I've always seen the acceptance rate as a kind of warning light for question quagmires, rather than an absolute measure of a user's cooperation.
I've had lots of accepts from users with 30-70% accept rates, which is below the "green" level. Sometimes these people really just don't get good answers or any answers, often because their work takes them into obscure or unpopular areas (SQL-CLR is one obvious example that comes to mind). It's not a big deal. Even if the question author is really a flake, and even if you're just doin' it for the rep, two upvotes are worth more than a single acceptance anyway.
The accept rate does come in handy when I open a question and it turns out to be a very vague, fragmented, or just poorly-written question. When I see the 0% accept rate on those questions, I tend to avoid them, because it usually means that either (a) they don't respond to comments or requests for clarification, or (b) they post "chameleon"-style troubleshooting questions and other people have given up on them.
The common theme in both cases is, I need to see the actual question before the accept rate means anything. If it's a good question then the history is irrelevant. Accept rate only helps me decide, if it's a bad question, whether I should spend any time trying to help the author improve it, or whether I should just downvote it and be on my merry way.
So in this case I'm in favour of the status quo: only show the accept rate when you're actually looking at a question. It shouldn't matter outside of that context.