It's time to go beyond the freedom-of-speech-versus-protection-of-children debate. The larger issue is sharing: how can diverse user populations share a single communications medium without constantly being at each other's mercy ? If we insist on the Net as a global free-for-all, some people will opt out and others will try to impose their notion of order on everyone else.
Instead, we need to provide a variety of lenses that magnify parts of the Net while covering other parts in a rosy haze - shielding people from the disorder just beyond their view. Content labelling can provide those lenses: software can read labels and automatically do useful things like highlighting documents labelled one way or blocking access to those labelled another way.
In the past year, MIT's World Wide Web Consortium developed a new piece of Internet infrastructure called PICS, the Platform for Internet Content Selection ( w3.org/PICS). The original goal was to empower parents and teachers to provide kid-safe lenses. As a bonus, however, PICS provides a general labelling infrastructure that is now available for all kinds of uses.
PICS establishes conventions for label formats and distribution methods without dictating a labelling vocabulary or determining who should pay attention to which labels. It is analogous to specifying where on a package a label should appear and in what font it should be printed without specifying what it should say. Once vocabularies and labels are expressed in PICS format, selection software can interpret them and do useful things such as blocking access or verifying the page's integrity. Any independent organisation can specify a vocabulary for labelling. For example, the Recreational Software Advisory Council ( www.rsac.org) took its computer-game rating vocabulary and adapted it for the Internet, calling it RSACi. An RSACi label has four numbers indicating the levels of language, violence, nudity and sex. Another organisation, SafeSurf ( www.safesurf.com), has developed a vocabulary with ten rating levels.
Paradoxically, setting technical standards for labelling has encouraged diversity, because selection software and labelling services can be developed independently. Software houses and online services can remain value-neutral by providing generic selection software without providing any rating labels; values-oriented organisations can offer labels, even if they lack the expertise to write selection software.
But this new labelling infrastructure raises two major concerns. First, publishers can label their own sites but can't prevent anyone else from distributing labels about their sites. That means the Christian Coalition can distribute a label advising people that an AIDS education page is inappropriate for children, even though the publisher may think it's vital to save children's lives. Of course, it also means the Simon Wiesenthal Centre can label hate speech without cooperation from neo-Nazi groups. Quite simply, this is the real free speech issue: you're free to put up a page, but you shouldn't try to prevent someone else from expressing an opinion about it in the form of a content label. The second concern is that individuals may not always control their own lenses. Parents and teachers should guide their children, but what happens when employers or even governments attempt to set access restrictions? Fortunately, the mesh nature of the Net makes local site controls much easier to implement than more global controls. An employee who has restricted access at work can get full access from home. Governments will find it difficult to act in loco parentis for an entire country because there are so many different access points, including telephone lines, diskettes in suitcases and network connections.
Labelling is useful for many purposes, and new labelling vocabularies that have nothing to do with the original goal of creating kid-safe lenses are springing up.
A code-signing vocabulary will allow software distributors to indicate properties of their products, such as whether they have been scanned for viruses or which hardware they are known to work with. Users can check the digitally signed labels and verify that the software has not been modified since the label was created.
Reputation vocabularies may develop. The Better Business Bureau could label commercial sites with especially good or especially bad business practices. There could even be labels for Usenet authors according to the quality of the messages they post; posts from those with poor reputations could be screened.
PICS may be the silver lining in the content regulation cloud. There had been proposals for an even more general infrastructure than PICS to handle metadata (a fancy word for the kind of information that goes into labels), but industry consensus was a rainbow on the horizon. Moral crusades are a dangerous way to conduct industrial policy, though; the next cloud may bring only rain.
Paul Resnick, from AT&T Research, and Jim Miller, from the World Wide Web Consortium, co-chair the PICS Technical Committee.