> how would we make money from open source software
There are many existing examples, so this is clearly a solved problem already. You charge for support, or for feature requests, and so on. That's how SUSE and RedHat make their money.
The flaw with looking at proprietary software's monitisation is that it usually just boils down to "pay for the binary". This obviously won't work with free software, you need to charge for development rather than access (though you can also use a seat-based model where you only provide support for machines that have valid licenses).
SUSE or RedHat are excellent examples when it comes to monetizing OSS for large businesses. What about consumer software? I do not think OSS will survive monetization when it comes to dealing with individual consumers.
And then there's the question of SaaS. OSS exists but a lot of high quality alternatives are paid. I don't think services like Todoist, Pocket, Evernote etc would exists on the open source model you described.
This is only half the story. Several business models around open source are working and proven (e.g. hardware vendors writing kernel drivers) but there is also a lot of unpaid volunteer work that everyone seems to take for granted. GIMP is one example. And probably a large number of the packages on NPM. Or think about OpenSSL, everyone was just assuming it must be well-funded, while nobody was actually funding it.
That definately is one solution, but perhaps made possible by restricting distribution of software in the first place to get a leg up (see comments regarding improvements and distribution restrictions regarding installer scripts in SLS http://www.linuxjournal.com/article/2750). This suggests that it isn't a solved problem, because the initial conversion to an open source model (or free software) with support on the side may have required a different model to start the venture.
None the less, it's admirable, and hopefully a net benefit for everyone.
My point was more about Stallman and co calling foul with regards to software freedom, codifying their own ideal, but not giving directions to reach that ideal. This feels like a safe pulpit to sit upon, where their view isn't falsifiable, useful when they want to say "I told you so", and eventually taking all credit for everyone else's efforts in between to make the end goal possible.
> perhaps made possible by restricting distribution of software in the first place to get a leg up
This is incorrect, you can download the full ISOs for SLE from the SUSE website, with 30 days worth of updates. The source code (and the system required to build it) is all publicly available on https://buid.opensuse.org/. I beleive RedHat have something similar.
I'm also not sure that an interview from 1994 with the creator of Slackware is a good indicator of the current state of distribution business models. Though even in 1994, both RedHat and SUSE were selling enterprise distributions.
Since the past determines the future, the relevant part of the 1994 interview is "... Instead, he claimed distribution rights on the Slackware install scripts since they were derived from ones included in SLS...", which as I understand it, is the restriction of distribution I was referring to (perhaps redistribution is more accurate).
This suggests that the business model benefited from restricting redistribution and modification of the source code, so breaks the assumptions that the business model was purely based on making money from open source, and so doesn't fully support the idea that proprietary software is unnecessary, in the case where we take SUSE as an example of saying it is "already solved".
There are many existing examples, so this is clearly a solved problem already. You charge for support, or for feature requests, and so on. That's how SUSE and RedHat make their money.
The flaw with looking at proprietary software's monitisation is that it usually just boils down to "pay for the binary". This obviously won't work with free software, you need to charge for development rather than access (though you can also use a seat-based model where you only provide support for machines that have valid licenses).
(I work for SUSE.)