As I understand the physics, the whole process could be made FAR FAR FAR more efficient if the rocket was boosted to about 40000 feet with a subsonic airplane, a' la' X-15 and such. It's above 75% of the earth's atmosphere (dramatically reduced drag), is already going 600 mph in the correct direction, and is 8 miles closer to the ultimate goal 250 miles up). This might not sound like much of an advantage, but if you've ever worked out the mathematics of the Saturn V (or space shuttle, etc), the VAST majority of the fuel was used up in the first 20,000 feet, maybe even the first 5000 feet. It would be even better if the first stage could be an air-breathing supersonic ramjet, but that's not my field of expertise.
Cypherpunks isn't the right place to discuss this in detail, but... Efficiency != Cheap Kerosene is cheap. Steel fuel tanks and rocket motors are quite cheap. Making big dumb rockets is well understood. However, aircraft integration is not. If you use an 'off-the-shelf' aircraft, it has a human in it. That means the whole thing must be safe. If you don't, you have a drone aircraft which isn't cheap at all. Remember, the cost of materials scales linearly with size. The cost of a complex system scales as the square of the parts count. These arguments are hashed out (admittedly without consensus) regularly in the sci.space newsgroups.