The reason we did our preview launch in August was simply to getsomething out the door. As soon as we picked the date and a minimal feature set, we got rid of loads of other features we were playing with (some of which we’ll bring back, some of which we won’t). We also identified and built a bunch of infrastructure we’d need to host real users. And we got more done in less time than I’d ever seen any team do. It was magical and fun. Before we had the date, frankly, we were drifting.
More real world advice on Ev’s post at Medium.