Dan Blanchard relicensed chardet from LGPL to MIT after using Claude to reimplement it from scratch, claiming the result is an independent new work. The original author objected, arguing this violates the spirit of copyleft. Armin Ronacher and Salvatore Sanfilippo (antirez) both defended the relicensing on legal grounds. This essay argues both conflate legal permissibility with social legitimacy, ignore the directional difference between the GNU/UNIX precedent (proprietary→free) and this case (copyleft→permissive), and speak from positions of privilege that align with their conclusions. The author contends copyleft becomes more necessary, not less, as AI lowers reimplementation costs, and proposes extending copyleft to cover API specifications and test suites as the next evolution of open source licensing norms.
Table of contents
The analogy points the wrong wayDoes the GPL work against sharing?A self-refuting exampleLegality and social legitimacy are different registersWhose perspective is the default?What this fight points towardSort: