that talk made it really clear how tangent-point energy is easier to compute than mobius energy but not why it's a good proxy for it, it seems like a semihack