It's a toy application on a real infrastructure. Wait until developers start writing some real value-adding/lifestyle Facebook apps. Real Facebook applications will intermediate between users without requiring their direct input (think Personal Assistants). That being said, 300M page views/month is something most indie developers would sell their friends for - though it might be a bit embarassing that it's such a LCD app.
I wouldn't expect to see value-adding apps too soon... unless FB is willing to make some significant changes to their platform. There's simply no incentive to make them right now -- virality continues to be the main criterion for app success.
The #1 value added app for lots of people, whether they know it or not, is exportability insurance. Make an app that tracks everything externally so that if Facebook starts to suck, you can migrate the next platform.
The existence of that portability might actually ensure things never start sucking.
I think five years from now we won't be logging into Facebook but we'll be talking to our mobile devices and they'll be scheduling reservations for us (it will all be autonomous agents doing tasks/chores for us on behalf of our openID)
I think those are too low. At $0.05 CPM, 300 million page views would earn just $15000 a month. I doubt that that many pages could be served for a few thousand dollars. Look at their hardware and bandwidth costs and work out what a profitable CPM would be from there.
That said, is this YC-funded?