So I saw this gif the other day, and was wondering, is this real or fake? And supposing there is no energy dissipated by the friction, why does such thing occur?
Can't say I've done anything as drastic as calculate anything, but a quick intuition for why this might work is to consider all the "free" travel the middle section of the bumpy track gives you.
When the ball rises over the top of the first bump it gets to fall a long way (and speed up again !) and this gives it another gravity boost. So it has a couple of those and gets some extra energy.
All the flat surface gets is those tiny little ramps on the end.
Even though the bumpy path is longer, it's also faster (maybe).