Some Thoughts on Tag Directions

Noah Mendelsohn
June 3, 2005

Overview

Our main subject for the upcoming face to face meeting is "TAG Directions". Tim set down some ideas in the note at [1]; here are some of my thoughts.

As Tim suggests, we should give serious consideration to documenting the architectural underpinnings of the Semantic Web, and perhaps of Web Services as well. I do think it's also useful to step back and take a bit broader look at our goals. IMO, our overall goal should be to maximize the success of the Web over the next 3-5 years and beyond, and we should set our priorities accordingly.

Specifically, we should consider:

• Ensuring that the Old Fashioned Web (OFWeb) is properly used and that it continues to scale. Maybe or maybe not AWWW Vol. 1 has covered everything we need, but we should recheck. I think some areas may still need attention.
• Technology and information content is changing: we need to be sure that the web is scaling to embrace such new content, and that the Web is not being inappropriately undercut by non-interoperable alternatives.
• We should look for opportunities to grow the web to provide services or capabilities that were not present before. The Semantic Web is an example of such innovation.

Some more details on these are provided below. I propose that we invest in these three areas, aiming for the balance that will best contribute to the overall impact of the Web in coming years.

The Form of Our Work Product

Last year we published AWWW Volume 1. Rather than assuming that our new work should be a volume 2, I suggest that we first set an agenda regarding technology and content. We can then decide whether to republish volume 1, establish a volume 2, focus mainly on findings, or whatever will be the right vehicles to carry our new messages.

If we do have additional volumes of AWWW, then I think they should be organized by topic rather than chronologically. We should give Volume 1 a subtitle (perhaps "Identification, Protocols, and Interaction") and make sure that Vol. 2 covers something clearly distinct. We can rearrange content from a republished volume 1 into volume 2 if necessary to create a sensible structure.

Bottom line: my thinking on how to split volumes is still pretty murky, but I'd lean toward doing it thematically, and giving ourselves license to republish or rearrange Vol 1 if useful.

Details of the 3 Themes Listed Above

That's the big outline from my point of view. Here are some notes on the three themes mentioned above.

• Ensuring the Success of the OFWeb

  While much of the core technology for the browsable Web has been covered, I think we need to be alert both for remaining problems, and perhaps for new challenges arising from changing circumstances. Examples include our open issues on XMLVersioning-41, mediaTypeManagement-45, putMediaType-38 etc. Maybe or maybe not those are the right specific issues, but we should be alert to breakage in such areas, and continue to offer guidelines for robust use of Web mechanisms, updating AWWW as appropriate.

• Responding to Changing Technology and Uses of the Web

  For the Web to remain the premier shared information space, we need a robust architecture that scales to handle changing content, technology, and application models. I think the TAG should take a hard look at some of the following, to see whether they suggest work that's needed either on the Web Architecture itself, or in our writings about it:

  • An interesting fraction of the bits moving on the internet are in P2P systems that are not well integrated with the Web. What can or should we do to ensure that the Web successfully integrates such content moving forward?
  • Multimedia continues grow in importance. Whether HTTP in particular can scale to supplying and maniupulating truly high quality video streams seems an open question. If it can't, then there's some reason to believe that we'll face choices between encouraging the use of non-HTTP protocols and excluding such content from the Web. In any case, explaining how integrate such rich multimedia seems like an approriate area for the TAG to consider.
  • Computer interfaces are becoming richer and more dynamic. Some of the models being created to drive such interfaces involve nonstandard content conventions (Flash, etc.) It seems important to ensure that the Web remains the unifying worldwide information space as expectations for richness of content and interaction evolve.
  • New application models are being built using standard Web technology. Among these is "Asynchronous Javascript and XML" (AJAX), which is widely deployed in applications such as Google Maps (http://maps.google.com). At one level, these represent a successful exploitation of the standard models set out in AWWW. At another level, they can challenge our assumptions of, e.g. what gets named with URIs (as you pan through a map of the US for example.) We should at least consider these models to make sure that we are comfortable with the way the Web is being used, and perhaps to expand our explanation of Web Architecture to give principles for their construction (e.g. ensure that a URI can be minted to capture any interesting state of the application.)
  • Mobile devices are becoming ubiquitous, and the W3C is making an increased investment in the supporting technology. The TAG should be alert to architectural issues that may arise in supporting the mobile web.
  • RSS and Atom are growing in importance. To some degree they represent appropriate applications of the Web Architecture, but like SOAP, they start to move into the space of standards such as multipart MIME that are already widely used (I.e. both RSS and mutipart are being used as containers for multiple substreams). Maybe or maybe not these developments require attention from the TAG.
• Growing the Web Beyond Browsing and Information Sharing

Initiatives like the Semantic Web and Web Services bring capabilities that the Web has never had before. Sem Web in particular represents a very large investment for W3C, and we should indeed devote TAG resources to helping it "reach its potential". I'm inclined to suggest that we consider the Semantic Web in particular at two levels:

1. The Semantic Web employs existing Web constructs such as URIs, sometimes HTTP, etc. I think our fundamental focus should be on ensuring that these core mechansisms of the Web are indeed usable in the Semantic Web. httpRange-14 is thus a key question, because it forces us to clarify our underlying story on URIs to the point where they become usable for the semantic web.
2. We could also invest in writing up architectures for the Semantic Web itself: how to use OWL, RDF, etc. In this area, I'd want to consider (a) is the technology sufficiently stable and widely deployed that now is the time to freeze such an architecture and (b) to what extent are we the right ones to do it, and to what extent should that come out of the Semantic Web Activity (with us consulting)?

In the case of Web Services, we should at minimum continue to ensure that the Recommendations are designed to appropriately integrate with the Web. Maybe or maybe not we should go further and dive more deeply into WSA architecture; as with Semantic Web, I'd be inclined to leave that to the pertinent "Activity", with us providing guidance. As Tim notes, the development of such an architecture has been tried at least once in the Web Services area.

Summary

Those are some of the ideas I've been kicking around. The big theme is: figure out what's going to be important to make the Web successful over 5 to 10 years, then let the form and content of our publications follow from that.

Noah

[1] http://lists.w3.org/Archives/Public/www-tag/2005Apr/0054.html