Hi all,
I geuss this is a pretty basic question. But I am usually working with Mesh type objects. So my understanding of (poly)surfaces is not great.
So for example why does the Patch command create a surface with a grid of control points, while Loft creates a surface with controlpoints only at the original curves? I am confused since both are labeled as ‘Surface’ in the properties, but they behave very differently.
What are the possibilities to add control points to a lofted surface?
Short answer is that the Level 1 and 2 Rhino training linked to in the help menu explain the basics of this thoroughly.
You are referring to over 600 pages of study material.
That is not a short answer at all 
Well I would strongly recommend going through it.
Here’s a short version you can use as an introduction to these important concepts:
Rhino uses NURBS for surfaces. NURBS stands for Non-Uniform Rational B-Splines. A NURBS surface is parametric with the parameters usually represented as u and v. Think of the parmeters as providing a (u,v) address for each point on the surface. In each of the parametric directions a surface has an order and a number of control points. The minimum number of control points is the degree + 1.
A polysurface is two or more surfaces joined together. When the surfaces are joined Rhino treats the polysurface as a single object for most operations. However the characteristics of the individual surfaces do not change.
Different commands use different algorithms to create surfaces.
InsertControlPoint and InsertKink can be used to add control points to a surface.
InsertControlPoint will almost always change the shape of the surface (degree > 1) but will not change the locations of the existing control points.
InsertKnot will not change the shape of the surface but will adjust the locations of existing control points so that the surface shape does not change.
