P.Z. Myers has a truly absurd attack on a recent post by Jennifer Fulwiler. I was interested in the fact that he seemed to try to contradict a particular proposition (Fulwiler noting that her atheist friends often would conclude that Catholicism is much more fair and reasonable than they had thought when certain misconceptions were cleared up) with a particular proposition (that there are other doctrines that some atheists, not misconceiving them, consider unreasonable), when, reading through the post I suddenly realized that the latter wasn't intended to be a particular statement: almost every point Myers makes is false unless we postulate that all atheists are atheists for the same reasons and have the same views. Since the evidence that this is false is extraordinarily easy to find, it's difficult to distinguish anything in the post from outright raving. This is especially so since it's a bizarre strategy to take in response to Fulwiler's post: she is identifying things Catholics should not assume of atheists, and therefore is not committed to any substantive positive claims about atheism in general (and all her positive claims about atheists are illustrations and examples from her own case prior to her conversion and from atheist friends she's talked to). The rational response to such a thing, if one disagreed with it, would be to defend claims contrary to claims made by Fulwiler; but Myers responds with the claims (he doesn't actually provide anything in the way of arguments) opposing Fulwiler's claims by contradiction. Making such a rookie mistake in critical reasoning while calling someone "bubble-headed" is (at best) ill-considered. (One does have to smile, though, at the preciousness of the little rhetorical tricks, so transparent a baby could see through, scattered throughout Myers's post; for instance, using the phrase "she says" to insinuate, quite without evidence, of course, that she might not really have ever been an atheist.)
But this phenomenon of a certain kind of atheist talking as if atheists were a monolithic group does raise the interesting question of whether there is any viable and handy classification of the different kinds of atheists. Luke Muehlhauser once had a post titled 17 Kinds of Atheism but (1) this was a little misleading as a title because the categories could all overlap and the post was really about seven different kinds of distinctions one might make among atheists; and (2) it didn't pretend to be exhaustive. I'm not actually sure an exhaustive classification is possible, but it's worthwhile to see if there might be a more useful way of grouping atheists.
Atheism as such is simply the basic claim that "No divine being exists" (let's call this A!). Differences among atheists arise from two sources: (1) the kinds of meta-claim they make about A! and (2) the kinds of principles they accept which support A!. Relying wholly on the second would probably involve us very quickly in a very messy and complicated classification if we tried to be comprehensive, while the latter seems to allow for some relatively clean distinctions, so it makes sense to build the basic skeletal structure of the classification from the first. However, at the same time it makes sense to focus on meta-claims that tell us something about how the claim can reasonably be accepted. I would then suggest, as the very first division:
(I) A! is acceptable simply in virtue of itself.
(II) A! is acceptable in virtue of some other principles.
Now (I) can only be the case if A! somehow shows that its contradictory (let's call it T!) is unacceptable simply in virtue of itself. That is, T! would have to be nonsensical or absurd simply in itself. (I am making an assumption or two about acceptability here. One could hold that (I) is consistent with saying that T! is acceptable simply in virtue of itself, but it's difficult to see how this would work when we get down to details.) Class I atheists hold that theism is self-evidently false. This meta-claim doesn't commit them to holding that T! is gibberish, of course, only that it can be seen to be wrong just in and of itself. Likewise, it doesn't commit them to any particular position about whether T! is obviously false or subtly false; the idea is simply that T! itself is not coherent in some way.
This brings us to Class II atheists. Given that one can hold that conclusions follow from different principles in different ways, or even from the same principles in different ways, we need to focus on the strongest meta-claim that the atheist makes. Given that, I would suggest that we first divide Class II into the following divisions:
(IIa) A! follows from some principles in such a way as to be demonstrated.
(IIb) A! is not demonstrable but nonetheless follows from some principles in another legitimate way.
There are different ways of characterizing demonstration, but the one that seems most useful here is this: Demonstration is rigorous derivation of a conclusion from premises, all of which are definitely known to be true. That is, Class IIa atheists hold that A! is the conclusion of a sound argument that is known to be sound; in yet other words, Class IIa atheists think that theism definitely contradicts facts definitely known to be true.
Given that Class IIa atheists think that they have an argument known to be demonstrative, Class IIb atheists naturally fall into groups according to the following meta-claims (remember, we are not talking about every argument the atheist might make but only how far the atheist thinks his strongest reasoning can take him):
(IIb1) A! follows rigorously from premises that, taken together, are only probably true.
(IIb2) A! at most probably follows from premises that are at least probably true.
Class IIb1 atheists hold that they might have reasons that demonstratively show theism to be false, but will concede that they are making assumptions that, as far as they know, need not be made by a reasonable person in possession of all the facts (they can still hold, of course, that these assumptions are reasonable, plausible, extremely likely, etc.). And Class IIb2 atheists will concede that they don't have any sort of strict proof that T! is false, but will hold that T! is improbable for some reason.
Of course all of this involves ignoring atheists who accept A! but who can't say why A! is true (e.g., because they've never given it any thought); we can gather these atheists together under the label Class III atheists.
Luke's category of 'gnostic atheists' covers Class I, Class IIa, at least most of Class IIb1, and at least some of Class IIb2. His category of 'agnostic atheists' covers possibly at least some of Class IIb1 and probably most of Class IIb2. All the atheists considered here are 'broad atheists' in his sense; 'narrow atheist' is an obvious solecism. How his 'friendly', 'indifferent', and 'unfriendly' categories map on to these classes depends very crucially on the particular theory of justification in play. In the way these terms usually are used, however, they mean something incidental to A! in particular, since they have to do with a person's tolerance for something they deem error. For instance, Feuerbachian atheists, who hold that religion consists of misplaced ethics or moral psychology, wouldn't think that T! is justified in any real sense, and can sometimes be just as uncompromising in rejecting it; but because they hold that much of the rest of the structure of religious thought is reasonable and insightful (even though involving a confusion of two different domains), they are sometimes treated as 'friendly atheists'. The rest of Luke's distinctions are based on issues that are equally incidental to the perceived truth of T! or A!. (They might be important in certain contexts, but they would get their importance from something other than T! or A! and the reasons for accepting or rejecting them.) This is common to most of the classifications one finds on the internet, some of which seem rather random, much as if one classified theists according to whether they were Republican or Democrat, whether they were fideists or not, whether they attend church, whether they were always theists or converts, whether they think children should be sent to public school, and whether they get involved in internet debates about atheism.
Such are the formal distinctions, but with all Class II atheists one would also need some information about the content of the reasons. In general, arguments for A! fall into three roughly delineable groups: (a) arguments that T! is somehow meaningless or incoherent, (b) arguments that T! is superfluous or otiose or arbitrary, and (c) arguments that T! conflicts with some positive fact or evidence (defect and evil tend to be the preferred choice these days, but one occasionally finds others). Depending on how one understands meaninglessness, incoherence arguments tend to be associated with Class I and Class IIa, although one occasionally finds cases of Class IIb atheists that make use of very, very weak forms. If we rule out trivial understandings of superfluity, i.e., cases where something is superfluous only because it is meaningless (and thus cases falling in with Class I), then superfluity arguments are associated entirely with Class II, and most commonly with Class IIb. (Superfluity arguments can be fairly effective at tearing down other arguments, but it is very difficult to make them strong enough on their own to support robust conclusions; most instances in which the argument is associated with IIa are cases where the atheist in question has either obviously misjudged the strength of his argument or has an unusual account of what constitutes proof.) Conflict arguments are found associated with every subclass in Class II. These associations, however, only indicate the typical reasoning behavior of reasonable atheists who have put some thought into the matter; because people can be mistaken about what their reasons actually support, you can at least occasionally find any of these kinds of arguments in any of the classes.
The basic kinds of reasons the atheist actually has play an important role that can't really be ignored. Superfluity arguments tend to be stronger against polytheism than monotheism and incoherence arguments tend to be stronger against monotheism than polytheism; so whether an atheist is surrounded by polytheists or monotheists can potentially have an influence on what arguments he emphasizes, and this in turn can have an effect on which class he actually falls into. Likewise, superfluity arguments tend to be more common in contexts where empirical inquiry is given very high priority, while incoherence arguments tend to be more common in contexts emphasizing pure reason; so the overall rational expectations of the people around him can choose which arguments an atheist emphasizes, and this, again, can influence the meta-claims the atheist is willing to make about A!.
The formal scheme, incidentally, has analogues among theists and agnostics; but the differences in such positions do lead to some differences. For instance, the theists analogous to Class III here fall into two very clear types: principled (i.e., fideists) and unprincipled (usually people who are just theist because they grew up that way, and couldn't tell you why beyond that). It is difficult to draw such a distinction among Class III atheists, although conceivably there are kinds of atheistic existentialism and pragmatism that might be the atheistic analogues of theistic fideism. And the material scheme in each case, of course, is very different. Arguments for T! are much more diverse in character than Arguments for A!; theism is usually handled by Kant's 3+1 division (ontological, cosmological, teleological, moral), but I've noted before that, while handy for classifying the major arguments of the rationalists and empiricists of Kant's day, it can't capture all the significant arguments in history, or even all the significant arguments in play today, without being made so formless as to be next to useless.
Such is my first go at classifying atheisms. Are there any ways it can be improved? Any schemes that do a better job?