haha. Maybe I got caught up watching this a little longer than you guys, but at times the pattern does begin emulating a circle pit. Both directions.
It's sort of hard to determine a "push" from the perspective of spheres, but IIRC I saw some of that behavior as well. I mean pushing insofar as red pieces would occasionally move erratically away from the walls.
I do agree about the lone mad rager acting as an unstable epicenter, though. That's key behavior, and something I did not notice in this simulation.