Because I needed the visual aids.
My take, in short, was that the sociology of religion doesn't have to assume that god doesn't exist, but it's more scientifically productive if it proceeds as though it assumes god doesn't exist. Brad then responded to my thoughts on the matter, I asked him to clarify his comments and he obliged me.
Now, having reached this point in the conversation I have realized that I need some visual aids, which is where y'all come in. Congratulations- I'm using this post as a sort of expanded comment to Brad's original post. So, hey, if you weren't following the conversation, you are now!
Brad concluded his most recent comments by saying:
It's not clear to me, however, that once we focus on natural mechanisms (without putting the supernatural in our model), it really doesn't matter which assumptions we make about the supernatural. (At least formally--certainly our assumptions creep into our work, but that's for another post).
As such, I don't see the value of assuming that God doesn't exist for doing sociology of religion.
I think my response is that it doesn't matter for the production of theories so long as the supernatural is not permitted as a variable, causal or otherwise. However, the caveats in this answer are crucial. We can imagine two related but distinct factors in this discussion: whether or not we use the supernatural as a causal variable and whether or not we assume that a god or gods exists. Then, in a move that will surprise no sociologist anywhere, we are in a position to produce a 2x2 table. This table, as it happens, is below:
The way I see it* these four combinations of two factors produce four possible "systems," two of which are very similar. First, when a god or gods is assumed to exist and the supernatural is permitted as a causal variable we are operating in the realm of theology. Thus, praying that god will guide events to a satisfying conclusion in some mysterious and non-material way reflects both belief in the existence of such a being as well as in the supernatural as a factor determining events in the universe.
The second combination, acceptance of the supernatural as a causal variable with an assumption that a god or gods does not exist, falls into the category of magic. Magic represents an attempt to exert control over the material world and may presuppose that there are regular causal linkages that can be uncovered but, unlike science, does not necessarily rely on a comprehensible causal chain. The elements that mediate cause and effect in magic are not apparent and, indeed, may be thought of as explicitly supernatural. So, performing the right ritual over a prospective voodoo doll will allow the magician to harm an enemy, but the mechanisms that allow this ritual to work are unknown and, indeed, unknowable. Additionally, it is not necessary that one believe in a god or gods in order for magic to be worked or believed in.
The third combination, assumption that a god or gods exists paired with a rejection of the supernatural as a causal variable, is natural philosophy. Indeed, in the infancy of science the natural philosophers thought that by understanding the laws of the world we would grasp the thoughts of god. In their case, however, recourse to supernatural explanations was frowned upon. Indeed, if god must make a common event occur through supernatural means it implies that he is "cheating" or, put another way, is not quite omnipotent enough to create a universe where such an event can occur without his direct intervention.**
Finally, the fourth combination in which god is assumed not to exist AND the supernatural is not admitted as a causal variable is what I term naturalism. In this case natural explanations are sought if only because there are no other possible types of explanations.
The way I see it, Brad is asking if I think there is a difference between research done in the context of the "naturalism" cell and research done in the context of the "natural philosophy" cell. I would respond that, really, there most likely isn't. Natural philosophy is equivalent to deism in this context and a deist scientist and an atheist scientist are likely to produce similar sorts of theories as, in each case, the ingredients of those theories and explanations consist only of things drawn from the natural world. Thus, for purposes of our conversation, these two cells are distinct but not different. Theists of any stripe can assume that god exists and, indeed, worship a god quite eagerly, but still do very good science so long as that belief isn't permitted to introduce supernatural causation or factors into the models.
That said, I think it quite likely that scientists operating in the realm of natural philosophy might find different research questions interesting than those who operate in the realm of naturalism. I do not personally, however, find such a difference to be of very much concern. A diversity of interests in science is not a bad thing and, so long as all parties follow the appropriate procedures for exploring the world, can only produce a fuller understanding of our universe. As far as I'm concerned, knowledge is rarely a bad thing.
Or so I would say given the relatively minimal amount of thought I've put into this. Others are invited to disagree.
* For the record, my views are informed by Rodney Stark's 2001 article "Reconceptualizing Religion, Magic, and Science." Review of Religious Research 43(2): 101-120.
** Known formally as the "God isn't a wuss" objection to supernatural causation.