Really? So the Linux Kernel, Postgres, etc. haven't been able to scale? The only barrier that has been preventing an Open Source, volunteer run, approach from working in the physical world vs the software world has been the higher cost of entry due to materials and the machines you need to manipulate those materials. It's been pretty clear that those costs are coming down, both with 3d printing and electronics (raspberry pi, arduino, etc.) Volunteer time is just a function of how many committed people you can get together, software has proven it's most definitely possible to do this on a massive scale.

The bulk of work done on the linux kernel is done by people paid to work on the linux kernel.

And companies that want to take advantage of freely available volunteer run hardware/physical projects wouldn't want to commit their own resources to improve them for their needs? I don't see anything about this model that wouldn't directly translate into these sorts of projects. Why exactly couldn't physically based open source projects have both paid and non-paid volunteers, just like software projects do?

Yes. Today. This was not always the case.

Consider this; at what point was the first person hired to work on the kernel? At what point were there a majority of patches submitted by paid workers?

^And then consider that up until that point, it was all unpaid. Just... think for a while, just think please.

This is true today, but it has not always been the case, and the FOSS ecosystem goes far beyond Linux.

Publicly funded research may also contribute to open source, 3d printed prosthetics.

None of those are all volunteer organizations. There are plenty of old mostly volunteer organizations and a few large but short term ones. But, without a core of paid workers organizations don't last.