See, that's exactly what you shouldn't do if you want to encourage academics to release the source code.

Academic code sometimes only has been made to work on one machine (the author's). Someone reproducing the results should still have access to the source, even if they have to hack it to work on their system.