Vasters is correct in being skeptical about earning a living off free software; while it can be done, it is not as reliable as it could be. Vasters agrees that free software would be good for society, but sadly concludes that producing it is too impractical in today's world.

But it doesn't have to be. Society already has established institutions for paying people to spend their lives producing information for the public good. The university system.

Free software should be a professional specialty within academia. There should be computer science professors and phD students whose mandate is not to discover only original results, but rather to implement known results well in the form of free software.

Structurally, the economics of free software production are identical to those of science. Both free software and scientific knowledge are "public goods". Once one person produces an piece of information, the marginal cost of reproducing the information is zero, and the knowledge becomes freely available to all. That is why the government finds it worthwhile to give money for the production of scientific knowledge.

Like science, the production of free software is a collaborative venture where publication is central. Like science, the system functions best when individual producers have a substantial amount of creative freedom to choose which projects to work on. Like science, it would be impractical for practitioners to do what is needed on their own, or in their spare time.

It is inefficient for society to reinvent the wheel in terms of funding free software. We shouldn't try to duplicate the university system by creating a separate ecology of free software organizations funding developers, and free software foundations making grants. Free software developers should be employed as professors in the university system.

I predict that the first university to appoint a professor of open source will quickly generate a flurry of publicity, and will, if they choose, become a hub for open source development. Such a move would be immensely attractive to students, as many computer science students intend to be industry programmers rather than academics. These students would be better served by more hands-on experience as a real team members in a real open-source project, as opposed to the current system of spending all their time on theory and personal projects. The first universities to appoint professors of open source will see their the quality and quantity of their applicants go way up. A second-tier university might be instantly transformed into a first-tier one, at least as far as the desirability of the computer science department is concerned.

This won't happen on its own. Although the economics are similar, free software is not research, and there is much competition in universities over what sort of professor to give open spots to. What is needed is a campaign to raise enough money to endow a new chair of computer science. This will be expensive (about a million dollars at the cheaper places). But if we as a community raised 100,000 EUR in just seven weeks to buy Blender, then surely we can raise ten times that amount over the course of a few years. We won't have to keep doing this indefinitely. After there are a couple of free-software professors, the movement will take on momentum in the academic community, and other universities will begin appointing professors with "free software" duties on their own in order to remain competitive.

I am considering starting a fund to collect donations for the endowment of a Chair of Free Software at some university. For the sake of compromise, I think it would be wise to allow the university to require that, beyond teaching, 50% of the professor's time be devoted to traditional computer science research, and 50% to free software efforts. Please let me know if you would be interested; my email can be found on my web page.

(from another place i wrote about this:)

Another aspect to this issue is the funding of open source. Open source, it seems to me, is a "public good", very similar economically to scientific knowledge. Funding it is a tricky problem, because it benefits society as a whole, but only if it is freely available to the public. We already have a system for dealing with this sort of thing; part of our taxes go to paying professors, who produce the "public good" for the public. It seems to me it would be most logical to use the existing infrastructure to fund open source; i.e. those who want to program open source full time should be able to become professors of open source (or, I guess, computer science, specializing in open source).

Of course, right now the obstacle to this is that ProfessorsDontCode? (or rather, professors are not given tenure based on actual usable projects that they produce, but rather based on the excitingness of their research).

I feel, however, that this could change with a little help. I would like for someone (maybe I'll do it later) to start a fund with the goal of endowing a Professor of Open Source at a major university. I think we need in between $500K and $2 million to endow a chair. If we raise money at the same rate as the Free Blender project, then we should be able to do this in a couple of years.

-- BayleShanks

(and another writeup:)

I see programming as economically similar to scientific research; it is expensive to do, but once done, the marginal cost of reproducing the final product (information: knowledge in one case, programs in another) is negligable. Government funding of researchers is the system which produces most of scientific knowledge. I think the same should be applied to software. I am suggesting that programming become a function of the university system, and that we should have tenured faculty whose promotion is based not on the production of original, interesting knowledge, but rather on the production of useful software code.

Just as the knowledge produced by government funds is freely available for private industry to make use of (at least, it was, until the recent sad trend of universities patenting government-funding discoveries), the code produced should be public domain.

Some details: