Okey in order.
@Low-Q If I understood the video correctly, the device he uses to push the ball is not only a measurement device of force, it is a power device which measures the newton-meters. That should mean that no matter how he utilizes it, he will always have expended the same amount of energy when it is pushed all the way in, as he showed in both methods.

@Merg
As Low-Q pointed out, a huge difference in length moved is displayed between the two driving methods. This means that while the short path is struggling more than enough with only overcoming the static friction, the longer path is mainly concerned with the dynamic friction.

But yes, if he is using a good measurement device that describes the energy used accurately, unaffected by the radical change in resistance seen here, well then this is quite good proof.

It should though be stated that there is a slightly higher loss of energy in the form of vibration in the short-path demonstration. Other than that this is a good video.
Julian

I cannot see any good proof anywhere. I see it like pushing a car with the breaks on. And when the breaks are released, we have overunity?

The first experiment are highly inefficient. The second are more efficient. Take away the friction from the surface, and put on some wheels. Then you can test this again and see for how long the rig will roll with a locked pendulum versus a released one. My guess are that these two experiments will gain the same net acceleration and velocity - with low friction wheels on it.

I was going to post something different, but I see you are probably right.
The static friction is the main problem here, it is a brake which doesn't matter as much in the long path experiment.....

milkovic Idea
http://www.youtube.com/watch?v=nAA71WhkyaQ

Milkovic / Bedini lighting 12 LEDs
http://www.youtube.com/watch?v=lVVVXJfGItI

Bedini-Powered Milkovic Two-Stage Oscillator #2
http://www.youtube.com/watch?v=pIvSlYoqnBM