Post #3033890

Reading @simon's description of the AI chardet clean room relicensed implementation: https://simonwillison.net/2026/Mar/5/chardet/ One of the arguments against this being a valid reimplementation is that the person who guided the implementation (Dan Blanchard) has also been maintaining chardet for over a decade, and thus has too much knowledge of the code for it to qualify as a clean room implementation. This claim frankly offends me. Of course Dan knows all about the code, and more importantly in this case he knows all about what's being asked from this code, what purposes it serves, and the requirements for any replacement. He knows all that, AND THAT KNOWLEDGE BELONGS TO HIM. It doesn't belong to the fucking chardet code or to Mark Pilgrim.