If the only force was friction why does it stop working when the propeller is disconnected. In fact any amount of friction reduces it's ability to climb the treadmill so claiming the only force is friction doesn't wash.
http://www.youtube.com/watch?v=LMgDvC5lqsY
Just supposing in this video, where we are not shown what is to the left (equivalent downwind) of the treadmill, there is a ruddy great fan. This would explain:
a) the motion of the vehicle in normal test mode, propelling itself, we are told, faster than the progress of a belt under its wheels, driven only by that belt under its wheels - it could, in fact, be being pushed via air blown left to right, turning the prop (actually a turbine), this giving the required extra energy Newton didn't think of,
b) the motion of the vehicle backward, as would be expected, when its turbine drive is disconnected, and
c) the apparent turning of the prop from a standstill at about 0:50, after it's second pass round the frame of the treadmill. What seems to happen is that when spork is holding it by its gear joint, as he does most of the time (oddly), the prop comes to a rest. Then, on that second pass, he momentarily has to change hands, and grabs the machine by its frame between the wheels. At that point, with no great forward motion of the machine in his hands, the prop seems to begin turning of its own free will. Spork holds the machine again immediately by its gear, and it stops again. He also seems always to place it on the treadmill while holding it at this position, perhaps because if he didn't apply a little friction there, we would see the prop spinning without the wheels being in contact with the belt, and, while he manages to get away with convincing some people that the system can gain energy from nowhere while it's going, no-one will believe it sucks power straight out of motionless air.
So, ok spork, if you want to get some way nearer to a decent test video, we need you to pan both ways, all round the room preferably, to see the whole rig. You need to hold the machine by its frame sometimes as you hover it over the treadmill. It might also help if, having let go, you could do a little jig or spin on your heels - that way we're more confident that you're not holding a piece of thread attached to the machine to make it track 'uptread'.
However, nothing will really convince me except making my own, and I ain't gonna do that, now am I? I'd like to know if anyone here has taken you up on the idea, who was a sceptic, built one, and now says well blow me down with a wind-turbine, I'm converted.
Let's take another think about this from the physics again. You say that there is some kind of equilibrium state - if it goes too slowly, it's 'obviously' gaining energy until it speeds up, which is where the stopper is needed at the front. But you don't - presumably - claim that it would continue to accelerate indefinitely if not stopped, so it would also get to a terminal velocity. If somehow propelled faster than that, the system would tend to return to its equilibrium again.
There's nothing odd about that and, taking the vids at face value, that fits. However...
Now, imagine it's just going a little faster than that equilibrium, for whatever reason - maybe the operator gave it an accidental push. Right, so we can recognise what happens. The various inefficiencies in the system cause it to slow down. Slowing down, however, causes its wheels to be moving slower w.r.t. the treadmill (if you push it against the tread, the wheels spin faster, let it roll backwards, wheels slow down w.r.t. the treadmill's steady speed). Now, that backward acceleration of the cart must reduce the power derived from the treadmill, which can only translate into less power output at the propeller. Less power output at the propeller can only cause the cart to be pushing forwards against the air and up-tread a little less. If it pushes forwards up-tread with a little less force, the relative speed of wheels over tread is further reduced, which is supposedly all that's powering the propeller, so the propeller will slow a little more, which will.........I think we can all see where that's going.
As I put it much more simply before, to cause that cart to remain at rest in a system where the only input force is right to left....well, the sentence can't be completed without contradicting itself. It would remain at rest if there were no friction at the wheels/tread interface. You have disproved Newtonian physics. Furthermore, it's not supposed to be staying at rest, but moving up-tread. You have simulated a surface with
negative friction, using a few bits from a hardware store. All our energy problems are over.
