Mandatory upstream distribution and its problems

Posted Dec 16, 2020 1:31 UTC (Wed) by bkuhn (subscriber, #58642)
In reply to: Changing CentOS in mid-stream by pabs
Parent article: Changing CentOS in mid-stream

The issue of mandatory upstream distribution of source changes to copylefted software has been a hotly debated topic since I first got involved in copyleft licensing policy and drafting in the late 1990s. The two primary positions within this debate have been outlined well already in this thread.

I think most copyleft theorists believe that modifiers (modulo the danger situations with dissidents that pabs identified) have a moral obligation to make generally useful improvements available to the general public. The problem, and this goes back to the points pabs raised, is how to draft policy that respects the privacy rights of some types of modifications while assuring that truly generally useful changes that the world deserves, morally speaking, can be codified into a copyleft license.

Much of this debate reminds me more recently of the so-called “Ethical License” initiatives. In those initiatives, its proponents argue that since there are immoral things done by companies with software (such as helping USA's ICE lock children up in cages) that the licenses are not ethical if they don't prevent that.

The flaw in the logic is the idea that every moral and social problem that relates to software can be fixed with its license. Having spent years enforcing copyleft licenses and drafting them, I've concluded that you have to be selective how you use the tool of copyleft. And, it's not that I'm just critical of new ideas: I'd argue that GPLv2§2(a) is a great example of a well-intentioned clause meaning to get to a good policy outcome that is just annoying to comply with in practice. We should all comply with it because it's required, but I would never include that clause or one like it again in a future copyleft license.

Similarly, when I hear people clamoring for features for copyleft licenses, I worry about future-proofing. It's hard to do. Anyway, I encourage anyone who wants to talk about real-world drafting of copyleft to join the copyleft-next project.