Thanks! Interestingly, figuring out the issue was only made possible as a result of the change that I had made previously, where I replaced the 4 single transparent bricks with one 1x2x5 transparent brick, to make the internal action more visible!

Yeah, as soon as he did that, I thought "Man, why didn't I think of that?"

Now that I've seen and understand the failure mode, I can see how giving the flaps more room to move would keep them from jamming if the ball moves off-center. I'm not sure what else that extra plate might do to the design though. Certainly something to keep in mind if the pyramids end up not being 100% reliable. (I won't know until I use the new design at a show.)

Funny you should mention that... I had the same thought this morning, thinking that regular 1x2 cheese slopes might be a good alternative to the pyramid slopes. I tested it, and while it worked ok, I didn't think it worked as well. In the meantime, I've been told that another alternative is to place a single 1x2 plate on the side closest to the in-feed. This creates an uneven surface on the top of the piston, and the studs on the 1x2 plate will now grab the balls at an angle (rather than from the bottom), generating more lateral resistance and keeping the balls centered. I just tested this approach, and I find it to be (so far) as effective as the pyramids. (And more effective than 1x2 cheese slopes.)