Every time someone makes another article or blog entry about Why the Semantic Web Won't Work, someone tells me that I have to respond in my own blog. For the most part, I resist the temptation to respond to each of these, since it would make my own blog reactive rather than proactive. But every once in a while, I guess I can do it.
The thing I like best about critiques of the Semantic Web is that they usually have some excellent insight, and really help sort out what can happen from what we merely wish would happen. This article is no different, and in fact, is particularly insightful.
The first insight has to do with getting people to agree on vocabularies. This is a common misconception of the Semantic Web - that it relies on a top-down edict of common vocabulary. It isn't surprising that this misconception continues to hold sway - after all, the very use of the word "Ontology" really supports this interpretation.
Of course, who am I to say it is a misconception? In my version of what the Semantic Web is and isn't, this notion of an "Upper Ontology" plays no role. I have great respect for the researchers who work on these things, but for all the reasons that Downes gives (and more!), I don't think they have any relevance to a semantic web of any kind. Sir Tim has a slide that he uses to illustrate this point; the Semantic Web standards of the W3C were designed so that a federated vocabulary can grow from the ground up, much in the way that Flickr and del.icio.us tags have grown up.
The other insight that Downes provides is, to my mind, even more profound. That is that businesses don't want to share data, since they don't get their profit margins ahead that way. This is why you won't see the IMDB exposed as a SPARQL endpoint anytime soon. I'd like to give some socialist argument that someone will find a business plan that involves sharing as a fundamental value that will beat all the selfish plans, but I don't really have the outline of that business plan.
Interestingly, it is the hobbyists who seem to be making moves toward a more integrated web. They flocked to tags in Flickr and del.icio.us; they invented microformats. They love the mash-up. They are willing to write lots of API calls to put together cool services that combine information. Even Yahoo! got into the act with its pipes. It is still a bit rough, but the promise of what it can do is pretty apparent to anyone.
To my mind, the Semantic Web (writ large, in W3C-style) isn't going to come from companies wanting to forge it. So in that, I am in total agreement with Downes. But the urge to share information is what drove the web itself into existence; that drive is still going on.
In my own consulting, I regularly watch people from all sorts of background re-invent various aspects of the semantic web in a wide variety of ways. Some use XML, some use databases, some make special-purpose APIs. It is clear to me that the drive for flexible, federated data is so strong that something will come of it. The Semantic Web will happen, one way or another.
Will it be based on the standards of the W3C that have been designed for it? Or will it be built out of a series of loosely interoperating, custom hacks? This is what I see as being the important question. As a technology evangelist, I am pushing for the quality solution. When the Semantic Web arrives, we will wish it had a coherent, consistent underlying framework. But even if it doesn't, we'll still get the work done - it just might take more hacking.