[JoeAltmaier]

Each surface polygon is flat. They can be 'inflated' via the OPs technique without violating the bound of an enclosing sphere, right? Each recursive expansion has an inflation factor that scales. Hm. But the sphereical section bounding each polygon doesn't scale, it becomes 'flatter' as you recurse. So there's a limit.

[ehartsuyker]

Actually, not. The definition of convex is that given a point A and a point B and a line between A and B, all points on the line AB are in the interior space of the solid.

Inflating two adjacent surfaces creates a valley along the pre-existing edge between the two of them and fails the above definition.

[JoeAltmaier]

Yet that's what the OP describe. Remember, the edge was a "mountain" to begin with, you have some wiggle room. That's the observation that the whole paper is based upon.