I've come up with two tests which can be applied to issues like the file format discussion in order to reach the determinations which I believe to be most correct.
==Two Tests for the freeness of activities related to project content==
===Impoverished/principled reader test===
Imagine people who are sufficiently impoverished that they can only afford zero cost software or who are sufficiently concerned about their freedom that they only use legally licensed Free Software. Does taking the action discriminate against these people? Does taking the action give them a materially lesser experience?
===Impoverished/principled author test===
Imagine a collection of authors or publishers who are sufficiently impoverished that they can only afford zero cost software or who are sufficiently concerned about their freedom that they only use and distribute legally licensed Free Software. Would they be able to take the same or materially equivalent action related to their own content?
=== Examples ===
====Using a free software flash module to support clients which can't handle the HTML <canvas> tag====
This passes the impoverished reader test: The impoverished reader can be given the <canvas> tag which has equal or better functionality OR the reader could legally use Gnash.
This passes the impoverished author test: The impoverished author could legally perform the exact same action while paying no fees nor using/distributing any software which was not freely licensed.
====Parallel distribution of Video in both Ogg/Theora and Flash Video because Flash video is more widely adopted====
I think this passes the impoverished reader test: The impoverished reader can view the Ogg/Theora file, and the differences in quality/bitrate are neither likely to be material nor were contributing factors in the decision to offer Flash Video.
However, this clearly *fails* the impoverished author test: The impoverished author can not legally engage in parallel distribution himself without paying codec licensing for encoders and fees for the distribution of material in the licensed format. The author could distribute exclusively Ogg/Theora, but that wouldn't be equivalent because it has significantly less adoption (and that was the reason to consider parallel distribution in the first case).
====Parallel distribution of hypertext in an eBook format where only *reading* tools were non-free, and some free ebook formats====
What if a format is totally free to authors/publishers but isn't useful without a non-free reader? I believe this has been for some ebook formats, at some points in time. We'll presume that the free format isn't materially worse than the non-free one.
This passes the impoverished reader test: The impoverished reader can use the free ebook format.
This passes the impoverished author test: He's free to turn out ebooks in both formats just like we are.
It was re-reading some of Geni's posts that made me think of describing this "impoverished author" test. Much of Wikimedia's mission isn't merely providing read-only content at no cost to the public, it's also a mission of enabling authorship by building a collection of works that others can build upon. It's isn't good enough that our content be available to free software users, it also must be free for authors to emulate, modify, and/or republish. As such, both tests are equally important.
I wouldn't presume to apply these tests to things unrelated to the content (all that stuff our users are creating and posting in the projects) such as Wikimedia office activities: Our mission is one of enabling the world through freely licensed educational materials, not the creation/promotion of freely licensed office materials. (Although there are practical and ethical reasons outside of the Wikimedia mission why preferring freely licensed solutions is generally good...)
Thoughts? Holes? Better restatements?