I agree that the demo is mixing concepts (esp.: relies on actual gravity to work) and tempting confusion. As a teacher, you'd have to be careful to clarify thi

I agree that the demo is mixing concepts (esp.: relies on actual gravity to work) and tempting confusion. As a teacher, you'd have to be careful to clarify this so you don't confuse students. I note that he says the demo is about GR, but then uses his rig and two handfuls of marbles to explain how almost all solar system orbits go in the same direction. (He doesn't need GR for that.) In this context, the rubber-sheet model is a very nice testbed for what is in effect a demonstration of stochastic initialization with positive or negative net angular momentum determining how it plays out. I liked that a lot. It also hooks up with various strands of current research: use of numerical gravitational simulations, solar system dynamics at long time scales, and the surprising discovery of many exoplanets with retrograde motion relative to their host star. If you'll grant that that is OK, you might have really lost it when he stuck the plastic strut underneath the sheet and said that was like dark energy! So that makes me want to re-think his rig in a more granular way. It seems like one key question is: does the curvature you get with a mass in the 2d sheet have the same local properties as the curvature you get via GR in Minkowski space? Is this what is really driving the paths of objects in the simulation to be ellipses, hyperbolas, etc.? What is a notable example of some GR prediction that is NOT accounted for by his rig?