So in this case there appears to be a bug in the boolean handling where it's keeping the wrong surface piece.

This kind of ghost surface piece here:



When you turn on its surface control points, the control points for the actual surface are way up here:



So the trimming boundary is totally off of the surface - the trimming boundary itself is in the right spot for the boolean result but the boolean seems to have decided to keep the wrong surface piece as the underlying surface part of the face.

This is a simple enough case that I may have a good shot at tracking it down and fixing it, I'll give it a try in a bit.

- Michael