Over on CIDU, there's a lively discussion about the definition of an atheist and the definition of an agnostic.
There's a posting by "mitch4" that seems useful, at least in defining what the semantic argument is all about:
Some possible belief positions:
a. "I think we cannot have certain knowledge of either P or not-P".
b. "I am certain that [I know that] P."
c. "I am certain that [I know that] not-P."
d. "My view is [I think that] [I believe that] P"
e. "My view is [I think that] [I believe that] not-P"
f. "I do not hold the view [do not think that] [do not believe] that P."
Some observations of the formal relations of the above:
1) a. contradicts b.
2) a. contradicts c.
3) a. is consistent with d.
4) a. is consistent with e.
5) e. implies f.
6) f. does not imply e.
7) a. is consistent with f.
A number of people (and I would join them) have asserted a. and have self-described that part of their belief system as "agnostic".
A number of people (and I would join them) have asserted either e. or f. and have self-described that part of their belief system as "atheist".
On the basis of formal observation 4 (or 7) those people maintain that they are not inconsistent; and further that they can be both atheist and agnostic.
Detcord and perhaps others think that the term "atheist" requires not just e. or f. but the stronger c. Then on the basis of formal observation 2, they take the terms "atheist" and "agnostic" to be inconsistent.
Neither Detcord nor I nor Google nor Webster can really determine whether e, or f. or the stronger c. is the ruling sense of "atheist". So we can't settle whether "atheist" and "agnostic" are consistent or inconsistent.
But we can easily affirm -- it's just logic! -- that a. and e. are consistent, and that a. and f. are consistent. And that people asserting either of those pairs should be found not-guilty of contradicting themselves -- regardless of the terminology they overlay on those specs.