The plane does take off.
My scanner is a couple years old, and hasn"t been used in just as long. I"ll look for another one somewhere tomorrow to get the free body diagrams to y"all.
1) Friction
The plane takes off thanks to friction, or rather the occasional lack of it. There are two places with friction, despite how we want to picture it: between the plane and belt, and between the belt and its inner components (motor, wheels it"s wrapped around, etc).
If the friction is higher between the wheel and belt, then the plane literally yanks the belt in the direction it wants to go. The plane ultimately will have more force than the belt on occasion, because it is the belt that reacts to the plane. All a pilot needs to do in this case to take off is hit the gas.
If the friction is higher between the inner mechanics of the belt, then the plane is pulled back some, but due to the reactionary nature of the belt, it will still ultimately lose. The plane will fight it until it takes off. If it"s acceleration isn"t sufficient, then the pilot just pushes the stick forward more.
2) The plane"s taking off does depend on the pilot.
...but then again, it does so when there is no belt, as well. The pilot has to make sure that the plane"s engines kick the belt"s ass.
3) If the pilot sucks, and the two sources of friction are equal, but the plane still takes off.
Such an equillibrium is always uneasy. The whole thing will oscillate with the plane"s wheels spinning out until the plane takes off anyway.
One interesting thing we found is that whenever the plane"s engines give it a good kick in the ass, and it overcomes the belt, then the wheels actually spin faster than they are driven to spin by the force of the plane"s thrust.
All this means is that the wheels spin out often, and the belt adds to their angular momentum. At these points, the wheels don"t matter at all.
In other words, those who have said that the plane lifts off have been exactly right, and almost entirely right about the reasons why. The only difference between their hypothesis and reality is that friction can"t be ignored, because it is ultimately the dynamics brought about by the friction itself that defines the system.
When I add the diagrams, I"ll just edit them into this post as links. Failing my being able to figure out why this POS I have isn"t working, I"m pretty sure there"s a scanner at the library here.