The Problem with the Pendulum Rocket Fallacy is that it assumes the rocket is a rigid body, but this is not the case for Kerbal rocketry. When the rocket is no longer a rigid body, the two cases become very different. For engines on the top, gravity acts on the center of mass of the rocket, and the thrust acts in an opposite direction in the opposite direction of this force. (That is, gravity points down from the center, and thrust points up upwards of the center.) These forces are thus in a stable equilibrium. When this is reversed, with gravity pointing downward above the thrust pointing upward, this causes an unstable equilibrium. The rigidity of a rocket renders moot any unstable equilibrium that might exist in this case. But because the rocket is not rigid, any slight wobbles will necessarily cause a net torque on the rocket. I performed a test for this. I made two 30m tall rockets, one with engines at the bottom, the other with engines at the top. for the engines at the top case the rocket tended to go straight up, but for the engines at the bottom case the rocket curled up and crashed into the ground.
