Well, it's definitely different. I prefer the carrier aircraft idea for all reasons except size. The Stratolaunch being built in Mojave can only carry a Falcon 4 or Falcon 5, and it's the biggest yet (about 200K lbs thrust). Something like the B-1 or British Vulcan would scale up quite nicely, and the Vulcan's thick low-aspect ratio wing is structurally an easy thing to scale up tremendously. Thrust isn't a problem because you can use as many engines as it takes, perhaps using GE-90's with about 3 times the thrust that the Stratolauncher's old 747 engines use. You could even design for supersonic flight using something like a vast array of GE F-404's we'll be retiring with the F-18 B's and C's. But the cost would probably be several bilion and you're probably not going to get much beyond the payload capability of a Falcon 9 or Falcon H, which while nothing to sneeze at shouldn't be the end point of space launch. However the Stratojet also shows that you can hang something like a fully fueled Falcon horizontally from some hardpoints and it's good for 1.4 or so G's with a human-rated safety factor, otherwise they wouldn't be able to fly with it. That means there really are no significant structural penalties in a horizontal lift-off. Hypersonic flight vehicles that pull several G's with pressurized tanks tend to be pretty tough.
What I'm trying to get around is the size limits on reusable launchers set by the carrying capacity of crawlers, the tower height, and the processing building, and the scaling laws likely are part of what makes the SLS cost/LEO pound so much higher than the Falcon 9 or Soyuz. Part of that is the engines, structure, and manufacturing, and part of that is the tower, VAB, and vertical processing. Once you severly cut engine, structure, and manufacturing costs, as SpaceX has done, you're left with the vertical operations costs. If you start re-using the entire rocket, as SpaceX intends to do, the vertical costs will start to dominate the bottom line. The Falcon 9, Soyuz, and Zenit avoid that by being fairly short and doing as little as possible upright, with a simple boom that raises the rocket. That gives way pretty quickly to a tower as the rocket size increases, and tower costs would probably follow a cube or fourth power law versus height. At some point most of your employees are tower-painters.
In contrast, a single maneuver right after lift-off is pretty simple. You know the rocket's launch mass and moment of inertia and can exactly calculate the thrust profile and impulse required. A lot of military missiles use a stage 0 booster and post-launch maneuver just to avoid back-blast from the main engine. The Navy even launches ICBM's from underwater. Some of the rocket hobbyists already have difficulties erecting their towers and rockets, and then have to climb precarious extension ladders for launch prep. If they'd add some high-thrust low-impulse solids they could leave it horizontal, blast it ten or twenty feet in the air while rotating, fire one opposite motor to kill the rotation, and then ignite the main motor, using anything from a simple timed sequence to a sophisticated control system (which the BATF really frowns on). It might be fun to play with some Estes motors and a simple controller just to show something weird and new in a hobby launch.
Doing something like that on the Falcon 9 would only take two Orion abort motors (which have 500,000 lbs thrust each and a 5.5 second burn). It's pointless on the Falcon, but they're not blowing a billion dollars on facilities. Air-starting the first stage shouldn't be shocking because we've been air-starting second and third stages, in flight, since the 1940's.
If you go liquid fueled for all the motors, you could have an intact abort and landing capability, and only about half the main engines at the base would need to rotate to lift the back end. If you want to avoid the weight penalty of liquid engines at the forward end, just use expendable solids, although that would probably rule out an intact landing unless you went ahead and seperated the stages and did an upper-stage fuel burn till it was light enough to set down. It's really not a major engineering challenge, and the result would have vastly more control authority in pitch and yaw.