I’ll shortly announce the addition of my popular class on Christian and Islamic counter-apologetics to my monthly online course offerings. Which means it’s time to discuss the few issues I do have with my preferred course text for that: Malcolm Murray’s The Atheist’s Primer, which you can get at Google Play or Amazon.

Overall Murray’s Primer is the best, short, most complete and up-to-date survey of the foundations of modern atheism. He explains in 25 chapters why atheism has already met the Burden of Evidence and it’s theists who remain wallowing in a backwards, irrational, contrary-to-evidence position.

For a course text I need a book that addresses a wide array of theistic arguments (more thoroughly than even any book of mine, for example), including responses to liberal theologies (not just conservative or fundamentalist), that is expert enough and good enough to be worth reading and referencing—and that’s affordable and still in print (and with not too enormous a page count). On that combination of criteria there simply is no other book on the market but Murray’s Primer.

Here I will focus on some of my disagreements with it, solely so anyone using it can benefit from knowing that; but these disagreements are not very major, nor very many. The book overall is excellent.

Brief Summary

With one chapter each, Murray’s Primer covers the questions “Why not agnosticism?” “Which god are you denying?” “Religion without god?” and the many problems with sacred texts (including liberal interpretations of them, not just conservative or literalist), as well as all the standards: the “Ontological Argument” and “Why is there something rather than nothing?” and then of course the “Design” and “Fine Tuning” arguments; two chapters on the “Moral Argument” and one on the “Problem of Suffering,” the problems with “Omnipotence” and “Omniscience” and in particular relating to free will; “Time and Immutability”; the “God Is Love” cop-out; “Faith vs. Reason” and the proposal of “Fideism”; even liberal theological stances, from the question of “Ultimate Concern” to “Postmodern Theology”; “Pascal’s Wager” and issues of “Non-Falsifiability”; “Mysticism”; “The Meaning of Life” and “Death”; and finally “Error Theory” (explaining why people still believe in false things like gods and magical immortality).

Murray covers multiple atheist perspectives in each chapter, but often settles on his own, which might not wholly agree with mine, but that’s fine for the purpose. Anyone who wants to know where I am on an issue can search this blog or the TOC or index of my book Sense and Goodness without God. Also applicable are the positions I describe as central to my rejection of popular Western religion in Why I Am Not a Christian. Which are as applicable to any other popular religion, with only minor adjustments in the underlying trivia. The real use of Murray’s book is to see what the issues are, how atheists have dealt with them, and to see at least one example of Murray’s approach, which you can use as a model to explore your own.

Pascal’s Wager

As an example of what I mean, you can compare his treatment of Pascal’s Wager with mine, to see how we differ in what we make of this argument; although his resolution is no less valid, mine brings out a great deal more wrong with the wager than he does. He also makes a small math error in his tables at this point, in particular Table 19.3, where he shows the effect of adding epistemic probabilities to utility tables:

As here written, the results should be +1 and -1 (not -3.5 and +3.5), the opposite of what he argues. His table’s shown results here require the assumption that the probability that God exists is nearer 6% than the 10% he selected to represent instead (then you’d get around -3.5 betting on God and near +3.5 betting against, just as his argument intended). As neither number is meant to be actual, just illustrative of a point, his argument can be fixed with this revision. Because his only point here is merely that “it could be the case” that the probability God exists is lower than warrants betting on it; and some set of numbers would illustrate that. Which, Murray points out, returns the argument to what that probability is; so Pascal’s Wager hasn’t really gotten us anywhere.

However, this realization also returns the argument to the actual expected utility of heaven and hell, since one could just as arbitrarily “raise the utility” to exceed any drop in probability, a point Murray overlooks; I pick it up in Pascal’s Wager and the Ad Baculum Fallacy. There I point out this simply convert’s Pascal’s Wager into the Liar’s Wager, wherein no matter how unlikely you think some outandish promises of reward are, we can always tell a bigger lie to overcome that probability in a standard expected utility equation. Which reveals a fundamental flaw in standard utility reasoning, one similar to what Murray is calling attention to: it makes no sense that the more ridiculous our lie gets the more you should believe us. Which is why Pascal’s Wager fundamentally makes no sense. Rather than help defend us against liars, it requires us to be more vulnerable to them; which is exactly the sort of thing every religious con-artist wants. Blaise Pascal was really just another P.T. Barnum.

Agnosticism vs. Atheism

Another place where we disagree involves Murray’s discussion of atheism and agnosticism, which I find slightly muddled. I think it could benefit greatly from reframing the discussion in a probabilistic epistemology—as all sound epistemologies must be. I’ve discussed this somewhat already in my article on Misunderstanding the Burden of Evidence.

Instead of using “a word” for your position (“atheism,” “agnosticism,” “theism,” etc.), start by simply assigning a probability to God’s existence—or else imagine someone doing so—and then defend it. Then revise what you estimated once you realize it can’t be defended and you can only reliably defend a different probability. After completing this exercise, what probability do you end up with? Most people who call themselves agnostics will end up with a probability to “God” well below 50%. Some will have settled on a probability nearer or spanning 50%; but most, honestly, will end up well below that.

Which is actually what people in common parlance mean by an atheist: someone who doesn’t believe God exists is by definition someone who thinks the probability that God exists is low. Murray, by contrast, uses “agnostic” to refer to people whose probability assignment hovers close to 50%, as if the existence of God were so likely that a coin flip could decide it. That is correct if by “agnostic” you really only mean the wafflers, not the doubters. Yes, “I’m actually not sure” means “the probability is close enough to 50% that I can’t make a reliable call.” But most of the time “I’m not sure” means something a lot less, like 20% or 10% or even 1%. After all, “I’m not sure” I will win the lottery, and it has odds far lower than even 1%. So merely saying “I’m not sure” does not actually in practice tell us you mean “I think I have somewhere near a 50/50 chance.” And in fact, rarely will that be what you actually mean. Likewise anyone else.

Disbelief, by contrast, is meant to mean more than being uncertain. There is in actual fact no difference between saying “I don’t believe x” and “I deny x” except at most to indicate differences in degree. But still in either case, all you really mean is “I believe the probability of x is low.” Most usually, in fact, you mean “low enough that I can go about my life assuming it’s not true.” So the difference between “atheist” and “agnostic” really becomes a quibble over how low a probability has to be for you to assume that. And even then, given that everyone is always both an atheist and an agnostic in this sense (given some gods or other), there is little social utility in even caring about the distinction (see my illustration of this point in Atheist or Agnostic?).

The more so given that the only kind of theism anyone actually cares to know your stance on is among the least probable theisms ever conceived (e.g. either “Christianity” or “Islam,” as actual theists who are neither are so rare as to be socially irrelevant). And even any other conceptual theisms almost no one believes in are so improbable they are never entertained in any science as even a harebrained possibility. You’ll sooner find quite ridiculous proposals in peer reviewed physics journals today (like that the Big Bang was caused by the collision of giant space-carpets or that life on earth came from alien planets in distant galaxies) than any proposal that a god was responsible. Because there simply are no plausible god-based explanations of anything that any science studies. Anyone who doesn’t recognize this is, quite frankly, a little delusional.

Be that as it may, it’s possible some people who identify as agnostic feel that the probability of a god existing needs to be lower than whatever low probability they assign it before they will call themselves an atheist. But that’s just arbitrary benchmarking. The only useful progress that can be had in any such discussion—agnosticism vs. atheism, soft vs. hard atheism, strong vs. weak atheism, positive vs. negative atheism, lack of belief vs. denial, and so forth—is to inquire what their probability assignment is. And then to inquire why it is so high. That is, why is it higher than they think would warrant identifying as an atheist? Or why is it higher than they think would warrant saying a lack of belief in a claim amounts to a denial of that claim? And how improbable does something have to be for you, before you would say you “deny” it’s true, and why does it have to be that low? Is this even consistent with the rest of your behavior? (Do you require such low probabilities for everything else in life you deny is true? Or can you justify your use of different benchmarks?)

Those are the only useful questions to be discussing. Debating whether someone “is” or “is not” an atheist or an agnostic, by contrast, is completely useless and a waste of time. Everyone is an atheist with respect to some gods. Everyone is an agnostic with respect to some gods. And those two sentences remain true for any possible definition of atheist and agnostic. So the only thing that separates theists from nontheists is the lack of belief in all gods. After that, apart from what everyone actually wants to know (“Do you believe in Jesus?” or “Do you submit to Allah?”), the only interesting question left is, “How likely do you think the most likely god is? And why?” Everything else is trivia. (See also my article Who Is an Atheist?)

The Natural vs. the Supernatural

Another example is Murray’s discussion of natural causes vs. supernatural causes, which I find even more muddled. Although that is commonplace. Very little thought is often given to what one actually means by those terms, even in the most excellent echelons of academic philosophy. I believe this distinction would benefit from more careful attention to the semantics of practical usage.

Contrary to what Murray says, science can prove a supernatural cause, the same way it proves natural causes: by meaningfully testing and yet failing to falsify every alternative hypothesis whose prior probability is not already vanishingly small. Even in the domain of natural causes, science can never conclusively refute every alternative; it often doesn’t even waste time trying. For example, no scientific study bothers to take any trouble to rule out “the evidence was rigged by a CIA conspiracy,” yet a conclusion is declared on the evidence received—and that alternative hypothesis just isn’t even considered. But only because its prior probability is vanishingly small, not because it “could not be proven” if it were true. In principle it could be; that’s what makes the theory intelligible and not just a vacuous collection of syllables.

Supernatural causes are epistemically identical to that: they are dismissed not because they could not be proved, but because they are even less probable than CIA conspiracies to rig data. See my article Naturalism Is Not an Axiom of the Sciences but a Conclusion of Them for why; and for links to further discussions of what people actually mean when they distinguish natural from supernatural causes.

Likewise, once a supernatural cause were proved, we could not reject it simply because “it is still possible” a natural cause like “evidence rigged by a CIA conspiracy” explains the same evidence. By definition, proving a supernatural cause scientifically would mean establishing that that explanation was far more probable than any natural cause. It’s just that there are no supernatural causes; which makes for fewer of them than CIA conspiracies. So we never have any examples of science proving any, because there aren’t any to be found. In fact, our not finding them with science is precisely why we conclude there aren’t any.

Supernatural causes are thus not defined as undiscoverable causes or unprovable causes or unscientific causes. In actual practice, when people mean supernatural causes, they mean an irreducibly mental cause—like a bodiless mind, an irreducible soul, a magic aura that cannot be reduced further into nonmagical parts, and so on (I survey many more examples illustrating the difference in my old article Defining the Supernatural).

We totally could have proved the Genesis theory of origins—had it been true, the evidence we would have collected of it would be extensive by now. It just didn’t turn out that way. We found billions of years of cosmological processes followed by billions of years of geological processes and billions of years of single-celled organisms evolving, followed by hundreds of millions of years of multi-celled organisms evolving, before we even got humans, and they evolved over millions of years from apes! And we didn’t find an earth at the center of the cosmos surrounded by a spherical bronze-plate firmament way up in the sky on the other side of which was a celestial kingdom full of angels. We found billions of planets orbiting billions of stars stars occupying billions of galaxies. No firmament. No angelic heavens. No special center of God’s attention. And so on.

Any supernatural theory you can think of (“ghosts exist”; “a god interferes in the world”; “people can be magically resurrected”; “I can levitate a coin with my mind”) is totally testable. And fully falsifiable. They’ve just all been falsified—or have coyly avoided ever being tested, which is essentially an admission that they won’t pass. That’s a point I clarify the logic of when explaining why historians no longer entertain supernaturalist hypotheses in Proving History (index, “Smell Test”). It’s not because we couldn’t have proved them, had they been true. It’s that they all have been falsified or so resolutely evaded being tested as to ensure they were always bogus. (See my talk on this point, Miracles & Historical Method.)

Semantics vs. Evidence

Semantics is about what you mean. Evidence then determines if what you mean is true. Once you define God, that definition entails observations, which (as atheists know) are not confirmed but contradicted. So the theist must invent excuses (for why the evidence their definition predicts is not observed), which only add complexity to their definition, and thus reduce the probability of the claim, by making it more complicated. Theists mistakenly think adding these excuses increases the probability of theism, by allowing it to dodge all the evidence against it. But it’s actually the other way around (see “The Cost of Making Excuses”). Theists don’t know this because theists suck at logic. That’s why they’re theists.

You should simply ask the theist: Why do you need to make all those excuses? Shouldn’t God just be a simple, clear truth, as obvious as the existence of other people or our need to breathe air? And shouldn’t the universe simply match our resulting expectations? Why is it that God has to have all these excuses to behave differently than every other rational being on the planet? Why in fact must he behave so extraordinarily oddly that he has to make the universe look exactly like a universe with no God in it? And why do you think that’s a credible thing to think God does? Indeed…Didn’t you just make that up? Or do you have actual evidence that your excuses for God are true, and not just stuff you made up to avoid admitting the evidence doesn’t match what your original definition predicts?

The next thing a believer might turn to is to avoid defining God at all, precisely so as to avoid having to accept the consequences of that definition (that the predictions it entails don’t come true). For which your response should be: If you cannot define God, then you cannot know if such a thing exists or not. Because if you don’t know what it is, you can’t know if it is. Worse, if you define God in such a way that it makes no difference at all between a world with God and a world without God, then you cannot know that God exists, because you are then admitting there is no evidence that God exists. For you can only have evidence God exists if something would be different if God existed than if he didn’t, and we then observed that thing being indeed different.

Thus, semantics is the start and often the end of any debate. Believers want to avoid the consequences of any definition they choose, because the evidence then either disappears (and they have no reason left to believe) or is conspicuously contrary to prediction (and they’d have to even more adamantly concede they shouldn’t believe). And there are only two ways to avoid those consequences: avoid admitting any definition is true (and if they do that, call them out on it: hold up a mirror to them so they see what they are doing and how disturbing it is) or make the definition even more complex and thus even more improbable in order to explain away the failed predictions (in which case: do the same thing as with the first).

Murray covers the semantic issues well, and in numerous different ways. All valuable to study. But he doesn’t exactly single out this one central point about the difference between semantics and evidence, and how evidence decides what’s true once you’ve settled semantically on what proposal you intend to test the truth of. He of course agrees that this is how it works. But he doesn’t provide a clear mechanic for how it works. And that’s what I sought to do with the example-tree in Bayesian Counter-Apologetics.

Conclusion

I have only focused on a few points where I think Murray’s Primer falls short, which does not do justice to the rest of it—which is almost the entirety of it—which does a bang-on job, and on so many more angles than I usually even find time to treat. I highly recommend every atheist keep a copy on their bookshelf (digital or actual; I have both). And every atheist who likes arguing with theists, or exposing their lies, errors, and delusions, should read it, to have a handle on all the angles it covers.

There may be other respects in which I differ from Murray. If anyone reads his Primer (or already has) and wants to know if I concur or disagree with any point he makes, ask me in comments below (be sure to quote Murray so I can go straight to the page in my digital copy to refresh my mind on the point). All comments go to moderation and may take a while to appear. Only pertinent comments and questions will be approved. But also note that here and for every article on my blog, anyone who supports me regularly on Patreon or PayPal may have their comments post immediately without going into moderation first (if this is not happening, please contact me to make sure I have the correct email address in my comments whitelist).

§

To comment use Add Comment field at bottom or click a Reply box next to a comment. See Comments & Moderation Policy.

Discover more from Richard Carrier Blogs

Subscribe now to keep reading and get access to the full archive.

Continue reading