Raw

Premier podcaster ( and Talisian colleague) Paul has a good post over on the Nodalities blog [ feed] discussing the notion of platforms, comparing relatively closed portal-like systems such as Facebook with inherently more open systems such as the Talis Platform. A nice line from the end-user perspective:

In these days of RSS feeds, widgets everywhere, more one-stop-shops than any of us could possibly ever need, why can't Facebook (or MySpace, which I have never used, or LinkedIn [ I'm here], or Bebo, or whoever) operate on my terms, allowing me to place the functionality that I use in contexts of my own choosing; whether on or off their site. Openness is great. But it's a two-way thing, and Facebook don't seem to get that yet.

[[Facebook [ I'm here], MySpace [ I'm here], LinkedIn [ I'm here], Bebo [ I'm here]]]

As Paul says, the platform word has been applied to quite a lot of different ideas - in the past I've often asserted that the Web is The Platform, and I don't see any reason to change that view. But clearly with the kind of complexity we commonly see around interlinked documents and data something systematic is needed to allow the developer to work at a suitable level of abstraction. Web standards are a key part of this, as are commoditized functional units that will be reused in a large proportion of applications (e.g. Web clients/servers, triplestores...). Aggregating these components into an integrated combination is the obvious next step.

Again as Paul says, there is still value in what Facebook are doing platform-wise, and as Kingsley says in comments over there:

Semantic Data Web middleware can take Facebook Data "on the Web" and Inject "into the Web"

Although wider awareness of what Semantic Web technologies enable for the Web of Data will undoubtedly change the ratio of traditional silo apps and native semweb apps, middleware is bound to play a huge role in the foreseeable future.

At this point in time there is a lot to recommend things like e.g. APP Stores for content-oriented applications, not least that their RESTful nature makes them true Web systems and hence entirely amenable to Semantic Web integration. Personally I'd favour starting with content-capable systems which already have significant semweb capabilities built-in (such as Virtuoso and of course the Talis Platform) even if these are not required for initial application use cases. But if clean, standards-based interfaces are in place there will still be good extension stories over and above reworking things internally - easy migration to other infrastructure is one, and the kind of middleware Kingsley refers to is another.

PS. Paul sets a challenge in comments over there:

So here's the thing; how to make 'linked and open data' the next cool and fashionable must-have for the typical Facebook user... and answers can't include 'triple', 'RDF', api', 'data', 'uri', 'REST', 'triple', or any of the other arcana to which we turn in explaining what we mean to one another.

See also: Pre-Release Tips for Web 3.0 System Builders , stats on Facebook usage 

@en