release at the end of September 2000.

Stephen Balkam
Executive Director, ICRA

1. Terms of Reference

In January 2000 the ICRA Advisory Board was constituted at the request of the ICRA Board with the following purpose:

To advise the ICRA Board on the structure, content and technical specifications of a revised rating and filtering system.

To review recommendations from the Rating System Working Group from the perspectives of the constituencies of members geographic, sectoral and professional interest areas.

To incorporate views of the ICRA Reference Group and other individuals and organisations with an interest in the development of the rating and filtering system.

The full terms of reference are appended as Annex 1.

2. Membership

Invitations to join the Advisory Board were accepted by 12 individuals from five continents with much relevant experience in a range of fields – computer science, media law, education, free speech advocacy, child welfare and children’s use of the Internet, media content regulation, and broadcasting. A number of members came with specific experience of looking at Internet content issues and labelling, rating and filtering. In particular, three had previously served as members of the RSACi Board, two had been members of the Steering Group for the European INCORE project and three had been members of the Bertelsmann Foundation Internet Content Experts group.

Nigel Williams, Director of Childnet International, was elected by the group at its first meeting to chair the Advisory Board¹. Three ICRA staff took a lead role in providing the Board with excellent support – Stephen Balkam, Ken Handley and most recently Germaine de Haan (who also took minutes of the face to face meeting in The Hague).

The AB was entirely independent of any company or commercial interest. All time was given voluntarily and generously by members, with expenses being paid for the telephone calls and the meeting in the Hague.

3. Approach to task

The AB met on six occasions. We had one full day face to face conference at the offices of the Film Classification Board in The Hague, Netherlands on 24 March. which was expertly organised by Cornelius Crans and his staff. The AB also had five telephone conferences, many at inconvenient times for a number of the members because of time differences. There was also a very lively exchange of views between meetings on the AB mailing list with members contributing quite detailed papers and reflections. While the AB did reach out to a number of interest groups in the course of our work, there was insufficient time to and budget to engage in wider consultation. In particular there has been no specific contact with the ICRA Reference Group and we would encourage the ICRA Board to consider how they wish to involve this wider group of people interested in the development of the system.

4. Structure of Report

The remainder of this report is split into three sections-

some important overarching preliminary comments
recommendations about the structure and design of the proposed labelling and filtering system
recommendations about a number of important issues that are likely to impact on the perception of and success of the system

5. Overarching comments

The AB was very conscious that it was being given an extremely ambitious remit given the limited time and the geographical spread of the participants. While many have discussed in an abstract how a rating/labelling² system might work online, the AB was now being asked to help design it. The AB saw itself as sitting at the intermediate stage between academic discussion of concepts and the detailed brief to software designers. We recognised early on that we could not hope to produce very detailed definitions for each element of the system – rather we could comment on the value of the main components.

We were driven by two overriding concerns –

the importance of protecting freedom of speech online and avoiding government censorship; and
the need to offer effective tools to parents to protect their children from content they as parents considered undesirable.

The AB acknowledged that the generally accepted international definition of a child was a person under the age of 18. We accepted the new ICRA labelling and filtering system could be used for children of any age and, indeed adults. However, in our view the primary target market would be protecting children under the age of 12 at home and in other locations where they might have Internet access, particularly schools and libraries, where research has indicated parents value the use of filtering tools because they cannot be there to supervise their own children.

We agreed that the system would work best when it was voluntary in its use (by both content labellers and by parents) rather than imposed by Governments or other agencies.

We were very challenged during our deliberations by two fundamental questions posed by Jean Armour Polly – Why did RSAC not attract a much higher proportion of content publishers to label their sites? and why did more users not avail of this free parental control system? While a number of possible answers were suggested to these questions, the key result of the discussion was that we agreed ICRA had to focus on meeting the needs of its core markets to succeed. The system had to be simple for content publishers to label, and be simple for parents to set up and use, while meeting their key concerns about exposure to inappropriate content. Furthermore, the new system’s availability had to be strongly promoted and aggressively marketed by ICRA in order to ensure more widespread utilization by both publishers and parents.³

We recognised that these objectives could conflict. Parents wanted both a comprehensive system in terms of types of concerns included but also a system that was simple to use. Comprehensiveness could lead to a long questionnaire for content publishers that might just turn them off from using the system at all. Conversely, added features could make the system more attractive to users, and thus be of greater interest to more publishers because of the bigger user base. This conundrum lies at the heart of the design process for the new ICRA system.

As an AB we crystallised this issue in a question “if we recommend including this feature or approach, will the added complexity be more than made up for by the additional functionality and attractiveness of the system?” .

Fundamental to the success of our approach will be the ability of the new ICRA system to import third party lists – at their simplest these might be “YES” lists of sites considered good for kids; and “NO” lists of sites not deemed suitable for kids. Indeed, we would argue that ICRA should go further and for some issues where convergent definitions are difficult (e.g. racist or intolerant content) ICRA should actively engage in outreach activities to encourage the production of such lists and have these available to be used when the system goes live.

Although the RSAC system may have under achieved, there are in excess of 100,000 sites with an RSAC label. Thus the AB was concerned that inasmuch as possible the new system should be “backward compatible” with the existing RSAC system. In particular, that the new system when installed in the browser would still read the existing RSAC meta tags.

During our discussions it became clear that the heart of the system was the questionnaire which had to be filled in by content providers. If you did not collect the right kind of data from content providers then no user interface could make up for the deficit. We thus focused our efforts on what should be included in this questionnaire or matrix (see Section 6 below). We have also thought carefully about how we could make the questionnaire as easy as possible for content publishers to use. We have therefore advocated the inclusion of two shortcut response buttons for “adult” content and “no problematic content for kids”⁴.

But during our work we also saw that a range of other issues would be very important to the ultimate success of the system. Although we did not have time to consider these broader issues in detail, we have highlighted a number of concerns which we hope the ICRA board will consider further – these are listed in Section 7 below.

6. The Content Questionnaire

We recommend including the four categories used in the RSAC system but advocate doing this through a set of objective descriptors rather than a scalar 0>4 set of levels. The existing scalar system suggests equivalencies between different types of activity that seem unjustifiable (e.g. explicit sex and sex crimes).
Some members argued for reducing the granularity of the definitions (i.e. having fewer descriptors for each category). Others suggested this would create false equivalencies like those present in the existing scalar system. The group agreed it should recommend to those producing the detailed definitions to focus on having as few definitions as will enable clear distinctions between content likely to be of concern to parents of young children
Nudity – We recommend more work to clarify certain definitions (e.g. revealing attire).
Sex – We recommend reviewing the number of definitions and their clarity.
Violence – We recommend excluding sports violence. This should be dealt with through concerned individuals and groups producing third party “red/no” lists.
Language – We recommend focussing this on swear words – hate and racist speech may be better dealt with through third party “red/no” lists.
Harmful activities category – The AB considered carefully the need for a category /set of definitions relating to activities that might be potentially harmful for children. Topics discussed under this heading included: alcohol, drugs, tobacco, cruelty to animals, racist/intolerant material, religious cults, information on how to commit suicide, production of fireworks, bombs or dangerous chemicals, and others. The AB concluded that all of these issues are of concern to at least some parents some of the time, and as such are worthy of careful thought about how to handle. We noted that our two members who represented video and film classification bodies put high priority in seeking to evaluate these kinds of issues in their work. Moreover in some cultures (eg Europe and Canada) issues like intolerance and racism are primary concerns of parents.
It is very difficult to include new categories of descriptors later in the process so very careful thought should be given before rejecting topics completely. However, we could not reach consensus on definitions for any of these areas that would be convergent to the majority of content producers. We therefore recommend, that unless ICRA can produce truly convergent definitions, these issues are better dealt with by third party “yes/no” lists integrated into the system .
Personal information and Privacy– There was broad agreement on the AB that web content which required children to divulge personal information was of real concern to parents, and could raise significant safety issues. However, the AB members were concerned that privacy experts in a number of countries had already spent much time attempting to deal with the issue of information being sought from children through web site forms. While we were attracted in principle towards seeking to cover this area, we decided in practice to leave it existing efforts like the Children Online Privacy Protection Act (COPPA) in the USA.
Chat and Viewing of Personal Details by others– The AB felt that a concern currently missing from the primary focus of privacy legislation was interactive content, i.e. content that encourages or facilitates the publication by children of their own personal details in a public space, (e.g. discussion boards). It was further noted that in many countries, parents are very concerned about dangers to children in chat rooms and that a number of significant cases of exploitation have occurred.⁵ We have used IRC/ICQ chat as an example to test whether definitions could be produced and are satisfied that they can, but believe careful thought needs to be given to the potential of other kinds of interaction between children (e.g. video, or audio).
We recommend including a set of definitions which allow distinctions between chat rooms that are moderated and unmoderated; the type and degree of moderation; those that are intended for children under 12, and those open to any age with allowance for the type of age verification. The aim would be to have a cluster of no more than five questions on this topic which would elicit the key features of any chat that was offered. We also advocate designing the system so that content providers can (and are encouraged to) label the specific page where chat access is provided so that other pages without chat might be available to the user.
Financial transactions – We recognised that there was some concern among parents about the opportunity for children to purchase online. We could not reach consensus on whether the simple existence of the ability to purchase products or services should be the subject of a separate meta tag. Some members felt that while some parents were concerned their children might spend money online, since this could only happen with some offline parenting intervention (e.g. through providing a credit card or e-cash of some kind) it was unnecessary to include it in the system. Other members felt that since the area of buying online would increase in importance, a label which indicated on a purchase page how the purchase could happen (through e-cash, credit card, or by other mechanism) would add value to the system and should be investigated further.
Context – The AB was of the view that there was considerable value in allowing content providers to indicate that the labelled content, whatever other meta tags it might attract from the ICRA system, was actually being provided for purposes of art, education, or medical information. Providing for such “redeeming context” would overcome objections that some material if labelled literally according to the ICRA definitions would be inaccessible to parents even though many would consider it beneficial to children (e.g. sex and violence in the Bible; nudes in works of art). However, it was very difficult to provide convergent objective labels for context – how do you define art??
Thus we concluded that it was best to allow content providers themselves the opportunity to claim context for their content and for parents and template producers to decide if they generally wished to believe such claims. A suitable wording for such a claim might be “The author/publisher of this site warrants that this site/page contains material that is educational (or artistic value or medical information) AND is suitable for children under the age of 12”. Adopting this approach does reinforce the need for careful monitoring to ensure accurate labelling and context claims.⁶
One button “adults only” – We have been concerned throughout our work about making the questionnaire as easy and as quick as possible for content providers to use. One idea which emerged during our discussions was the possibility of a single button for content providers to click indicating their site was intended for adults only. While this clearly meets the aim of encouraging easy labelling the major difficulty arises that providers will interpret this opportunity in different ways. It will also lead to a mixture of detailed data for some web sites and just a broad label for other sites. The AB then considered whether this one button approach might be a kind of “input template” (i.e. it would automatically fill in the answers to all the questions in the questionnaire).
We agreed that the one button for adult sites should be included but be entirely transparent (i.e. it would fill in the questionnaire in a specific way but the content provider would be asked to confirm the choice of operators given). Furthermore, we recommend consulting with the “adult sites” industry on this proposal⁷ with the possibility that they might generate a form of input template⁸ to their members.
One button “no problematic content for children”– Some on the AB felt that there was also a case for trying to make it easier for those sites whose content was likely to be labelled as having no content of particular concern to parents. We saw the advantage of creating a method of simplifying input for many, many sites. We toyed with the idea of defining this button as “suitable for children”. This could have the advantage of becoming a badge of distinction for children’s sites. However, on reflection we decided that it was extremely difficult for a self-labelling system to guarantee the answers to appropriate questions to ensure we could stand over the quality of sites for children. Thus we felt this task was much better dealt with by third party “yes” or “green” lists of children’s sites. On balance therefore we support the inclusion of a one button “no problematic content for children” in the same transparent manner as with the one button “adult content” approach above. ICRA needs to reach out to those producing high-quality child friendly material (especially educational institutions) and encourage them to label their content using this approach, or develop their own “input template” just as is proposed for the adult sites industry.

7 Other Issues of Concern

In the course of our deliberations, we touched on a number of other significant issues of concern as to how the ICRA system will work:

Marketing of the system –We were concerned that part of the failure of RSAC was the lack of continuous marketing of the system. ICRA will need very effective marketing to ensure it is broadly used. Such marketing must be addressed at both content producers and users.
Monitoring of the accuracy of labels –We were concerned that there was potential for misleading or inaccurate labelling of sites and would encourage ICRA to look carefully at mechanisms for independently auditing the accuracy of the system, and sanctions for mislabelling.
Language –We noted that the current priority languages for the implementation of the ICRA system did not include Chinese. As this language probably has the second fastest growth of Internet users after English we questioned whether it should not be included at this stage. This would also be a clear signal of ICRA’s intent not to be anglo-centric.
User interface –We have not had time to focus on the design of the user interface. But we are very concerned that:
- It should be as easy to use as possible
- It should be as prominent in the browser as possible. A key concern about users experience of RSAC is the extent to which it is buried in the browser settings. Some members noted the successful uptake of AOL parental controls arising from the constant encouragement on start up to use the settings and a one click link to setting them up
- The facility for users to add their own favourite sites or download appropriate “yes” and “no” lists should be available from launch date to overcome the problem of the preponderance of unlabelled sites.
- Just as the AB has advocated the importance of having “one button” easy to use solutions to encourage rating of particular types of sites, so we also believe that offering parents the opportunity to choose a pre-configured “maximum safety” button would be a good idea. This would be a form of template⁹
Relationship with other labelling efforts– We had the opportunity through AB member Bruce Rigby to be acquainted in outline with a number of efforts to label content through meta tags, in particular the Dublin Core Metadata Initiative (DCMI) and Instructional Management Systems (IMS)¹⁰. While we had hoped to pursue contact with the groups behind these initiatives, this was not possible in the time available. We would however strongly urge collaboration between ICRA and these initiatives. There could be opportunities to link and thus encourage those labelling with one system to label with others; to examine how the systems can be made compatible for users; and to generally increase the profile of content labelling among producers.
Development of templates –We do see the value of developing user templates with preconfigured filtering settings. However, we would caution that if this approach is to work, there need to be at least two downloadable options available from the outset. ICRA may therefore need to make it very easy for organisations to produce templates, and for users to install them.
Field testing –We believe it is crucial that the ICRA system gains wide acceptance quickly. This is much more likely if the system is launched after extensive field testing with users in relation to both its content and design. AB members could point to experience in the offline world in relation to new labelling/rating systems which support this recommendation.¹¹
How labels are generated –ICRA should review the process used by RSAC of a central server generating all the labels. As the nature of content on the internet changes (e.g. the trend from static web pages to databases of reconfigurable content) other forms of distributed registration may be helpful.
An ICRA search engine? –We wondered if ICRA succeeded in labelling a substantial number of sites it might create a search engine/directory of labelled sites or partner with an existing engine(s). Indeed, a partnership with an existing engine in their sign up procedures for sites might help produce critical mass.
Implications of new platforms, technologies and convergence –We were very concerned that the new system should not be specific to one delivery platform and should aim to work with new wireless technologies, digital television, and any other developments that emerge in coming months and years. Specifically, ICRA should monitor carefully what platforms the adult industries are migrating to that are open to use by children. This would be a good indicator of the need for an ICRA-type system on that platform. ICRA should also monitor the opportunities for convergence with existing media – it may well be that the system could transfer and be taken up by television and video in a number of markets.

8. Conclusion

The Advisory Board is grateful for the opportunity to input to the exciting development of the new ICRA labelling and filtering system. We recognise that it will be a very challenging task to launch the system successfully across the world and encourage rapid uptake. A number of members have indicated their willingness to advise in more detail on specific issues as they emerge. The group as a whole would welcome feedback on this report and would encourage the ICRA Board to publish it in a public area of the ICRA site.

Notes

1 For brevity and convenience “Advisory Board” has been abbreviated to “AB” throughout the rest of this report.

2 Although the words rating and labelling have become almost synonymous in some countries (especially USA) we determined that it was much better to talk about labelling of content as this did not imply any evaluation of the desirability or worthiness of the content.

3 See also Section 7(a) below.

4 See paras 6(l) and (m) below.

5 Most recently documented in the report (8 June 2000) for the National Center for Missing and Exploited Children by the Crimes against Children Research Center “Online Victimization – a Report on the Nation’s Youth” which showed that one in five youths aged 10 to 17 had received a sexual solicitation in the last year. Report available at www.missingkids.com

6 See para 7 (b) below

7 One of our members did engage in private conversations as an individual with the adult industry to gauge their interest in co-operating with this kind of effort. There was a very enthusiastic response to co-operating with this kind of concept and other ideas like a “labelling bot” to propose a preconfigured questionnaire to adult sites.

8 The concept of input templates has quite wide applications beyond material specifically aimed at adults. For example, a health service could produce a template to enable it to easily publish many pages which had some nudity in a medical context.

9 See para (f) below

10 For further details see http://purl.oclc.org/dc/and http://www.imsproject.org/metadata/

11 For example, the Classification System for Canadian Television Programming was introduced to broad public acceptance following extensive field testing; the US TV parental Guidelines Rating System had a very controversial launch following little field testing.

ANNEX 1

ICRA Advisory Board

Terms of Reference

Purpose

To advise the ICRA Board on the structure, content and technical specifications of a revised rating and filtering system.

To review recommendations from the Rating System Working Group from the perspectives of the constituencies of members geographic, sectoral and professional interest areas.

To incorporate views of the ICRA Reference Group and other individuals and organisations with an interest in the development of the rating and filtering system.

Membership

The membership of the Advisory Board shall be limited to fifteen members appointed by the ICRA Board of Directors. The members shall be from non-industry organisations and represent different geographical, sectoral and professional interests. Membership is for one year from appointment, renewable.

Chairing

The Advisory Board will, at its first meeting, select a chairman from amongst its membership.

Support

The Advisory Board will be supported by the ICRA staff.

Decision Making Powers

The working group has no formal decision making powers and is advisory in nature.

Timescale

The Advisory Board will meet to a timescale specified for its task and linked to other ICRA development processes, particularly the plan for the development of the system and the decision making schedule applying to the Board and the Rating System Working Group. It will hold at least one face-to-face meeting per year.