Hey Greg,
Let me share my take on modeling work, not defending Rhino, but just an approach that works well in Rhino, also in Alias, NX advanced surfacing, or Catia (Ive never used Catia, but based on peers experiences).
In most of our work we use patches made out of 2 clean curves, usually single span if we can, with matching degree, point count and relative position of those curve points along the surface. We use those as inputs to make surfaces with the EdgeSrf command. Then we ChangeDegree on the cross (V) direction to degree 3 or 5 and that adds more points evenly. Then we ‘curve’ that direction by using the MoveUVN interface. Along with Smooth (with locking boundaries checked) to get clean and smooth curvature. Then tools like BlendSrf and MatchSrf also help for neighboring patches continuity. The new option in V6 for BlendSrf at creation and EndBulge after the fact editing are also great. And history, in a few instances is also helpful.
We also do a lot of early exploration in SubD, not in Rhino, and then import that fast, cool but bumpy stuff in to do the process above described in Rhino for clean rebuilds. For the last decade or so we do most of our explorations in SubD. Nurbs is a complete waste of time for this, and I think it’s only used by people who haven’t ventured into SubD to expand their skills, or who do not need to do fast iterations and explorations and focus only on the one-shot final/cleaner output of Nurbs.
We are designers, and we really obsess about form, and I can’t even think how we could get any work done in a solid modeling app, even the ones with some limited surfacing like Fusion360, which we use for some internal nick-knacks and for import/export to client files. F360 does have more capability with Tsplines but I personally find it slow to work with, unreliable and very ugly in its end results.
I can’t imagine using a solid modeler more intensively for our work. For example: Modeling with fillets is a total hack in my opinion, ok to do internal mechanical components, something fast, or for a quick 3D preview of stuff to be rebuilt later. Never acceptable to do something visually pleasing, or to release for tooling. Same goes about the standard solid modeling tools like extruded, lofts, and Booleans. It all looks like a Solidworks tutorial to me.
Rhino is in a class so above all the simplistic (but robust) B-Rep solid modeling packages, that unless you can embrace their optimal modeling approach you will think it’s a toy. Most people using solid modeling think that way, but that’s because they don’t know what they don’t know, and because they are happy modeling and editing numerically simplistic prismatic shapes or questionably acceptable topology when they try to get fancy. I also think that for many tasks, parts, and industries the level of modeling you can achieve in a solid modeler is just fine. So I don’t want to sound elitist. They do very different things for a very different audience of users and clients.
I hope this helps, maybe try some of the tools and techniques I mentioned and see if the expand your horizons a bit more. The problem with these workflows is that they are hard to do and mostly undocumented. And like someone at McNeel once told me: “The people who know how to do the stuff are too busy to share it”. I feel guilty of this, and at least I wanted to chime-in with a few pointers while I’m laying on my couch taking a break from ‘doing the thing’.
Best,
Gustavo
PS: Dear McNeel people, do not misinterpret this post as praise. I’m just thinking out loud. We’ll be back to your regular beatings on Monday 
