Web inventor Tim Berners-Lee Unplugged: Semantic Web better than APIs for data access
Last night at the MITX (Massachusetts Innovation and Technology Exchange) Technology Awards held at the Four Seasons Hotel in Boston, MA, the inventor of the World Wide Web Sir Tim Berners-Lee was awarded the organizations 2007 Lifetime Achievement Award (last year, Nicholas Negroponte was the recipient). Prior to the main event getting underway (many other awards for innovation and leadership were handed out to Massachusetts-based hi-tech companies), knowing that Sir Tim was "in the house," I asked about his whereabouts and was led to a VIP reception where he was holding court with several attendees including Fortune Magazine senior editor David Kirkpatrick (who later moderated a great discussion about the mobile Web). As that reception wrapped up, Sir Tim stuck around to answer some questions on video.
He and covered a fairly broad range of topics. We started out with a report card one of his most important initiatives as director of the World Wide Web Consortium (W3C): the Semantic Web. For those of you not familiar with the Semantic Web, I asked Sir Tim to state its value proposition.
You can listen to what he has to say about, but the general idea is for there to be a layer of data on the Internet that he calls the "data bus" and the way the data bus works is not too different from how we've heard Microsoft's WinFS filesystem described where connectivity between related data items is organic rather than synthesized. For example, whereas today, a mashup developer may have to call upon two APIs to show where a specific Starbucks is on a map, the Semantic Web approach might involve little more than a simple query of that data bus using a query technology called SparQL.
On the other, a lot of big Internet companies would probably prefer developers go the non-standard API route because of the way API-dependencies can result in developer loyalty (ok, "lock-in"). After all, once code is written and reliant on APIs (and it works), API extrication (in favor of using SparQL against RDF) will invariably entail a rewrite. That is unless developers are anticipating the Semantic Web and modularizing their code in such a way that they have query modules that abstract query specifics. In that case, so long as the module returns the same information, it's only the guts of the module that have to be fixed (trust me, it's much more complicated that I'm making it seem).
The message (regarding data access) from Sir Tim was of course very much about standards. If you subscribe to the notion of the Semantic Web, then you also believe that data access should involve standard mechanisms for data connectivity and queries (as opposed to APIs). That discussion of standards (we talked about the royalty-free issue as well as open source) was a great lead in to the next issue that I most wanted to hear from Sir Tim about: standards in the RIA (Rich Internet Application) space.
The big question there is whether the existence of exisiting non-standard (non-de jure standard, that is) RIA development platforms (eg: Flash and Java) along with the arrival of new ones (like Silverlight) is something that requires the attention of the very de jure-standards focused W3C. Not surprisingly, the stovepiping of the Web is something that is very near and dear to Sir Tim's heart. Check out the video. Or, if you don't have time to watch but want to hear the interview. We've stripped off the audio and made it available as a downloadable podcast or you can just hit the play button above on the Flash-based podcast player ( read more about subscribing to the podcasts so they show up automatically on your PC or MP3 player).