Dave writes: "Does it seem to your readers that it's often more efficient to create separate families than try to make one that "flexes" correctly to make all the possible types we need? It seems that way to me at times."
So what do my readers think? Post a reply in a comment and let Dave know.
I do remember an interesting discussion at one of AU's Un-Conferences. A few of us remarked that we were seeing a trend away from trying to create super-families. I can relate to Dave's comment, which seems, to me, like voicing some specific frustration perhaps. I think that it is natural to be inclined toward building very parametric families. As soon as we see what is possible, why not?
However, as fast as some design solutions change a single solution family might get built quicker than the one we make trying to anticipate how it might change. If the family is rendered irrelevant as soon as the design changes then investing a lot of time in making it parametric may not make sense. This is where our intuition and experience make a big difference. These will help us decide when we know enough, are confident enough that it is worth putting the effort in.
Keep in mind that a family can evolve. It can start out simple, singular and then grow into a much more parametric version as the design evolves, settles down. I see designers focus on the big picture and then seconds later fret over the smallest detail. That's the nature of design, back and forth from macro to micro. It's hard to be general and at the same time be highly specific. That's the tension we are dealing with, transitioning to Revit.