I've seen a lot of companies using GPL/AGPL for exactly the same purpose - to prevent other commercial companies from reusing their code. It is especially common in areas where added-value customizations are valuable - i.e., if you have GPL (or, usually, dual-licensed commercial/GPL) core for, say, document handling system, you can market it as open-source, take contributions, and build add-on business on it, but if a competitor comes out and wants to use your core, they now have to publish source for all their add-ons, so you can have them for free. That puts you at the advantageous position. It's not the unique practice, I've seen many companies doing this. Not sure that's what FSF meant to happen, but that's what is happening.
I would have thought the FSF/Stallman would be perfectly fine with this. Their main objective is to maximise the freedom of users by advocating for software that can be inspected, modified, shared etc. As far as I know, they are not opposed to someone making a private income off this kind of software.
However, it's probably true to say that a number of possible methods of extracting an income would be closed off to you under these conditions. They're closed off not because the copyright(left) is anti private income, but because they're probably user oppressive practices (ones that would be very quickly ripped out of the codebase e.g. sending user private information back to home-base for subsequent sale to advertisers etc.)
That's my layman's understanding of the situation anyhow (happy to be corrected if wrong).
Thing is, a very common model along these lines is to dual-license under copyleft for free, and proprietary commercial license for those who don't like copyleft "virality" (e.g. the way Qt used to do it until switching to LGPL).
This works great; but the proprietary part is completely antithetical to the whole "free software" concept.
I've been thinking about the dual-licensing approach. Strategically, one could argue that it still aligns with FSF goals. My logic is roughly:
(objective) We want people to use software that doesn't oppress them -> People naturally want to use the 'best' software -> Copyleft licences ensure the 'best' software is non-oppressive (via open source, freedom to modify etc.) as derivative improvements can be merged back into the mainline.
The key part is that the 'non-oppressive' software is 'the best' (or at least of decent quality) so that people will naturally use it and society as a whole will remain un-oppressed. Mightn't one extend this logic and say: Although we don't get the benefit of useful modifications being mainlined for closed-source, secondarily licensed derivatives, more money for the project means more resources/devs -> ultimately means improved mainline software?
I mean, I'm unsure if I agree with that argument, but it seems at least plausible.
I think it's a perfectly plausible argument. The problem is that it is a pragmatic argument. It is possible to be pragmatic in pursuit of one's ideals - indeed, I would argue that this approach is the one that usually works best (or at all) - but FSF, and Stallman personally, seem to frown upon short-term pragmatic deviations from ideological purity.
> if you have GPL (or, usually, dual-licensed commercial/GPL) core for, say, document handling system, you can market it as open-source, take contributions, and build add-on business on it, but if a competitor comes out and wants to use your core, they now have to publish source for all their add-ons, so you can have them for free.
This is a bit tricky to pull off unless you get a copyright assignment or a license permitting relicensing from all contributors, or your GPL licensed core has a large exception carved out of it (say, for proprietary plugins).
You can use the code without regard to GPL, since you own the copyright. Of course, contributions usually require copyright assignment in this model, via CLA of some kind.
