I am going to be annoying and answer your question with a question: consensus among who? How do make a decision like this?
On the one hand, you have Wikimedia users, who don't really care about the appearance of promoting FOSS or not.
[Citation needed]. User's aren't one person, but quite a varied group. However I believe a lot of them care about promoting FOSS. Just look at the MP4 RFC. Obviously not all Wikimedia users voted in that (especially if you're including non-editors in the user category), but a significant chunk did.
After all, we are talking about a project, where people, out of the goodness of the heart spend their time writing high quality educational articles without compensation. I would imagine that a good portion (not all) of the types of people attracted to doing this are also the type of people who would be attracted to the ideals of the FOSS movements.
They just want to things to work and look good.
Well yes. Don't we all?
They also believe that the only consensus that matters is the one on their local wiki.
To be fair, they don't really have an opportunity to participate in a larger consensus.
On the other hand, we have MediaWiki developers, some of whom would rather throw up their hands and not specify a real font stack,
Well that's an antagonistic way of phrasing it.
By the same count, MediaWiki developers would rather throw up their hands then make #expr accept localized numbers (bug 30318), or any other of the thousands of WONTFIXed bugs.
rather than touch the non-free fonts *most users already have* with a ten foot pole.
Whether or not the non-free fonts are installed on the users system seems irrelavent to the concerns that the non-free font crowd have raised.
There's a big difference between saying "Do whatever you think is best" compared to "Do X if possible", even if both statements evaluate to doing the same action.
They seem to think that an RFC or discussion on Wikitech-l is what represents a consensus that must be respected.
I'm not sure if people actually think that (or if they do think that, that they actually said that). One person (MZMcBride) implied that kind of heavily, most of the other "revert"-ish threads are not using the "because concensus (or lack thereof) on wikitech-l" as an argument.
There is a historical precedent for development decisions being made not by consensus, but by a developer hierarchy (e.g. people like Brion or Tim making final calls). Now a days it seems like things have shifted more to managers at WMF make the decisions. How controversial decisions are decided in our community is a complex topic unto itself. I'm not even sure what the current best practise is, to be honest.
And they don't feel responsible for what gets released on Wikimedia sites necessarily.
I strongly dispute that statement. If they didn't care, this thread wouldn't be here.
In essence, the controversy is a bunch of people saying: I feel responsible for what gets released on Wikimedia sites, and I believe that the problems of TR don't outweight the benefits, or that it otherwise doesn't meet the high standard of what normally gets released, etc
We can gain more consistent, accessible typography across languages with an iterative approach that continues to build on what we've done over the last five months. Or we can go back to the drawing board to try and please everyone, which is an impossibility if you ever want to make progress.
Sorry to put this so bluntly, but to be frank, its debatable whether this actually represents "progress".
--bawolff