25. What we do not know

The question: what has been left unsettled in this book?

The answer in two lines: seven places, which we count out here plainly by name. And saying 'I do not know' establishes nothing and brings nobody nearer to the truth of what we say — it removes one of the grounds for thinking ill of us, and puts nothing in the pan. And whoever reads a book's modesty as evidence that it is right has weighed with what is not a balance, and that is a mistake on our side and not on his.

1. Why this measure of pain?

We have no decisive answer.

And the cleanest thing in this field is that believing specialists say it before we do. Peter van Inwagen wrote his lectures on the problem of evil as a defence and not a theodicy: a story that may be true so far as our knowledge reaches, which he does not claim to be the real reason, his aim being to establish that the argument does not succeed as a proof, and not to explain why evil occurs. And he is not to be reported as saying that he 'does not know' — that is a denial he did not make; what he denied was any claim on his part to set out the real reason. And Marilyn McCord Adams — and she is a believer — sets aside the free-will defence and the 'greater good' theodicy in the case of horrendous evils, because they treat of the good of the whole and not of the good of the very person struck by it.

And on the other side, William Rowe — the author of the best-known evidential formulation of the argument — coins in that same paper the term 'friendly atheism' and holds that the believer may be entirely rational in his faith. And Paul Draper frames it as a comparison and not as a refutation: the distribution of pain and pleasure is more to be expected on the hypothesis of indifference than on theism, and he states that this does not establish that theism is false.

So the position is: a heavy probabilistic argument, replies that fall short of settling it, and the question of the single person struck, to which we have no answer. And a man who has lost his child is not told 'there is a wisdom in it'; the quick answer here is worse than silence, and it is read as haste to be done with the conversation.

2. Why the hiddenness?

It is not settled, and the burden on the believer is heavy here and we do not lighten it.

And Schellenberg's argument is not 'had He existed He would have shown Himself'. It is finer than that: on his account a perfect love entails standing always open to a relationship, and a relationship requires belief that the other party exists, so that the existence of the non-resistant man who does not believe tells against it. And the condition of non-resistance is written into its premise — so whoever answers it by saying that 'the denier is being obstinate' has answered a different argument.

And it is kept apart from the problem of evil in so many words by its author, so it is not to be answered with the answers to that.

As for the replies: in an earlier draft we ascribed to Michael Rea an objection that is not his, and we have put it right, and this is the statement of what is correct:

1. Rea's reply is aimed at the premise about love and not at the premise about belief. The substance of it: the transcendence of God makes our expectations about the form a perfect love takes unreliable. We picture the perfect lover as present in the form in which we ourselves know presence, and His presence may take forms we do not count as presence — in the text, and in the community, and in the world — so that with us it is read as absence when it is not absence. *And its example that can be seen*: a father whose son has learned that love is a telephone call every night, while the father expresses it by staying late at work. The love is there, the expected form is not, and the son reads absence. And the limit of the example is that the father is limited and has an excuse, while the claim concerns one who is perfect and has none — and Rea answers that transcendence cuts the comparison off at the root, and that is the place of the dispute.

2. As for the other objection — that belief in the sense of assent is not a necessary condition of every form of relationship — that is not Rea's, but is ascribed to Poston and Dougherty and Kvanvig. And a reader said that he did not understand it and could not picture it, so here is its example that can be seen: a man who does not believe that his missing father is alive, and who writes to him letters that he keeps, and lives on hope and not on assent. So there is a standing turning towards with no belief in it. And its parallel in the question: the man who prays the prayer of one who hopes and not the prayer of one who assents — is there not between him and the one he calls upon a connection of some kind? And if the answer is yes, then belief was not a necessary condition of every form, and the premise falls.

3. And even so it is not settled: hanging the matter on a thin notion of relationship — hope without assent — weakens the claim about perfect love itself: is this what the perfect lover wants for the one he loves?

And the collected volume on the question — chapters by believers and atheists and agnostics — agrees on nothing, and its two editors declare the question to be open. And we used to say 'eleven chapters', and we struck the number out: we did not set it against the book's table of contents, and a number that is not set against its source is not written.

3. Does any single argument stand on its own?

No. Not one.

And not one of the arguments for the existence of God binds a man who does not grant its premises. And Oppy holds that the arguments on both sides do not succeed in this sense, and he is an atheist — and he holds along with that that his own position comes out ahead on the theoretical virtues, as we said in the twenty-first chapter; and the place of this last is his other book (2013) and not his critical survey (2006), and we have not yet set it against a page. So he is not to be reported as holding the two pans level.

And Swinburne — the widest of those who have defended by probabilistic reckoning — has for his declared conclusion that the existence of God is more probable than not (greater than a half), with no certainty and no binding proof; and he states that exact numerical values cannot be given, so 'by a little' is not to be added to him any more than 'overwhelmingly probable' is. And it is a conclusion in dispute and not a thing granted.

And here are two places we have put right against ourselves, and we mention them because hiding them would be easier: The first: we used to say that Swinburne 'counts evil as evidence against theism and subtracts it from the balance'. And we have struck it out — we did not set it against the page of the chapter, and what is in the secondary sources is the opposite of it: that he holds the occurrence of evil not to lower the probability of the existence of God once the theodicy he offers is taken into account. And a claim ascribed to a man who holds the contrary of it is the most damaging thing in this field, and it is the very reason on which we built the whole references section. So it was struck out until it can be set against the page, and it is not to be restored except by that. And the second: we used to say that '97%' is a figure not in his book. And the sentence is sound to the letter — the book meant is *The Existence of God*, and it is not in it — but it is read as a denial of the figure altogether, and the figure exists: Swinburne has a figure of this kind in another book, and on another question — the resurrection of Christ, not the existence of God. So what is correct is to say: the figure is in another book and on another question, and whoever transmitted it in the matter of the existence of God carried a figure from one field to another. And an incomplete denial is an incomplete transmission, and that is the very thing this chapter finds fault with.

As for accumulation — that the weakness of each argument be made good by another — it is legitimate in itself, but it requires that each argument have a likelihood ratio above one.

So whoever adds up seven whose ratio is one or below and says 'they have become strong' has erred in the reckoning and not in taste: what does not raise the ratio does not raise it by repetition, and what lowers it subtracts and does not add. As for the weak argument whose ratio is above one, that does accumulate in fact — and that is precisely the Bayesian reckoning on which Swinburne's work is built. So the dispute is over which of them are above one, and not over whether accumulation is sound. And we used to say 'whoever adds up seven weak ones', and by that we voided by accumulation the very thing we had made a condition of accumulation, and handed the opponent a reply in one line.

So which of them are above one? — and the question is a fair one and we answer it

A doubting reader said: 'You said "that is not so in all of them", that is, in some of them. Which of them? And you named none.' So here is what we have, and we put it forward as a claim and not as a reckoning:

And this is a table of estimation and not a table of reckoning, and we set no figures against it because we do not have them. And whoever rejects the whole of it may do so, and we have nothing with which to bind him.

And a methodological objection to the whole reckoning — which we had not mentioned

And the heaviest thing brought against Swinburne is not an internal reply about the value of one of his factors, but a claim in method: that theism is to be judged by the standards of any other theory and is not exempted from them, and that by those standards it is an unsuccessful theory; and that the adding of a new metaphysical category tips the scale towards the naturalist on grounds of simplicity; and that whoever sets his factors after he knows the result has built the reckoning on his result. And Sean Carroll has framed that in setting theism against the models of physics — and he states that it enters the balance and is not put outside it, so its exclusion is not to be ascribed to him.

And its absence from our draft was a fault we own: whoever sets before his reader all his opponents' replies except the heaviest of them has chosen for him.

*And where it breaks down*: that theism be judged by the standards of scientific theory — simplicity and predictive power — is itself a claim in method that needs a ground of its own: it is not granted that those standards are the ones that govern in metaphysical questions, and whoever lets them loose brings down with them a great many questions nobody accepts bringing down. And Carroll himself denies that there is in his paper any proof that God does not exist. So the reply is heavy and it is not settled, and that is why its place is in this chapter and not in another.

4. Figures we have left aside deliberately

And they are the best-known things put about in this field in Arabic:

• Ten to the power of ten to the power of one hundred and twenty-three in Penrose: it is a ratio of volumes in phase space under a known measure, and it is the denominator of the fraction and not its numerator. And reading it as a probability is not an outright distortion, but it needs further premises that go unmentioned when it is put about. And Penrose puts it forward in search of a physical explanation, and in 1979 he proposed a hypothesis — the Weyl curvature hypothesis — and not a settled explanation, and he himself counts it a proposed constraint that stands in need of an explanation.

And we used to say 'and he proposed that explanation himself', and we have put it right: we ascribed to him an achievement he does not claim, and that is our own fault in the note beneath it. And what suffices for the purpose is that the door of physical explanation is open and has been put forward by the owner of the figure himself, so that the claim that 'nothing is left but design' falls by that alone.

• Weinberg's anthropic bound on the cosmological constant: it was put forward by a declared atheist to explain matters by selection within a multiverse, and not by a designer.
• The claim that the fine-tuning is severe: the widest recent quantitative survey of it narrows it rather than widening it, and finds the permitted range wider than what the enthusiasts pass around.
• And the strongest philosophical formulation of fine-tuning — Collins's formulation — is probabilistic and not demonstrative by the statement of its own author, and tells only against single-universe naturalism.

And why do we leave them aside? Because a claim that is denied by the very man it is ascribed to is more damaging than silence about it. And the example: the man who called a witness in court, and the witness stood up and denied what had been ascribed to him — he lost the case and lost whatever was left of his words in it.

5. And what we do not know about ourselves

Motivated reasoning is not the opponent's failing alone. A man who sits down to reply never reads the objection with a neutral eye, and we are among them. And skill in this field increases the capacity to defend, not the readiness to revise.

And the remedy for it is procedural and not homiletic: that the strongest form of the objection be drawn up and then put to a man who actually holds it, and that he be asked: is this your objection? And we do not claim to have carried this out in every place in this book. What we were able to do was to take every objection from the book of its own author and not from the summary of his opponent, and to state with every reference what it does not say of what is ascribed to it.

And our notes were counted up against us and found to lean: most of them restrict what is put against us, and the fewest of them restrict what is put for us. And the observation is correct, and the reason is not a complete excuse: that most of what our Arabic reader passes around in this field is a truncated transmission from Western researchers, so that the greater part of the setting-right fell on that side. But the effect remains what it is: a reader comes out with 'nothing establishes anything' and then finds the burden laid on one side. So whoever sees this as a lean should count it a lean, and in this revision we have struck out its plainest forms: the title of the twenty-first chapter, the ordering of the criteria, and the closing of this chapter.

6. The moral objection to the content — and it is the largest absentee

And this place was added in this revision, and the reason for it is that a doubting reader said: 'Your five places contain no place that is moral and about the content of the text. The problem of pain with you is about God, not about the Book. And most of those I know did not come out of cosmology; he came out of a verse.' And he was right, and this is the largest thing we have missed.

And where the matter stands with us:

1. The question stands and is not denied. That a man should read, in a text claimed to be from God, a ruling his conscience finds repugnant with a repugnance that no explaining removes — this is an objection of the first rank, not a passing quibble, and it is the fourth criterion in chapter 22 applied in its own place.

2. And we have no general answer to it. The answers put forward are three, and each has its price: that it be said the ruling was understood other than as it should have been — and this is sound in some places and is misused in every place, so that it becomes a key that opens every lock; or that it be said our moral balance falls short in this field — and that is an answer that overturns the criterion at its root and makes 'I know with certainty that this is injustice' a sentence with no application; or that it be said the ruling is tied to a state and a time that do not hold generally — and this needs to be shown in each place on its own, not let loose as a claim.

3. And what we say: we have not settled it, and it is the heaviest thing in this chapter against us. And we did not bring it into this book as a section of its own because it is looked into in particular places and not in a general rule, and because the running of it on a particular text is the work we said in chapter 22 is other than this book. And whoever counts this an escape has a plain case, and we do not hold it against him.

7. And the final adjudication between claims of revelation

The balance we have given you — the five criteria — for the most part rules out and does not raise up. As for the final adjudication between competing claims of revelation, it has not settled even among the specialists themselves: Alston, the most rigorous of those who have defended religious experience as a ground, counts it a real difficulty in his own position and not in his opponent's, and his answer to it is that holding to the practice is rational for want of any neutral ground — and it is an answer that preserves the man who holds it and does not sort the claims, and licenses for every holder of a practice what it licenses for him. So whoever offers you a balance that sorts the claims in a single page has promised you what those who have worked at it their whole lives have not reached.

And finally

So what we say in honesty is this: there is no proof.

As for the weighting, that is a second claim, and it is not granted by our having finished with the first. Whoever claims it is bound to name the piece of evidence whose ratio is above one and to set out his reckoning; and we have named in the table what we have, as an estimate and not a reckoning, and we hold nobody to it. And the most that is in our hands is stated plainly by the man who holds it: that it is more probable than its contrary (greater than half) — and nothing is to be added to it, neither 'by a little' nor 'an overwhelming balance', for its author sets down that exact numerical values cannot be given — and it is a matter of dispute and not a thing granted. So whoever reads 'a weighting' from us, let him read it within that limit and add nothing to it.

Our first conclusion said 'the weighting stands' after lines that had in them 'No. Not one' and 'that is not so in all of them' — that is, it announced the result after denying what carries it. And that is the plainest fault in this part, and it has been struck out.

So what we say is three things, and no more:

• That the claim 'created, then fell silent' — the positive claim, not the suspension of judgement — is asked for its warrant as we are asked for ours; while the man who suspends judgement we hold to nothing.
• That you have a measure in your hand with which to weigh what you please, our own words among them, and most of it brings down and does not raise up, and you have seen how it holds in an exercise with no religion in it.
• That what has not been settled is seven matters that we counted out by name, and among them are some that weigh more heavily on us than on our opponents.

And you may come out of this book still holding your doubt. Nothing in it makes it a condition on you that you be convinced, and your staying where you are is not to be read as a failure in it. And that a man should know where his knowledge ends is no defect in knowledge; it is a condition of his being believed in what he does know — and not evidence that he is right. And between the two there is a difference, and we ran them together in an earlier draft.