The first move that atheists often make is to assume that religion is attempting to do what science wants to do. Even when Carroll is trying to show that they are not a priori incompatible he uses this analogy:
"An airplane is different from a car, and indeed if you want to get from Los Angeles to San Francisco you would take either an airplane or a car, not both at once."He suggests that there could be a universe where science and religion were compatable--starting and ending in the same place. Unfortunately, Caroll says, religion doesn't end in the same place, so they aren't compatible in our universe. The assumption is that science and religion are just two routes to the same goal: reliable theories about the world, knowledge, coherent models of the natural world, explanations of strange phenomena... however you want to phrase that goal.
Carroll later writes:
"The favored method of those who would claim that science and religion are compatible — really, the only method available — is to twist the definition of either 'science' or 'religion' well out of the form in which most people would recognize it. Often both."But then Carroll chooses to twist the definition of religion into a kind of science!:
"Scientifically speaking, the existence of God is an untenable hypothesis. It’s not well-defined, it’s completely unnecessary to fit the data, and it adds unhelpful layers of complexity without any corresponding increase in understanding."It's very easy to make the case that religion is bad science. Just like playing tennis is a bad soccer, or that playing poker is a terrible strategy for winning at Go Fish. What is the benefit of saying that a sermon couldn't hold up for 5 minutes at a biology conference? None, unless you think that religion is an old, primitive form of science. If your religious opponent accepts that premise, then it's game over.
Of course, there are lots of examples of religion being a primitive form of science, the kinds of superstition that seemed to be verified by experience, say, a dancing ritual to change the weather--and the failure of crops being due to impiety toward the gods, etc..
It is worth noting however that many of these ideas aren't stupid. Some have suggested that elaborate rituals before a hunt would have helped focus the mind on success and increased the teamwork skills of the tribe. In short, it did what it was intended to do. If we look at Rene Girard's scapegoat theory, we may see that sacrifice rituals gave off psychological benefits--eliminating community aggression--in a way that was very powerful. When sacrifices occurred they had actual, felt psychological pay off in a community.
This form of religion is obviously a lot like science. Observe the world, look for patterns, see what works, try to 'divine' answers to mysteries, pass on the superstitions about what concoction is best for a snake bite, what pattern of actions leads to the highest birthrate. Naturally, modern science developed out of these kinds of habits: astrology became astronomy, alchemy became chemistry, numerology became mathematics, various theories about spirits/souls/humours became modern physics, biology. So it goes without saying, the witch doctor ought to become the medical doctor because he is really within the same trajectory--identify the problems/mysteries, experiment, go with the most effective results.
Some of the major modern religions however--Judaism, Christianity, Islam--aren't this kind of religion. Early Christians were considered atheists because they didn't do the whole observe/experiment/repeat process. These religions did not develop out of a need for better crop production. They are based on recieved revelatory experience.
Now, there remains a lot of the aforementioned superstition in all religious traditions. I visited Turkey once and depending on the region the Muslims had various folk spirituality and occult practices tied up in their Islam. (I think the same is true for Catholics in Mexico.) People still can get hyped up about coincidences, unexplained events, etc.. But my impression is that at core these 'revealed' religions don't require trust in those kinds of 'signs' and often have a difficult relationship with them (Jesus gets mad at people looking for signs, Jews outlaw divination but use lot casting in temple worship, etc..)
If someone recites a creed "I believe in God the Father, Creator of Heaven and Earth...I believe in Jesus Christ...I believe in the Holy Spirit..." that isn't a hypothesis, a theory. We aren't postulating--and it's dishonest to assume that it is. Why then don't we reject the Pledge of Allegiance because it lacks 'explanatory power'?
Atheists are prone to group all religions into the same 'failed God hypothesis' but I think there is a significant difference between certain religions that need the divine or spiritual for explanatory power and others that don't assume God as a way to explain things. It wasn't like someone sat down and said, "Why is the world here? Why is nature like this? Oh, maybe 'Jesus is the Messiah.'" I think Christians way too often take that bait, assuming that God was realized as a way to make sense of the world.
All that said, I think there are lots of other fruitful routes for the atheist--questions about revelation and authority, for example. But the 'failed science' argument for some religious traditions is disingenuous, confused and a rhetorical false start.