Whats super interesting to me is how this system becomes dynamically stable. In the pure thick shaft version, it looks like the jet is centered on about 1/4 of the way down the shaft - so it should be imparting torque on the whole thing, but the fluid dynamics are enough to keep it at what appears to be a pretty constant angle.

Also, as a note about the golf ball comment (this seems to happen quite a lot), nothing he said was wrong, but there's slightly a bit more nuance going on. For a sphere (since we're talking golf balls) in a given flow, there will be a point along the ball will the flow will detach and form eddies and the mentioned low pressure zone behind the ball causing additional drag. By adding dimples, you introduce turbulence into the boundary flow that basically gives it enough momentum to keep up with the ball for longer, letting it stick the ball and therefore separate later - therefore lesser losses through that pressure drag.

The reason why we don't dimple everything flying through the air is that adding turbulence to the boundary flow also increases the amount of drag on the object (so called skin friction drag). The act of adding dimples may either increase or decrease total net drag depending on the exact parameters of the object and the flow.

It just happens that for golf balls flying through atmosphere, typical drive velocities are right in the range where adding dimples causes noticeable net reduction in drag.