A lot of the code we've written in this project has been concerned with getting the data we want to display. Actually adding the infinite scroll feature itself requires only a minimal amount of code – a single handler that watches for the scroll event and triggers a new request for more data when the document is scrolled to the bottom.
As you can see this is a feature that would be easy to retrofit to existing functionality as an additional layer. This technique is best suited to data that can easily be arranged in reverse-chronological order, with new items appearing at the top and older items appearing at the bottom.
It's not necessarily a complete replacement for paged data, but certainly makes sense when dealing with things such as news stories, blog posts, tweets, or status updates. It works very well with social data.