Thinking about it, the discussion was about micro-poly displacement, something I always associate with subdivision surfaces.
He showed that the polygon smooth thingie sort of approximates sub-d surfaces, but it's nowhere near the same idea.