Updated: Jan 6, 2020
Almost a year ago, at the inaugural CopyleftConf in Brussels (the day after FOSDEM), I did a presentation with the somewhat provocative (and intentionally so) title "Extreme Copyleft: Boon or Bane."
The presentation was in reaction to trends that had been developing, unseen to many who are not avid readers of the license-review mailing list at the Open Source Initiative, where various entities or individuals were submitting new licenses for approval that were testing the boundaries of copyleft. A good chunk of the presentation was a graphical representation of how the boundaries had been, ever so gradually, moving from GPLv2, to OSL/AGPLv1, to GPLv3 and then AGPLv3. At the time, the two licenses generating discussion were the Server Side Public License (SSPL) and the License Zero Public License (LZPL or L0PL). Ultimately, both of those licenses were withdrawn from consideration for OSI approval, in view of significant opposition by mailing list participants. The main opposition fell along three lines: 1) suspicion of the business model that were to be used with these licenses (so-called "selling exceptions" to open source); 2) opposition to the notion that certain entities were "strip mining" open source by using open source licensed code without sharing revenues with the authors of that code; 3) belief that if copyleft were to be expanded beyond its then current boundaries, it ought to be the Free Software Foundation (originator of copyleft), or some other respected non-profit entity with a large and diverse supporter and user base, that should be doing so.
Part of my presentation was designed to alert conference participants to what was going on and to suggest that this was a developing trend that people needed to be aware of and take a position, particularly by participation on the OSI mailing list (which is dominated by a fairly small number of regular participants). As a bit of a provocation to the audience, I presented what I called the "Extreme Copyleft Public License" -- a license which was essentially designed to obligate users of the code to abide by copyleft obligations as the result of *any* exercise of the rights granted the author under copyright law. My intent was not to come up with another new license (there are plenty of those), but to have people think about what they believe copyleft to be and what, if any, boundaries constrain it. For example, does "Freedom Zero" ("The freedom to run the program as you wish, for any purpose") constrain a license which attempts to impose copyleft requirements on any and all exercises of copyright rights? I've always felt that it was inherently part of the Open Source Definition, although some approved OSI license might call into question that assumption.
This debate has been ongoing, and has recently reached a head as a result of the continued efforts of Van Lindberg, attorney for Holochain, to seek OSI approval of the Cryptographic Autonomy License. The license itself is complex, and seeks -- to a certain extent -- to impose obligations on data, in part through the vehicle of patent protection, but it has once again reopened this debate. Bradley Kuhn of the Software Freedom Conservancy -- the sponsors of CopyleftConf (coming up again in early February 2020) -- recently posted his thoughts on his concerns raised by the CAL. Bruce Perens, one of the founders of OSI, recently gave up his membership in OSI and withdrew from the mailing list because of his concerns about the direction that CAL represented.
The boundaries of copyleft will likely continue to be tested, as will be the processes for approving new open source licenses through the OSI and the criteria by which licenses are to be evaluated. Those with an interest in these questions should stay tuned -- or better yet, let their voices be known through OSI or other vehicles for letting the community know where you stand on the issues.