I agree the FreeCAD UX could use some improvements, but I think even more important would be improving reliability and feature coverage in the OpenCascade CAD geometry library it uses. Slightly more complex things like fillets, lofts, nurbs surfaces are very limited and unreliable today. Mixing any of these with booleans tends to create more trouble.
Blender on the other hand is very solid in the modeling department, in my experience. The mesh based approach certainly helps, but not depending on an external organization for the core geometry functionality does not hurt either.
>> but I think even more important would be improving reliability and feature coverage in the OpenCascade CAD geometry library it uses
I've been wondering what portion of the FreeCAD "problems" people report are actually issues with OpenCascade. I don't have time to deal with any of it, just a question that pops into my head when I see complaints about FreeCAD. When I have time I spend it in Solvespace, which is decades behind commercial CAD but is so much more fun.
Completely agree with this, OpenCascade is just not a good enough foundation to build a CAD package at the same level as SolidWorks or Fusion360. The unfortunate thing is that the level of investment to build a kernel of that quality is in the 10s or millions dollars. Without a corporate backer who wants it, it won't happen.
The only way I could see it happening is if a consortium of large companies decided that they wanted to drop the big players and build an open source one. It needs a visionary in the position to decide to do it.
I've been using various CAD programs, and IMO, FreeCAD UI/UX is quite ok. If you are Inventor or Solidworks user and try to switch to SolidEdge for example, it won't be easy. Not to mention some others like VariCAD (~30 years old professional MCAD)
There are many issues with FreeCAD being OSS alternative to professional MCAD software, but UI/UX is not one of them
Blender on the other hand is very solid in the modeling department, in my experience. The mesh based approach certainly helps, but not depending on an external organization for the core geometry functionality does not hurt either.