I would suspect SpaceX has the numbers. I don't think they're just guessing that what they want to do will work. Rocket engineers are big on numbers. They inte

I would suspect SpaceX has the numbers. I don't think they're just guessing that what they want to do will work. Rocket engineers are big on numbers. They intend to do a return to launch site--at least eventually. There will be several intermediate steps to prove the system. If it proves unworkable, always landing on a barge is better than not, I guess. RTLS will certainly require a good deal of margin (by rocketry standards), but first stages usually have some margin already, and SpaceX will have a nice range of options once the F9H comes on line so that they don't have to fly a rocket "fully loaded" except for certain extra-large payloads. I imagine they can also optimize most flight profiles for RTLS, by sending the first stage more than usually straight up--which is close to the case anyway. Lateral velocity is mostly the job of the second stage. While it's never been done before, it's not an unheard-of concept. One of the potential evolutions for the STS was "fly-back boosters".