May Contain Ponies
Since the last entry on the .xxx domain proposal, and subsequent discussions in another forum, I've been thinking a bit about how content rating should work.
To my mind a key requirement for technologies to rate content on the internet is that it should allow for multiple content classification criteria originating from many different, disparate sources. This is the biggest problem that I see with the .xxx domain: it requires that everyone everywhere agree what sort of content should be classified this way. Such agreement is clearly not likely in all but the most extreme corner cases.
In answer to the question posed in my last post, it seems that the W3C's preferred replacement for PICS is ... RDF. No, not not that RDF, although there probably is a bit of the second kind of RDF going on around the first kind of RDF.
RDF is flexible enough to describe all kinds of metadata, not just content ratings. And when describing content ratings for a site, I believe it can rate against multiple classification schemes.
By way of experiment I obtained a RDF-based content rating for this site using the Family Online Safety Institute classification scheme. They have an online tool to generate the rating and it is pretty straightforward; just fill in a form and they email you the RDF file and instructions on how to add it to the site. You can go to the label tester to see the results.
The labelling guidelines are well described, with the possible exception of the language ratings. These are sufficiently vague that I ended up simply picking the middle-of-the-road rating in lieu of the missing "pretty clean but occasional f-bombs" rating.
But the point is that however imperfect this rating system is, it does at least allow co-existence with others. So I could also use a hypothetical pony-based rating system which classifies sites in terms of exposure to ponies, and rate my site using both FOSI and PONI schemes.
To pick a possibly more useful hypothetical rating system: how about a rating system to declare content that contains images and recordings of deceased persons? I understand that such content is distressing to some Aboriginal people. There are probably many other useful rating systems not yet in existence.
The only missing piece of the puzzle is how to rate other people's content. The problem of finding a third-party rating for a given site is one that I haven't seen a solution for. Similarly the ability to attach a rating to a hyperlink (for example, replacing the ubiquitous "[NSFW]" tag) would be useful, and I'm not sure how this would work.
Brendan raised the point that content labelling should be voluntary and I obviously agree with this. It's a logical extension of the provision that multiple rating systems should be allowed. I would probably go one step further and say that it should be possible for others to label your content, whether or not you choose to do it yourself.
There are lots of security issues relating to this — and RFC 3675 addresses most of them — but I don't think these are insurmountable. The question of which rating is authoritative for a given site is very similar to the question of which site is appropriate for a given search term. Search engines seem to have sorted out this problem to a reasonable level of approximation.
The ability to aggregate multiple ratings for a given site into some coherent whole seems like a key requirement for any technology that allows distributed rating. When a single agency — either the content author themselves, or a third-party — is solely responsible for rating content, there is the potential for abuse or mistake. This can only be mitigated effectively by allowing multiple independent agencies to rate content. In fact this is a general metadata problem, and I'd be interested to see how this will be addressed.