I agree. I think a similar situation obtains in academia. Economically, reusing the existing academic infrastructure to fund FOSS development would be a perfect

I agree. I think a similar situation obtains in academia. Economically, reusing the existing academic infrastructure to fund FOSS development would be a perfect fit, because both FOSS and research are "public goods". But the success criterion for research involves novelty. Some funding for FOSS skates by nevertheless but it's an uphill battle, due to the problems you mention (but even worse, as a production-quality implementation of something that someone else has already published a demo/prototype of is less novel than an 'original research result').