Rug pulls, forks, and open-source feudalism

I emailed Stallman about the ethics of using AGPLv3 with a CLA to allow selling exceptions. Here's his reply:

https://news.ycombinator.com/item?id=42601846

  I see what you mean.  The original developer can engage
  in a practice that blocks coopertation.

  By contrast, using some other license, such as the ordinary GPL,
  would permitt ANY user of the program to engage in that practice.
  In a perverse sense that could seem more fair, but I think it
  is also more harmful.

  On balance, using the AGPL is better.

I'm confused. His answer doesn't seem to be about the CLA.

What I understand from RMS's answer is that he recognizes that the solution of using the AGPL + a CLA for allowing to sell exceptions creates an imbalance/unfair situation where only one entity can engage in the activity of selling exceptions, but ultimately he cares more about user freedom, and finds the solution of using the GPL to avoid the imbalance worse because anynody could modify the SaaS code without redistributing it, which means that the users freedoms are not respected. Basically, the GPL doesn't protect much the SaaS code and is somewhat similar to a permissive license in this setting. The AGPL protects the code better.

RMS doesn't like the GPL for SaaS software for exactly the same reason they created the AGPL in the first place, and developer inconvenience is less of a concern than potential user freedom breach.

The entity to which the exception is sold could itself close the software for its own users. But so could it if the code was released under a permissive licenses and this is, critically, why RMS finds this acceptable: he doesn't want to consider releasing software under permissive licenses unethical. This is a limit he doesn't want to cross. After all, one can't be blamed for all the sins in the world and it's the company closing the code that would be doing non-free software, not the original authors.

AGPL+CLA doesn't enable more cases of users losing freedom than a permissive license, so this is okay for RMS.

Now, it is a view strictly focused in terms of user freedom outcome and that's probably how anything RMS says should be interpreted by default. Nothing prevents you from considering that there are other aspects to consider and that the imbalance AGPL+CLA creates is unacceptable.

On a side note, it makes me think of the Qt business model.

> On a side note, it makes me think of the Qt business model.

That's the major example Stallman cited in his article about selling exceptions to the GPL.

https://www.gnu.org/philosophy/selling-exceptions.html

> The KDE desktop was developed in the 90s based on the Qt library. Qt was proprietary software, and TrollTech charged for permission to embed it in proprietary applications. TrollTech allowed gratis use of Qt in free applications, but this did not make it free/libre software. Completely free operating systems therefore could not include Qt, so they could not use KDE either.

> In 1998, the management of TrollTech recognized that they could make Qt free software and continue charging for permission to embed it in proprietary software. I do not recall whether the suggestion came from me, but I certainly was happy to see the change, which made it possible to use Qt and thus KDE in the free software world.

> Initially, they used their own license, the Q Public License (QPL)—quite restrictive as free software licenses go, and incompatible with the GNU GPL. Later they switched to the GNU GPL; I think I had explained to them that it would work for the purpose.