Giving Clarence's puzzle some more thought. Having the entire span fill with air or a trail of bubbles seems fanciful. There is already a span wise flow from tip to root on the windward (lower pressure) side, and in any case the pressure gradient that would suck air also sucks water. My guess is ventilation may progress a few cm along the span, in effect shortening the fin. A propeller provides a additional mechanism to move air around as its spins in a vortex. With an effectively shorter fin its easy to exceed the angle of attack where separation occurs and the relationship of increasing angle of attack to force flips, and you spin out.
I believe similar occurs with weed on fins. Just a guess but the behaviour fits the theory. 
