A post over at Pharyngula finally pushed me over the edge into accepting the claim that agnosticism-ness is orthogonal to atheism-ness, i.e. whether or not a person is agnostic is independent of whether or not they are an atheist.
I have understood this claim for some time: those who are assert it are defining agnosticism as being about knowability, while atheism is about presumptive belief1. But I was hesitant to accept it, because it seemed to me that this was not the definition by which most people understood agnosticism.
I have changed my mind. And the reason is because I do not think there is any other useful definition of agnostic.
Using the orthogonal definitions, which I now believe to be the only acceptable definitions, very few atheists are not also agnostics. A non-agnostic atheist would be the rare bird who says something along the lines of, "Your claims may be unfalsifiable, but I have received personal revelation which says they are false!" Yeah, I don't know anybody like that either. Virtually all atheists explicitly reject falsifiable gods (like Thor or Yahweh) and tentatively reject unfalsifiable gods, while recognizing the by-definition unknowability of the verity of unfalsifiable claims.
The only alternative definition I can think of is one in which an agnostic not only insists that unfalsifiable claims are unknowable, but furthermore asserts that the odds of any given unfalsifiable claim being true are close to 50/50. First of all, I doubt many people actually think that. And second of all, if anybody does think that, they are grade A stupid.
I suppose we could also define it such that an agnostic recognizes the unfalsifiability of theistic claims, and at that point immediately stops pondering the question any further. But we've already got a word for that: "incurious".
1I insert the word "presumptive" to make it clear that I am talking about the "I believe there are five beers remaining in my fridge" type of belief, not the "I believe in Jesus" or "I believe emacs is better than vi" type of belief. (though emacs is indubitably better than vi, but I digress...) If I looked in my fridge and saw six beers, I would feel comfortable revising my estimate. On the other hand, if (hypothetically speaking) someone showed me scientific evidence that vi was superior to emacs, I would be forced to assume some supernatural explanation, e.g. that Satan, as a well-known vi user, tampered with the evidence to fool us into thinking vi was better, even though the man pages tell us otherwise.