At 30 seconds in, shouldn’t we reach 1/√2 or about 70% of the torque at 45°? Also, the gain for such a new system will depend on how good are cyclists at apply

At 30 seconds in, shouldn’t we reach 1/√2 or about 70% of the torque at 45°? Also, the gain for such a new system will depend on how good are cyclists at applying force mostly at the points where torque is maximal. I wouldn’t know, but would think the net gain is smaller than what a model of “cyclists applies the same force all the time” predicts. I also wonder about friction losses in this setup. Ignoring the additional chain (it can be engineered away in the final product), I would think those perpendicular gears and the rollers on the track will add significant losses.