Another possible method of steering is to build four or more reaction jets around the rocket; could be fed with a monopropellant or exhaust from the main engine or its turbopump.
If you can deal with slight decreases in thrust, you can also do thrust vectoring with injecting water or other suitable liquid into the nozzle in proper places. Could be mechanically simpler than gimbals in some cases.
...they don't start working until you're fast enough (quite high above the ramp, which is a sensitive area where you really need it) and they stop working once you leave the atmosphere (which is all too soon for an orbital carrier). Why bother with them?
There generally on the bottom stage so they don't reach Anything close to orbit. But what there really useful for is dampening rotations which is hard to fix with a thrust vectoring system.
This has been being done for a long time. Saturn V's were thrust vectored too. For any large scale rocked, the fins are far to large to move like on smaller missiles.
This works well to cut down on mass and drag and is especially useful for silo/sub launched missiles where fins don't fit well in the launch tube.
Before someone corrects you with a photograph, the ascent stage had fins in addition to the vectoring on the F-1s. However, my understanding, though I might be wrong, is that those were intended for stabilizing the vehicle instead of acting as control surfaces.
http://www.youtube.com/watch?v=V3L7crGudVU
Finless rocket? How does it steer/stabilize? Do the engines have fine control over which side their fire perhaps?
I guess fins are useless outside of atmosphere anyway.