[whatwg] Question about the PICS label in HTML5

Phil Archer parcher at icra.org
Thu Apr 17 01:37:30 PDT 2008

Marco, thanks for raising this and thanks to Dan Bri for altering me to 
your question.

I'm CTO at the Family Online Safety Institute [1] which includes ICRA 
and chair of the POWDER WG [2] so I'm probably reasonably well placed to 
answer your question.

As of today, PICS remains a W3C Recommendation so if HTML 5 is not going 
to support it in any way, that strikes me as questionable. PICS 
continues to be supported in Internet Explorer and, much to my surprise, 
this feature  was updated to work with the current ICRA vocabulary in 
IE7. This was based on research that MSFT did into its usage which found 
that 'just enough' people used it to warrant its retention. I don't know 
whether it will survive into IE 8. We do still issue PICS labels 
(alongside an RDF-based label, the syntax for which is not standardised).

With that on the table I've spent a lot of energy over the last few 
years working towards seeing PICS withdrawn in favour of something that 
can do what it does but smarter and better - which is what POWDER is 
about. From my point of view, and this is echoed by what MSFT told me at 
the time of the Content Advisor update for IE7, PICS is a legacy system 
that we need to support for a while longer yet - but no one, least of 
all me, is arguing for it to be updated - the major update for PICS is 
well established. It's called RDF.

So what is the Protocol for Web Description Resources (POWDER)? It's a 
way to attach content descriptions to collections of resources (e.g. a 
Web site) - a way to create RDF triples about lots of things at once. A 
POWDER document is a separate XML file that has a GRDDL transform 
associated with its namespace which, if performed, will yield RDF/OWL*. 
Description Resources have 3 key elements:

- attribution (who is making the description)
- scope (what is described)
- the actual description which can be provided as RDF terms (i.e. 
controlled vocabularies) or free text tags

We plan to use FOAF for the attribution side of things and provide hooks 
for authentication routes so that the trustworthiness of the description 
can be established.

What do we need for HTML 5?

Just the link/rel element. A POWDER link will be something like

<link rel="powder" href="powder.xml" type="application/xml" />

so no surprises there. The relationship type of powder may become 
'describedBy' - that's something we'll be deciding within the next 
couple of weeks.

Actually, one of the problems with PICS is that it has a peculiar syntax 
which a lot of WYSIWYG HTML editors muck up so a constraint on POWDER is 
that we make use of well-recognised HTML elements.

The e-mail thread you pointed to on the TAG mailing list is about being 
able to put that link into an HTTP Header. We actually recommend this 
for ICRA labels based on the fact that HTTP Link was in an older spec 
but got dropped. Mark Nottingham has an updated Internet Draft [3] and 
the basic plan there is that the value of the rel attribute is a URI. 
Relative URIs are relative to the IANA namespace, so we'll probably be 
working to register our relationship type there but I know I also need 
to follow up on a conversation I had with Hixie about this in Boston in 
November and get it in the HTML 5 list too so all bases are covered.

In summary, unless we make a complete lash up, POWDER's main tech 
documents will be at Last Call around the end of this month and into CR 
over the summer. PICS will be withdrawn in its favour around the end of 
this year. Support for extensible relationship types is important to us. 
Support for PICS less so.



[1] http://www.fosi.org/
[2] http://www.w3.org/2007/powder/
* Yes, the GRDDL transform will produce RDF/OWL but for full logical 
processing, a couple of 'semantic extensions' need to be implemented.

Phil Archer
Chief Technical Officer,
Family Online Safety Institute
w. http://www.fosi.org/people/philarcher/

Marco wrote:

I've been looking through the HTML5 working draft and I've been trying
to find a reference for the use of the current PICS labels.

I noticed that the new specs only give three accepted keywords for the
http-equiv attribute, which doesn't include the current "pics-label":

Being a custom attribute, it makes sense that you wouldn't want to
promote this type of thing.

Having said that, I was searching within the new specs for a way of
accurately describing one's content. I'm somewhat aware of the W3C
POWDER WG and the only post that I could find that was recent and
relative to this is here:

The closest type of markup that comes close to describing content that
is accepted with the HTML5 validator is the current ICRA label:
<link rel="meta" href="http://yoursite.com/labels.xml"
type="application/rdf+xml" title="ICRA labels" />

My question is: what is the direction for describing the type of content
you would have within the context of the HTML5 working draft?

Looking over the latest working draft, the closest I could see to where
this might possibly be applicable is section 3.3.4 - Transparent content
models. Would it be that you'd have markup right in the body that is
considered transparent but describes the content on the page? Or, would
you take something like the approach with the ICRA example above?

If anyone has insight into this, I'd certainly appreciate it.

Marco Battilana

More information about the whatwg mailing list