A quibble over godlessness [tpm]

[Originally posted at Talking Philosophy Magazine blog]

Atheism and agnosticism. If you ask some people, atheism is just a sexed up version of agnosticism. After all, atheism is about what you believe (or don’t believe), and agnosticism is about what you know (or don’t) — so when we say that we’re atheists, we’re just putting accent on the fact that God is really really really super unlikely.  But others will say that atheism and agnosticism are perfect companions. They’ll tell you that agnosticism is just a closeted form of atheism. After all (they’ll say), since agnostics dislike being called ‘theists’, they must be atheists — the one position collapses into the other.

To see an example of this contrast in action, consider the views of Bertrand Russell and Anthony Grayling. Russell argued for atheism in public, and only called himself an agnostic among philosophers. That’s because he thinks there’s a significant gulf between atheism and agnosticism. By contrast, in a difficult-to-parse exchange with Jerry CoyneAnthony Grayling begged to differ — an agnostic is either an atheist, or just plain irrational.

Grayling does himself a disservice by repeatedly claiming that “an assimilation of proof concerning matters of fact to proof of the demonstrative kind”, when that doesn’t really seem to be exactly what’s going on. This post is going to be my attempt to make sense of what Grayling is up to, and an argument about why he hasn’t got it right.



This fight comes down to a complaint in the theory of knowledge. Grayling’s claim is that Russell tacitly bases the distinction between atheism and agnosticism on “a quibble about proof”.

Russell thought that you can’t disprove the existence of a corporeal object — like, say, a mountain made of gold — in the same way that you can disprove 2+2=6. You can prove that 2+2=4, alright; but you can’t prove, strictly speaking, that God (or the golden mountain) doesn’t exist. You can only say, “A golden mountain is pretty unlikely”. Moreover, it would seem that the two kinds of proof can be measured on a common scale — that of certitude. For Russell, following Hume, deduction and induction involve different degrees of warranted certainty. The idea here is that we ought not have as much confidence in inductive proof as we do with deductive proof. Logic and mathematics occupy a kind of heaven, an epistemic ideal; inductive proof, like most of our commonsense knowledge, will always be in perdition. As a result, while it’s tempting to believe that “Rain is wet”, I am really only warranted in believing that “rain is probably wet”. Likewise, you’re only warranted to believe that God is really really really not likely, though you might brand yourself as an atheist.

This is where Grayling and Russell part ways. Grayling believes that Russell is wrong to think that you are warranted in being any less certain about induction than deduction. In other words, Grayling thinks that it is just as provable that rain is wet as it is to say that 2+2 = 4. Instead of putting deduction in heaven, and induction in hell, the two modes of reasoning stand side-by-side. They just seem to be adhering to different standards.

What deduction and induction have in common is that both are capable of their own kind of proof. Grayling seems to think that proof is defined negatively for all kinds of discourse, in terms of what is irrational to reject. As he puts it, a thing is proven so long as we can adduce “evidence of the kind and in the quantity that makes it irrational, absurd, irresponsible or even a mark of insanity to reject the conclusion thus being supported.” So, any belief that is scientifically invalid (e.g., “next time I go out in the rain I won’t get wet”) is just as irrational as a belief that is, for all intents and purposes, impossible to demonstrate (e.g., “rain does not wet anything”, or “2+2=6”). The parenthetical quotes are cases of propositions that are disprovable, and hence (I think it is fair to say) certainly wrong.

This relates to agnosticism and atheism in the following way. Grayling thinks that the concept of “God” is as absurd, irrational, irresponsible, and possibly insane as the concept of “2+2=6” or “rain does not wet anything”. For Grayling, if a person says that they are agnostics about God, they might as well be saying that the jury is still out on whether or not two and two make six. Russell, and a great many following him, would prefer to say that while ‘2+2=6’ is demonstrably false, other howlers (like “there is a golden mountain”, or “rain does not wet anything”) are only probably false.



Optimus prime mover.

Or, at least, that’s how I’ve interpreted Grayling. He hasn’t made it easy. The problem is that when he accuses Russell of “an assimilation of proof concerning matters of fact to proof of the demonstrative kind”, the accusation can be made just as effectively against his own view. For while it may be true that Russell is measuring the two kinds of proof along a single spectrum of certainty, there is a sense in which Grayling is doing the very same thing, but in another way! After all, he’s assimilating them negatively: by saying that proof of any kind is the sort of conclusion that is “irrational, absurd, irresponsible or even a mark of insanity to reject”, once supported by sufficient evidence.


A more serious problem is that Grayling hasn’t got the commitments of atheism squared away. Grayling argues that “if you seriously mean that you think it might be conceivable or possible that there could be evidence for a deity, [then you are] agnostic, not atheist” (“possible” meaning, I presume, “rationally possible”). In this, he provides an implausible formulation of what it means to be “agnostic” and “atheist”. For an atheist might think that there is no God, or even that it can be proven that there’s no God, while still admitting that God is a rational possibility.

Recall, the only thing that an agnostic needs to say is that they don’t know whether or not God is real. Atheists, in the strongest sense of the term, claim to know that there is no God; in a weaker sense, atheists claim to believe that there is no God, and live their lives accordingly. Grayling is effectively saying, “there are no weak atheists. Go strong or go home.”

Unfortunately, contrary to Grayling’s claims, atheists can possess warranted doubts against their atheism, even if they think it is proven that there is no God. For one does not need to be convinced that God is rationally impossible in order to be an atheist, any more than one needs to think that Yeti are rationally impossible in order to think that they belong in fairy tales. That’s because, while it is indeed quite irrational, absurd, and irresponsible to believe that God exists, that doesn’t mean that it is rationally impossible to believe in God’s existence.

For you to believe that God’s existence is rationally impossible, or “inconceivable”, you must mean “inconceivable by everyone, everywhere, even at their best”. After all, “impossible” is a tough-guy word, a heavy-duty blunt instrument — when you say something’s impossible, you mean business. To say something is “impossible” is to say that it is “necessarily not”; and “necessary” implies universality; and ‘rationally necessary’ implies universality across the class of rational people. So when you say that belief in God is rationally impossible, you’re saying that no rational person can believe in God.

So what’s the difference? Well, since we’re talking about a crowd of rational people, that presumably means that we must assume that even the ideally rational, bold, imaginative, and informed person is in the crowd; and we have to have faith that the most rational person would agree that God has been proven to not exist. By contrast: to prove a thing — even in Grayling’s sense of “proof” — is not to suppose that you believe it as an ideally rational and informed agent, or even that you would retain the same beliefs if you were closer to the ideal. It just means that, according to some standards of discourse, denying a proposition is daft. Never mind whether or not the standards themselves require revision.

Here’s the punchline. If you are rationally compromised (to any substantial degree), then you might have proven something, alright; it’s just that the thing you’ve proven, could be working with suboptimal standards. For, as a matter of fact, while you might think that belief in God is irrational, etc., you might also think that you might be (to some substantial degree) irrational, irresponsible, or insane — that is, to think that you’re an ordinary Joe believer, slumming it with the rest of us. Long story short: “rational impossibility”, if it means anything like what it says on the tin, is an idealized standard that belongs to the epistemic angels, while Grayling’s sense of “proof” belongs to mortals.

So why does Grayling think that God is rationally impossible, or inconceivable? Evidently, he has a narrow view of the impossible. Grayling says that God and Yeti are not the same sort of thing, because at least people can think of the conditions under which the existence of a Yeti might be confirmed (e.g., as being a furry Wookie-like creature). By contrast (he says), it’s not even clear what research programme could be contrived to figure out where there is a God, because the concept of God is a catch-all wish-fulfiller.

I agree that belief in God is madcap all the way, because the idea of “God” in the mainstream Abrahamic faiths is a nebulous blob, a Rorsharch for the credulous. (To me, this is like determining what feats of strength and vigor are possible by visiting a leper farm. But anyway.) Suppose that Grayling’s programme is plausible. If it is, then the fact that it is both irrational to believe in God and irrational to believe in 2+2=6, ought to tell us that it is rationally impossible to think that there is even any evidence that they might be true claims.

And yet, and yet, and yet… ! — people still reject Grayling’s account. Because (Grayling says) people hold some lingering fidelity to Rome, one final chain around our ankles, and that we have yet to emancipate ourselves from it. But that implication seems pretty unlikely when thinking about avowedly Godless heathens and Grayling-dissenters like Jerry Coyne and Richard Dawkins. So there might be a better explanation of why people may be resistant to Grayling’s programme.


First, “conceptual clarity” doesn’t matter in the way Grayling thinks it does. For my part, I agree that “God” is a bullshit concept. But that isn’t really very interesting or important when we’re trying to figure out whether or not the existence of God is rationally possible. Meanings can be clarified — and sometimes, even the relevant, supposedly deep ontological features of the concept can be reconceived — without improving the epistemic standing of the doctrine or the worldly practices of the conceivers. For instance, five years from now, the Roman Catholic Church might declare that God lives in a slum on one of the planets in the Hades Gamma Cluster, thereby turning God into a more exotic sort of Yeti. But this really wouldn’t make any difference to how skeptics think about the Catholic worldview. We’ll never see Hades Gamma, or develop any means of seeing whether he’s out there. The Hades Gamma version of the Catholic Church is just as irrational, madcap, and irresponsible as the Orthodox version.

Second, because doubt matters to scientific integrity. Both Coyne and Dawkins make a virtue out of retaining it, and they seem to do so for the sake of the institution of science. A measure of doubt is always sublime.