I haven't been following the Web 2.0 Summit, but did receive an O'Reilly network email yesterday about the release today of Twine, the supposed vanguard of the 'new' Web 3.0. I'm assuming this means that next year's conferences will be Web 3.0 Summit, Where 3.0, etc.
O'Reilly had a write up, as did Read/Write and I imagine a host of others before the day is up. Nick Carr wrote a relatively positive review, which surprised the heck out of me, with all of the "this is the Web 3.0" talk associated with the application. I haven't tried the application, though I've put in for a beta account. I imagine I'll be among the many in the 343rd wave to be invited to try the tool.
From the start, the positives have to be the commitment to semantic web technologies: RDF, OWL, SPARQL, and GRDDL. This is very cool, especially since the tool is designed with both a REST and SPARQL API to the data. The other social applications have been closed systems, only reluctantly opening up after months, or even years, of being hassled about the 'walls' to the user's data.
Another interesting aspect is that Twine is being marketed for more serious use. This really does set it apart from Facebook, which just went through a round of criticism not long ago about the fact that none of the applications one can install with Facebook do anything meaningful, or useful. Twine seems to fill that particular niche. (The question then being, will it be 'fun' enough to generate Facebook like interest?)
Twine also promises some interesting work with natural language processing, and I heard the use of 'concepts', so I have to assume they really mean that when they say it. I hesitate, though, when I hear about automatic generation of RDF via natural language processing, because it's rather antithetical to the purpose behind the semantic web technologies: tools to give people ways to mark their data intelligently so that machines could 'read' the data and form meaningful associations. Tools to automatically generate RDF so that other machines can automatically consume it strike me as "Google does RDF".
For all my interest in the technologies being used, I see two drawbacks to Twine.
The first is that, to me, the semantic web means the web in the wild, not centralized in a specific tool or environment. If this becomes a "Facebook and Wikipedia mashup", it might be successful, and it might be semantic, but it isn't the web. The whole point of the semantic web technologies is for each of us to annotate our data, wherever we are, regardless of tool, and begin to really drive out the tiny threads of true meaning on a global scale. If we have to leave our places where we're at and go elsewhere, this seems to create a disconnect, right from the start. I have this same quibble with the other 'mainstream applications using semantic web technologies', so the concern isn't targeted specifically at Twine.
The second issue I saw with Twine is the marketing hype associated with the tool; the uses of the terms: "Web 3.0", "semantic graph", and the "first mainstream Semantic Web application"–even the location of the office (San Francisco, natch) and product roll out (at Web 2.0 Summit). I can understand the reasoning behind the marketing. After all, this is the candy that lures in the kiddies, and adds billions of valuation to chaotic applications like Facebook. Lack of effective promotion is what the semantic web adherents have been criticized for in the past: we serve steak, with no sizzle.
At the same time, the marketing, location, and early associations also serves to enclose the application within an increasingly insular community. I can't think of anything more 'not' semantic web than to become part of such an insular community.
Still, my perceived drawbacks could be more of a anti-techmeme/technorati/silly valley markeing/VC/Web 2.0 bias than based on anything 'real'. I won't know anything 'real' until I try the app myself.
Instead, I'll leave my first non-look of Twine on a final positive note. One thing Twine will accomplish is generate fresh, new interest in semantic web technologies such as OWL, RDF, GRDDL, and SPARQL, which is a very good thing. I mentioned in my previous entry on a new book on RDF that these technologies are perceived to be 'old', and here we have a hot new application making them all sparkly new again. So, those who have been asking for a new book from O'Reilly on RDF will now most likely get it now, which is also a very good thing.
