A scientific basis for Open Source Software

Though it may require much more of it than is obvious at than at first guess. For example, sometimes the only viable path to show that something is broken is to fix it and then let others validate for themselves that the fixed version works better. A license which allows you to look but not touch isn't sufficient.

Ultimately, I think the real underlying requirement is that the software itself also be part of the continuing scientific dialog— but this isn't reasonably possible without pretty much the full DFSG. Consider— without the freedom to use the results of your modifications commercial, you'll spend time reinventing the wheel instead of furthering the art from the existing tools. — but sure, access to the code is an essential improvement.