I'm of the opinion of launching as soon as you can. You can only really learn when the app hits the wild. Realize though that apps are not websites, they can't be iterated as fast and a furry of bad reviews because of a buggy launch can haunt you forever. Some initial closed, pre-App Store launch is needed to make sure the basics are covered. No crashes, no major UX bottlenecks and ease of performing the 2-3 main things you want people to do.

Dan's idea is great if you'd like to save the US (or any other country of your choice ) for a big launch with press and a furry of users. Then you'd want some things further iron out. Some apps never do this and just organically grow their way to awareness by pure awesomeness and hustle.

Btw, love the idea and just signed up for the beta ;)