As more evidence has come to light, from YouTube and re-enactors, I’m getting more convinced that collisions could be quite common (at least when one side didn’t run away first). Although I’ve been heavily influenced by John Keegan and Frank Tallett, maybe their positions were too extreme in the other direction in saying that knee-to-knee formations and head-on collisions could never happen at all. However, I still don’t think that a fast close-order charge to contact gave the decisive advantage that Frank Jones argued for. In most cases a collision would either have a negligible effect or be counter-productive (at the speeds he described it would be very counter-productive).

From what you’ve described I’d guess that your experience of collisions is somewhere between the second and third videos I linked to above. In that case there might be an advantage if one side were all very good horsemen with deep seats who could stay on while the other side were poor horsemen who fell off very easily. But that seems like a bit of a gamble when you don’t know the enemy’s capabilities. Presumably using the disorder to fix and flank the enemy depends on outnumbering them so you have some spare troopers to take advantage of the disorder.

As I’ve said before, I think the most important thing here is psychology. The fear of being hit by an enemy horse probably overrides the knowledge that the collision is likely to be equally bad for both parties. I need to look more closely at folk physics to see how people think/thought about collisions between horses. At the time of the civil wars Newton hadn’t even written his laws of motion, and highly educated people believed all sorts of strange things about the humoural body and the chain of being. Even now the implications of Newton’s third law for cavalry charges don’t seem to be very widely understood.