Stallman on Ogg Vorbis using permissive licence

That appears to be what they are doing with the schroedinger project -- it is released under both GPL and MIT. Not so with the dirac-research project, which is released only under reciprocal licences.

Now that job and coriordan have pointed out these strategic considerations, it seems like they probably chose to do it this way because of these strategic considerations: they want the schroedinger implementation to be re-used in proprietary products for better market penetration of the codec itself, while they want the dirac-research implementation to have enforced reciprocality so that new innovations have to be open.

Just a guess.