Nivi has a good post on how RSS acts as an API for content. I think he's on the right track, I wanted to add a bit to his thoughts. I see RSS as more of a model for content, rather than an API for content. The API is the methods and parameters needed to access an application. The method by which that data is encoded such that the client application can make sense of it, the model for that data, is where RSS fits in. RSS (as well as other syndication formats), at the core, is an abstraction of what a piece of web content is, which is why it works so well in so many different cases.

What we've focused on with Blogdigger is building a platform with an open interface (API), available as RSS. Every bit of data in our system is accessible, not only as HTML to view in a browser, but as RSS (actually, some of our data is available only as RSS; HTML is sooo Web 1.0). We've tailored our API (the search interface) to allow for as much flexibility as possible, so you can do anything from boolean search for keywords to finding content by location (not to mention media, links, tags, site, subscriptions and a whole bunch more).