So this is Tsiolkovsky's rocket equation. Don't get hung up on the details; the basic fact is that there's an exponential curve with "mass fraction" on the Y and "change in speed" on the X. In other words, the greater proportion of your rocket is 'gas' the more velocity you get. "Velocity" in this case being "that which gets you in orbit or escape velocity."

People with a casual familiarity with rockets don't really get that your average orbital launcher is a giant tube of propellent with a flyspeck of useful payload at the end. The Saturn V, for example, is a skyscraper full of kerosene and LOX that flings the equivalent of a VW Beetle at the moon. So the real goal has always been figuring out a way to leave most of that Tsiolkovsky curve somewhere other than "on the rocket."

Rockets don't go up, by the way. They go over. They have to go up until the air is thin enough to go over, but once they're up there, they roll over nearly 90 degrees. It isn't altitude that gets you into orbit, it's speed relative to rest position. That whole "3-2-1 blastoff" big pyre of smoke thing is the first stage lifting the other stages up and over to the point where it can head out into the deep blue yonder.

That's the Scaled Composites SpaceShip One and White Knight, a combo launcher designed to ferry tourists up to suborbital so they can spend a few minutes experiencing zero gravity.

Something like 80% of Scaled's work is classified. They don't publicize this but they've been doing weird stuff for the Air Force for decades.

So the concept isn't new... what's new is Paul Allen's decision to fund really big rockets.

It isn't the concept that ooks me out. It's that hope is about only thing keeping the bodies from spindling around that wing, and that the violently-detaching load is right where the stress point is the greatest. I mean, that thing banks too tight in a turn and one tail is going to have more lift than the other and the whole plane is experiencing torsion around its drop point.

Them be some brave pilots.