It's an interesting case, but I'm skeptical of SFC's assertion that this case "could have serious repercussions for software freedom".
The main point of the "further restrictions" clauses in GPLv3 is virality. If you receive someone else's copyrighted program under the GPLv3, modify or add things, then license the new program to someone else, the new program has to be licensed under the GPLv3. You can't add additional restrictions.
But in this case, Neo4j was the copyright owner for the entire program. Everyone agrees it had the right to relicense the program under whatever license it wanted. Everyone agrees it could have avoided the GPL entirely and written its own non-free license.
It just so happened that Neo4j instead chose to take the AGPLv3 and tack a non-free clause on the end of it. SFC's argument is that this counts as an "additional restriction" that can be removed. The district court disagreed.
As far as I can tell, the district court's ruling on the matter ([1], quoting [2]) was very brief and gave the issue little consideration. I think SFC may have a point that the court misinterpreted section 7. But even if the ruling is upheld on appeal, I'm not convinced the precedent would have much impact on any future cases where it actually matters, where the person adding the "additional restriction" is not the copyright holder for the entire program. For instance, the ruling talks about "the form AGPLv3", which seems to be a reference to the term "form contract", emphasizing that for the software in question, AGPLv3 was merely a basis for Neo4J's new license, not something legally binding in its own right. Likewise, the ruling's reference to "Neo4J Sweden’s exclusive right to license its copyrighted software under terms of its choosing" seems inapposite to the case of a derivative work with multiple copyright holders.
Also, just for the record, even if the "you may remove" clause were read out of the contract entirely, it wouldn't neuter the GPL or anything. Distributing a derivative work under the wrong license would still be a copyright violation. It would just remove one of the options for addressing such violations. I can buy that such a precedent would have "serious repercussions for software freedom", but I'm just saying, it wouldn't be that bad. But again, I don't think that's the precedent that was set here.
The main point of the "further restrictions" clauses in GPLv3 is virality. If you receive someone else's copyrighted program under the GPLv3, modify or add things, then license the new program to someone else, the new program has to be licensed under the GPLv3. You can't add additional restrictions.
But in this case, Neo4j was the copyright owner for the entire program. Everyone agrees it had the right to relicense the program under whatever license it wanted. Everyone agrees it could have avoided the GPL entirely and written its own non-free license.
It just so happened that Neo4j instead chose to take the AGPLv3 and tack a non-free clause on the end of it. SFC's argument is that this counts as an "additional restriction" that can be removed. The district court disagreed.
As far as I can tell, the district court's ruling on the matter ([1], quoting [2]) was very brief and gave the issue little consideration. I think SFC may have a point that the court misinterpreted section 7. But even if the ruling is upheld on appeal, I'm not convinced the precedent would have much impact on any future cases where it actually matters, where the person adding the "additional restriction" is not the copyright holder for the entire program. For instance, the ruling talks about "the form AGPLv3", which seems to be a reference to the term "form contract", emphasizing that for the software in question, AGPLv3 was merely a basis for Neo4J's new license, not something legally binding in its own right. Likewise, the ruling's reference to "Neo4J Sweden’s exclusive right to license its copyrighted software under terms of its choosing" seems inapposite to the case of a derivative work with multiple copyright holders.
Also, just for the record, even if the "you may remove" clause were read out of the contract entirely, it wouldn't neuter the GPL or anything. Distributing a derivative work under the wrong license would still be a copyright violation. It would just remove one of the options for addressing such violations. I can buy that such a precedent would have "serious repercussions for software freedom", but I'm just saying, it wouldn't be that bad. But again, I don't think that's the precedent that was set here.
[1] https://storage.courtlistener.com/recap/gov.uscourts.cand.33...
[2] https://storage.courtlistener.com/recap/gov.uscourts.cand.34...