When the guy was talking about structural identity, starting at about 8:53, he mentioned how the swiftui needs to guarantee that the two views can't swap places, and it does this by looking at the views type structure. It guarantees that the true view will always be an A, and the false view will always be a B.
Not sure exactly what he means because views can't "swap places" like dogs. Why isn't just knowing that some View is shown in true, and another is shown in false, enough for its identity? e.g. The identity could be "The view on true" vs "The view on false", same as his example with "The dog on the left" vs "The dog on the right"