Charlie says that "the whole idea that you have a main site is dead". I couldn't agree more - I strongly believe that platforms are the future of the Internet.
One clarification I'd like to make in this whole discussion is Facebook is both a destination and a platform. It's important to understand that these are discrete things - something can be a platform without being a destination and vice versa. Clearly, Facebook offers a nice API for integrating your code into theirs, but this to me is not what makes Facebook a platform. MySpace, iGoogle, and a plethora of Web 2.0 portals allow you to "embed" your code - Facebook just allows you to do it more seamlessly. If anything, these are all simply platforms - or rather, vehicles - for traffic. Of course, as Charlie and myself and countless others have said, traffic does not give you a business model.
On the other hand, Facebook is getting flak for not being open enough with their data. As Fred says, being open in one way is not enough. Yes, the fact that Facebook does not (yet) allow Twitter to update your Facebook status is annoying, but as a platform, Facebook is important because it exposes the underlying relationship information to consumers of the API. This allows me to build an application that leverages this data without having to go out and collect it myself. While I appreciate being able to subscribe to my news feed in my newsreader, ultimately it's this social landscape already available in the API that is important.
This is precisely why Facebook is appealing to us as a platform and something we consider a key element in our launch strategy. As Charlie puts it, "at the end of the day, you shouldn't care where the user winds up... everyone can access your content or your service in a form native to the platform that its on, but will the full functionality of whatever you're up to." Facebook gives us a natural place for our users to wind up - both because of the potential traffic and the data it exposes - but ultimately we recognize that it is but one of many destinations.
Along these lines, it's why new social networks like Plaxo Pulse leave me scratching my head. Plaxo already knows who your friends are - the fact that it asks me to add a friend on its site that it knows I'm already a friend with is just ridiculous. Despite already leveraging the social networking data exposed via Facebook and others, Plaxo is trying to reinvent the social networking wheel.
I'd also reiterate the point Charlie makes about dogfooding. We built our site on top of the API so we can ensure that the API properly captures all of the functionality we want available in the API. If you're not able to build your site with your API, then your API is not an actual API.