Colin Howson evidently has two main aims: to discredit the Abrahamic religions on a wide variety of grounds, and to argue against the existence of God. The book is for the most part written for a general readership, although some readers will find various arguments difficult – for example, the argument involving Tarski's Undefinability of Truth Theorem, on pp. 201–205.
In pursuing the first aim, chapter 1 makes remarks about features of Abrahamic religions which dispose them to violence and intolerance, biblical portrayals of God's totalitarian regime including extreme punishments for the wicked, the Inquisition, sexual abuse of children in Catholic orphanages and schools, and so on. In chapter 2, amongst other historical claims, we find Howson telling us that it is doubtful that Jesus as a real historical character ever lived. There are further attacks on religion scattered through subsequent chapters. Although some of Howson's specific assertions are correct – e.g., the Inquisition did exist, and acted very badly – by and large the quality of Howson's arguments concerning religion is low. There are no signs at all of well-informed, skilled engagement with the Bible, or with the historical issues concerning Jesus and the origins of Christianity, although he says a lot about these. Despite frequent, confident, highly disputable claims (e.g. on pp. 6, 8) to the effect that Christianity claims this or that, the book does not reflect any attention to the question ‘How do we decide what propositions belong to the content of Christianity, as distinct from the opinions of this or that group of Christian believers?’. Possessing an answer would be vital for well-constructed argument in favour of many of Howson's assertions. Unfortunately, they are often mere blustering.
Howson's case against the truth of theism is much more worthy of respect. He argues that the existence of God is logically impossible, and (independently) that it is logically incompatible with the existence of evil. He also attacks various arguments in favour of the existence of God, including the ontological argument, the Kalam cosmological argument, and the fine-tuning argument. Michael Behe's case for an intelligent designer of various biological structures and processes also receives attention, though (as Howson indicates) Behe's hypothesis does not entail theism. In addition, Howson addresses the claim that morality depends on God, launches what amounts to a rather unsystematic attack on arguments for theism based on alleged miracles, and discusses Pascal's Wager.
Howson declares that the God of theism is demonstrably inconsistent with innocent suffering (p. 199). He says surprisingly little in support of this declaration. His remarks on page 119 amount merely to the assertion that it seems true to a great many people as well as him. The only argument provided, on page 49, consists of the early moves in J. L. Mackie's well-known paper ‘Evil and omnipotence’ (Mind, 1955). Howson neglects to inform his readers that subsequently, in The Miracle of Theism (Oxford: Clarendon Press, 1982), p. 154, Mackie himself later abandoned his earlier argument and the logical inconsistency claim (while nevertheless arguing that certain truths about evil provide good evidence against the existence of God). I believe the 1955 argument to be unsuccessful, for reasons given in my own book God, the Best, and Evil (Oxford: Clarendon Press, 2008), pp. 131–134.
Howson rightly points out that divine omnipotence and omniscience play vital theoretical roles in philosophically developed theism. He argues that each of these attributes is logically impossible. Omnipotence, he says, can be shown to be impossible via the paradox of the stone (pp. 45–46). The paradox has led to a large literature on the question of how ‘omnipotence’ should be defined, given that it is a technical term in metaphysics. Howson quotes Mackie and endorses his assumption that if God is omnipotent then it is not the case that there are things which he cannot do. But why should theists adopt a definition of ‘omnipotence’ that undergirds this assumption? After all, as Howson himself points out, theological tradition regards omnipotence as involving unlimited power. Howson owes his readers an explanation of why theists, or anyone else, should hold that if God possesses unlimited power then he can do anything, i.e. anything that is logically possible. In general, the fact that an agent cannot do something is often due to factors other than lack of power – for example, you cannot now tell me exactly how many atoms there were one hour ago in the Andromeda galaxy, but it is not lack of power which is the obstacle.
The thesis that omniscience is impossible is supported (on pp. 201–205) by a line of thought involving, first, an argument, based on a theorem due to Tarski, for the view that there could not be a being who knows all and only the truths, and then, second, an immediate inference to the conclusion that there could not be an omniscient being. I am unsure that the first stage is sound: the argument seems to be enthymematic, and it is not clear exactly what the suppressed premises are. But let us grant that there could not be a being who knows all and only the truths. Why should we infer that there could not be an omniscient being?
Many theists, for reasons entirely independent of Howson's line of thought, reject the view that ‘omniscient’ should be defined in terms which entail knows all and only the truths. Many theists hold that God lacks infallible belief with respect to future undetermined events. Furthermore, many philosophers who believe that God in fact determines all that occurs also believe that God could have adopted a looser providential policy that provided for future undetermined events. Both groups of philosophers and theologians would be likely to insist that the definition of ‘omniscient’ should accommodate God's being omniscient in such circumstances. In God, the Best and Evil, p. 39, I offered a definition which fulfils this desideratum. Philosophers are used to theoretically motivated changes in the meaning of terms in logic, mathematics, and empirical science – consider ‘true’, ‘set’, ‘mass’, ‘simultaneous’, ‘species’ – and so they should recognize that it can be legitimate in theology and in philosophy of religion.
There remains a substantial question: can a revised definition of ‘omniscient’, one which permits there to be an omniscient being who does not know all truths, enable the proposition God is omniscient to fulfil its roles in contributing to the explanatory power of theism in relation to such empirical evidence-statements as Our universe is fine-tuned? Prima facie, the answer is Yes; I do not know whether Howson would disagree.
The upshot: Howson has not established either that the existence of God is logically incompatible with the existence of evil, or that omnipotence is logically impossible, or that omniscience is logically impossible. Therefore he has not provided good reasons for believing that theistic hypotheses purporting to explain empirical evidence all have zero prior probability, and so zero probability relative to our total admissible evidence.
Howson claims on page 106 that there are other reasons why, when considering any specific empirical argument for the existence of God, the prior probability of theism should be regarded as extremely low. Unfortunately he does not identify these reasons, but confines himself to attacking Swinburne's view that, other things being equal, a simpler hypothesis has a higher prior probability than a less simple one. On this point, Howson's arguments vary in quality; the best is a technical one on page 135.
Apart from alleging that theism has a very very low prior probability, Howson states what amounts to another general objection to any empirical argument for the existence of God: theism is a very poor hypothesis because it cannot explain anything, any more than the exact sequence of Heads and Tails resulting from yesterday's 100 tosses of this coin can be explained by the hypothesis Santa Claus willed this sequence and whatever Santa Claus wills, happens (p. 129; cf. p. 71). Evidently Howson is relying on the premise that if theism provides genuine explanations then it yields predictions – not merely inferences to truths we already know – which inquirers will eventually be in a position to verify or falsify on the basis of evidence not currently available.
Consider Julius Caesar believed that he was a Roman citizen, which contributes to explaining a lot of evidence we have today about Caesar's participation in Roman politics. Does it yield (by entailing or conferring high probability on their truth) predictions which will be verifiable or falsifiable on the basis of new evidence? Surely only degenerate ones like No newly discovered manuscripts by ancient authors will contain credible accounts of Caesar's career while affirming that Caesar did not believe that he was a Roman citizen. If the bar is set that low then theism can rise to it, even if it is only by yielding predictions like No great voice from the sky, of the kind imagined by Cleanthes in Hume's Dialogues, will declare that the Olympian gods rule the world and that the Judaeo-Christian God does not exist; in that case, theism's fitness to provide personal explanations of empirical evidence is not threatened. If the bar is set high then the demand for predictions rules out many respectable empirical hypotheses providing respectable explanations.
When theists advance probabilistic arguments for the existence of God, they do not typically formulate the hypothesis so that it entails the known evidence: they formulate it so that they have to provide substantial argument for the view that the evidence is rendered probable by the hypothesis. For example, in Swinburne, The Existence of God (second edition, Oxford: Clarendon Press 2004), pp. 337–339, it is clear that the hypothesis is God exists, not something like God brought about such-and-such or God intended to bring about such-and-such. The provision of the substantial argument means that the explanation is not (as Howson claims on p. 194) a cheap fix. Admittedly, theism provides no account of how God is able to do what he does, in the way that anatomists and physiologists can give an account of how people are able to walk. But this does not suffice to render explanations in terms of divine causal activity trivial, degenerate, and therefore illegitimate. After all, sooner or later physicists must invoke causal powers without knowing how exercising the powers brings about the relevant effects; indeed, they may suspect that no further how-explanation is possible.
Howson's discussions of specific empirical arguments for the existence of God are intertwined with a useful exposition of his Bayesian approach to scientific inference. He identifies a couple of variants of the Prosecutor's Fallacy, which involves neglecting prior probabilities and arguing, say, from The evidence is enormously improbable relative to the chance hypothesis to The chance hypothesis is enormously improbable relative to the evidence.
On page 68 he claims that this fallacy occurs in Francis Collins's presentation of the fine-tuning argument. But if one reads the sentences Howson quotes in their context in Collins's book, The Language of God (Free Press, 2006), pp. 75–78, it is plain Collins does not ignore prior probabilities: he argues, in effect, first, that the hypotheses to be taken seriously in the face of fine-tuning are limited to (i) there being a Multiverse, (ii) there being one universe with chance giving rise to the fine-tuning features, (iii) there being some unified physical cause of at least many aspects of the fine-tuning, and design by a creator; and, second, that the prior probabilities of each of the first three are very low. Furthermore Collins's conclusion on pages 75–78 is not that the universe was designed by a creator; it is For those willing to consider a theistic perspective, [fine-tuning] certainly provides an interesting argument in favour of a creator. If Howson is analysing Collins's argument in Bayesian terms, the first part of the foregoing sentence should be regarded as amounting to something like ‘For those who assign theism a positive prior probability considerably higher than that assigned to hypotheses which are not to be taken seriously (such as The gods Poseidon and Athena together caused the fine-tuning) …’
On page 82 Howson claims that John Leslie, in Universes (Routledge, 1989), p. 121, commits the Prosecutor's Fallacy in his treatment of the fine-tuning argument. This claim is highly doubtful. The sentences Howson quotes, in their context, are making the point that there is a wide range of examples in which the only serious alternative hypotheses are Design and Chance, and even though the prior probability of Design is much lower than the prior probability of Chance, the probability of the evidence relative to Design is so much higher than the probability of the evidence relative to Chance that empirical inquirers unhesitatingly reject Chance. This point is correct, and is well illustrated by the examples Leslie gives on page 121. Leslie makes a similar point on his pages 9–10, where there are in effect three serious hypotheses –Fish Ensemble, Design, and Chance – and we reject Chance.
Howson offers no argument for supposing that the fine-tuning argument essentially involves the Prosecutor's Fallacy.
In addition to points already covered above – the zero or very low probability of theism, the explanatory impotence of theism, the ‘cheap fix’ objection, and the Prosecutor's Fallacy – Howson advances three other main objections to the fine-tuning argument. First, ‘it is not obviously unreasonable to place a substantial probability on the hypothesis that a future physical theory will fix [the values of the relevant physical constants]’ (p. 171; pp. 99, 104). Second, the argument fails because of ‘the sheer incoherence of computations of the “chances” of fine-tuning were there no fine tuner’ (p. 171; cf. pp. 107–113). Third, the probability of fine-tuning relative to theism is negligible: for surely God ‘could think of less complicated and hazardous ways of creating intelligent life’. These objections deserve serious consideration. Although they have received considerable discussion in the recent literature, Howson can hardly be blamed for omitting, in a book intended for a general readership, discussion of various attempts to deal with them.
