I shall start, if you will permit me, by indicating how I shall be understanding a couple of crucial terms, what I shall be meaning when I talk this evening of ‘the universe’ and of ‘God’.
1 ‘Collapses’ then needs scare quotation marks as, on the Everett interpretation, instead of really collapsing, every component of the wave-function lives on in some branch or other.
2 The use of ‘universe’ and ‘God’ then necessitates that if there is a God, he exists outside any and every universe. If God exists, it is not then strictly-speaking correct to say we live in a universe in which God exists. Rather, we should say we live in a world in which God (and at least one universe) exists. See note 31 on page 185 of Swinburne, Richard, The Existence of God (Oxford: Oxford University Press, 2004).
3 I do argue for it elsewhere, e.g. my Belief in God (Oxford: Oxford University Press, 2005), part I. In doing so, I follow in broad outline Swinburne, who argues that polytheism, for example, has a prior probability lower than classical theism (in e.g. Swinburne, Richard, ‘The Argument to God from Fine-Tuning Reassessed’, in Manson, Neil (ed.), God and Design: The Teleological Argument And Modern Science (London: Routledge, 2003), 107) due to its complexity. As well as its simplicity, the God hypothesis has its plausibility as a metaphysical necessity on its side relative to alternative supernaturalist hypotheses. Suppose that we allowed that a hypothesis positing the pantheon of Greek gods was as simple as classical theism and raised to the same extent the probability of the fine tuning that we observe. Unless the fact that the pantheon of Greek gods existed would be as plausibly necessary were it to obtain as the fact that the God of classical theism existed would be necessary were it to obtain, the God hypothesis would nevertheless be a better explanation of the fine tuning and the existence of the Greek pantheon is less plausible as a necessity (if it obtains) than the existence of the classical theistic God is plausible as a necessity (if it obtains). So, even suspending judgement on its relative simplicity, we are justified in starting with a presumption that classical theism is to be preferred as a potential explanation of the fine tuning over Greek polytheism and, as the same considerations apply mutatis mutandis to other supernaturalist hypotheses, over these too. Were the proponent of Greek polytheism or an alternative supernaturalist hypothesis to assert his or her religious beliefs as beliefs in metaphysically contingent things and assert that the metaphysical contingency of his/her explanation didn't undermine its truth, we might agree. Were he/she to maintain that its contingency didn't undermine its explanatory value, we should disagree. Greek polytheism and other supernaturalist hypotheses, if they present themselves as metaphysical contingencies, just push the fine tuning up a level. (And if, as just mentioned, they present themselves as metaphysical necessities, they are less plausible as such than classical theism.) We shall come upon this issue again in the main text when discussing the first naturalistic model for explaining the fine tuning of the constants; the higher-level fine tuning remerges on it, prompting the move to the maximal multiverse model. The notion of plausibility in play here is certainly a difficult one, but the same notion is in play in other areas of Philosophy too, whenever one judges of things that they are metaphysical necessities. (The fact that nobody has yet built a time machine such as that depicted in H. G. Wells' story The Time Machine is more plausible as a metaphysical necessity than the fact that nobody has yet built a space ship capable of interstellar travel.) This notion is not, I suggest, reducible to the notion of simplicity or even universally conjoined with it: it may be in some cases that that which is relatively complex appears more plausible as a metaphysical necessity than that which is simpler. There being nothing at all would be simpler than there being something, but – given that there is something – we know that there being something is more plausible as a metaphysical necessity than there being nothing.
4 There are many good discussions of these things, e.g. Leslie, John, Universes (New York: Routledge, 1989); Rees, Martin, Just Six Numbers: The Deep Forces that Shape the Universe (New York: Basic Books, 2000); Barrow, John, The Constants of Nature: From Alpha to Omega (London: Jonathan Cape Collins, 2002); Collins, Robin, ‘God, Design, and Fine-Tuning’, in Martin, Raymond and Bernard, Christopher (eds.), God Matters: Readings in the Philosophy of Religion (New York: Longman Press, 2002); and Holder, Rodney, God, The Multiverse, and Everything: Modern Cosmology and the Argument from Design (Aldershot: Ashgate, 2004).
5 I talk in terms of ‘landscape’ as does Susskind (Susskind, Leonard, The Cosmic Landscape (Little Brown and Company, 2005), passim) although with something slightly different in mind; Leslie talks in terms of a ‘local area of possible universes’ (op. cit., 138). There is room for confusion here between considering the landscape/area being talked of as a landscape/area of possibilities, or one of actualities. In fact Susskind's term ‘landscape’ refers to different solutions to string theory which are instantiated via Linde's eternal inflation model and so form part of the one actual universe in my sense of universe; he calls this the ‘megaverse’. These things are glossed over rather in the main text, but not in a way that affects the validity of the argument.
6 Of course gods other than the classical theistic God could explain the fine tuning, but the classical theistic God hypothesis is the simplest god hypothesis and the most plausible as a necessary truth, so it is the one that should be preferred. See previous note.
7 Of course, much depends on how ‘life’ is taken. Carbon-based life of the sort we are familiar with obviously occurs in less possible universes than life on more relaxed understanding. But the proponent of the argument characteristically urges that life per se requires some degree of structuring and in the vast majority of possible universes there is not sufficient structuring to allow it to arise however relaxed (yet plausible) a definition of life one operates with. Certainly the sort of life on which we'll be focusing needs this structuring.
8 Mawson, T. J., Belief in God (Oxford: Oxford University Press, 2005), 145.
9 Although see Sober, Elliott, ‘The Design Argument’, in Mann, W. (ed.), The Blackwell Companion to Philosophy of Religion (Oxford: Basil Blackwell, 2004).
For a very full and balanced discussion of observer selection effects, see Bostrom, Nick, Anthropic Bias: Observation Selection Effects in Science and Philosophy (New York: Routledge, 2002).
10 Bradley, M. C., ‘The Fine-Tuning Argument: The Bayesian Version’, Religious Studies 38 (2002), 382–385, discusses this point, although in discussing it in terms of objectivism, he does not draw out the requirement of what I call the trans-universality of value. A naturalist realism is, I take it, a version of objectivism, but would not be sufficient for the fine tuning argument to go through.
11 Examples of such theories would be Susskind's (The Cosmic Landscape) and Smolin's (Smolin, Lee, The Life of the Cosmos (Oxford: Oxford University Press, 1997)). To make the outcome we are concerned with a statistical certainty, we'd need to formulate these hypotheses so that they involved an actual infinity of oscillations or what have you. Otherwise the outcome would just tend towards a statistical certainty as the number of oscillations or what have you tended towards infinity.
12 Here and elsewhere I assume certain things about how one may speak of probabilities even when considering sets of possible outcomes with an infinite number of members. In doing so I set myself against some of what is said in Timothy McGrew, L. McGrew and E. Vestrup, ‘Probabilities and the Fine-Tuning Argument: A Sceptical View’, Mind 110:440, 1027–1038. This is discussed in Graham Oppy, Arguing about Gods (Cambridge: Cambridge University Press, 2006), 205–207. See also Swinburne, ‘The Argument from the Fine-Tuning of the Universe’, 185–188 referring back to 168–172 and Koperski, Jeffrey, ‘Should we Care about Fine-Tuning?’, British Journal for the Philosophy of Science 52 (2005).
13 This is assuming that the machine works as the terrorist describes it, of course.
14 Compare Lewis, David, On the Plurality of Worlds (New York: Basil Blackwell, 1986) and Tegmark, Max, ‘Many Lives in Many Worlds’, Nature 448 (2007). It will be noted that I am ignoring one common objection to multiverse theories; that they are ‘unscientific’. Of course I would dispute the claim that is sometimes made, that multiverse theories are unscientific as there is no evidence in favour of them; the fine tuning of a universe to the sort of life on which we're focusing is evidence in favour of them. ‘But such models don't provide causal explanation’, someone might say. Indeed, the maximal multiverse has explanatory power, yet the explanation it provides is not causal; so much the better then, I would reply, as anything which explained the universe causally would itself be part of the universe (if the causal explanation was a scientific one) or God or something similar (if not).
15 I take it that this is the solution to what Hacking calls the problem of the ‘inverse gambler's fallacy’. Pace Hacking (Hacking, Ian, ‘The Inverse Gambler's Fallacy: The Argument from Design. The Anthropic Principle Applied to Wheeler Universes’, Mind 96 (1987), 331–340). If you were a surviving victim of the terrorist we are imagining, and you knew the machine to work as the terrorist had described it, you could then conclude from your survival that it was more likely that the terrorist had tried his machine out many times than that he had tried it out only once. And the simplest hypothesis that has him trying it out many times is the hypothesis that he has tried it out an infinite number of times. The fact that we know that there aren't these sorts of actual infinities within our universe is why we are not intuitively drawn to such a hypothesis. We don't know that there are not the relevant sort in the case of the situations that concern us in the main text. There is then another objection to the theory, that actual infinities are not physically realizable. Copan and Craig explore this in Copan and Craig (Copan, Paul and Craig, William Lane, Creation out of Nothing (Grand Rapids: Baker Academic, 2004)). I ignore this in the main text as my overall conclusion does not depend on this objection failing.
16 There are a number of possible universes which are not fine tuned to life of the sort we're interested in, yet which plausibly have certain good-making features and which God may have chosen to create in virtue of these features. We have considered one such universe, a universe with just one hydrogen atom in it, and suggested that it would have the good-making feature of a certain sort of beauty. It's not then that fine-tuned-for-life-of-the-sort-we're-interested-in universes are the only ones that God might have good reason to create. But this doesn't matter for the argument; as long as a fine-tuned-for-life-of-the-sort-we're-interested-in universe would be ‘quite likely to occur’ (Swinburne, Richard, ‘The Argument from the Fine-Tuning of the Universe’, in Leslie, J. (ed.), Physical Cosmology and Philosophy (New York: Collier Macmillan, 1990), 155) on Theism, then whatever other universes are more or less likely to occur on the God hypothesis in addition is beside the point (although see discussion in main text concerning the explanatory power of the maximal multiverse hypothesis relative to the God hypothesis in raising the probability of the evidence to one). Of course, merely assessing the probability of the evidence on the hypothesis will not get one very far in assessing the probability of the hypothesis on the evidence, since an inherently improbable hypothesis may give a high probability to the evidence; hence the unavoidability of prior probabilities and – I would concede – the unavoidability of using simplicity, as it strikes one, to judge of these. But it is the unavoidability of just this sort of thought process which, as we shall see, forms the basis for the most powerful version of the Design Argument. So this can hardly be an objection to my argument.
17 Some suggest that simplicity considerations favour the God hypothesis over the maximal multiverse hypothesis (Moreland, J. P. and Craig, W. L., Philosophical Foundations for a Christian Worldview (Intervarsity Press, 2003), 487, and Holder, op. cit., 16). It is an implication of the final argument that I shall advance that, ultimately, we cannot advance above brute intuitions here. It has been suggested that since Kepler's and Ptolemy's laws of planetary motion involve equations with common variables, at least the number of free parameters in those equations can be compared and relative simplicity judged thereby (Dowe, David et al. , ‘Bayes not Bust! Why Simplicity is No Problem for Bayesians’, The British Journal for the Philosophy of Science 58 (2007), 709–754), although some (Forster, Malcolm, ‘Bayes and Bust: Simplicity as a Problem for a Probabilist's Approach to Confirmation’, The British Journal for the Philosophy of Science 46 (2003), 399–424) would deny even this. However this issue is to be resolved, it seems plausible that nothing similar is possible when comparing the God hypothesis to a multiverse hypothesis. Swinburne suggests, in essence, that a hypothesis is simple insofar as it involves few substances and few properties, but this is not unproblematic as the properties in turn need to be simple ones (not grue/bleen-type ones), which just pushes the problem on a stage. All that being so, the approach taken in the main text, to pump an intuition with a thought experiment and then move on, is, I suggest, the only approach to take. See though Richard Swinburne, Epistemic Justification (Oxford: Oxford University Press, 2001), 83–102. It may also be worth pointing out at this juncture that my final conclusion doesn't depend on the maximal multiverse hypothesis being simpler (in that it says that even if it's simpler, we should still prefer the God hypothesis).
18 This is also the view of David Lewis, the most prominent exponent – although for quite different reasons – of the maximal multiverse hypothesis. See also Bradley, ‘The Fine-Tuning Argument: The Bayesian Version’, Religious Studies 38:4 (2002), 389. Arguably the situation is not so clear cut as I make out in the main text: some of these universes – lots of them indeed – contain types of thing that don't exist in our universe, e.g. new sorts of particles (sorts that are physically impossible in our universe), so one might argue that my saying in the main text that on the maximal multiverse hypothesis, there is only one type of thing, universes, is too quick. My intuitions go as they do as I suppose that physical stuff (whatever universes are made up of) is fundamentally one type of thing and spiritual stuff (whatever God is made up of) is fundamentally another. This raises then the general problem of how we are to determine whether two objects, A and B, are two tokens of the same fundamental type or one token each of two types. Insofar as A and B may be discriminated between (and thus plausibly are numerically distinct), there will be some qualitative difference between them in virtue of which we make the discrimination and, in lieu of anything else, this could always be used as the hook off which to hang a claim that they are tokens of two types of thing. ‘We should not be talking of A and B as two peas in a pod’, someone might say, ‘but rather as one A-pea-in-a-pod and one B-pea-in-a-pod’. But, I take it, some concepts strike us as gerrymandered: in the situation I am imagining, it strikes the majority of us that two peas in a pod, A and B, are not tokens of two types of thing, but rather two tokens of the one type of thing, a pea in a pod; indeed peas are just one type of thing whether they're in a pod or not. In deciding which concepts are gerrymandered however, we will be drawn back to judgements of simplicity. See previous note and later discussion in the main text.
19 I am then assuming that the machine shuffles each pack separately and then draws a card in turn from each shuffled pack, rather than mixes the packs in together with one another during the shuffling process.
20 Walker attributes it to Kant in Walker, Ralph, Kant (London: Routledge, 1999), ch. 11; there are also versions in e.g. Walker, Ralph, The Coherence Theory of Truth (London: Routledge, 1989). There is a parallel here with worries as discussed by e.g. Davies (Davies, Paul, The Goldilock's Enigma: Why is the Universe just Right for Life? (London: Allen Lane, 2006), ch. 8) that multiverse theories make versions of Bostrom's simulation argument more pressing or they make more pressing the hypothesis that we are probably so-called ‘Boltzmann brains’ (see esp. discussion in R. Collins, ‘The Fine-Tuning Argument’, forthcoming in W. L. Craig and J. P. Moreland (eds.), The Blackwell Reader in Natural Theology (Oxford: Blackwell)), worries which do not arise on the hypothesis of ‘a God who is no deceiver’, as Descartes might have put it. As well as the line of thought explored in the main text (which applies induction primarily to time going forward [although secondarily, in a later note, to time going backwards too]), we might consider Penrose's (Penrose, Roger, The Road to Reality (London: Vintage Books, 2004)) argument, which applies it to space going outwards. As Holder put Penrose's argument in an email to me: ‘The creator had 1 in 10 to the power 10 to the power 123 universes to choose from, only one of which would be as ordered as ours. However, to make life you only need a solar system's amount of order. To make only a solar system, surrounded by chaos, by the random collisions of particles, which is all that is required to make life, the order required is much less than this, though still vast. It is 1 in 10 to the power 10 to the power 60. Since 10 to the power 10 to the power 123 swamps 10 to the power 10 to the power 60 completely, what that means is that although a universe with order 1 in 10 to the power 10 to the power 123 exists with probability 1 if all possible universes exist, the probability of … [creatures such as ourselves] observing such a universe is only 1 in 10 to the power 10 to the power 123’. He suggests a helpful analogy, drawing on the typewriter one. ‘Suppose you have a monkey typing and life corresponds to its coming up with “To be or not to be, that is the question”. It is much more likely to come up with that than the whole of Hamlet, so it is vastly more probable that “To be or not to be, that is the question” is surrounded by junk than the rest of Hamlet, let alone the whole of Shakespeare. Thus … [creatures such as ourselves] are far more likely to find ourselves in a solar system surrounded by total chaos than in a totally ordered universe'. It is of course less obvious that it is valuable to creatures such as ourselves not to live in such a spatial ‘oasis’ than it is that is valuable to us not to live in such a temporal ‘oasis’, which is why I think this variant of the argument is weaker than the one discussed in the main text. Matters are similar, I suggest, with regards to certain features of our universe which one might think are ‘over designed’. For example, the proton lifetime is at least 2 × 10 to the power of 32 years, i.e. at least ten to the power of 22 times the age of the universe, which is vastly longer than needed for life to form. It is less immediately plausible that a universe ‘special’ by being over designed in this way is evidence of God. We shall return to some of these issues at the end.
21 Walker, Ralph, Kant (London: Routledge, 1999), 171.
22 Goodman, Nelson, Fact, Fiction, and Forecast (Harvard: Harvard University Press, 1955), ch. 5.
23 Ralph Walker, Kant, 172.
24 But see Swinburne, The Existence of God, 88, n. 10.
25 But see next note for further reflection on this grant and its dubious status on the maximal multiverse hypothesis.
26 A possible counterargument (suggested to me by Swinburne) is as follows: Given that creatures exist for a period that is long enough for them to sustain thought concerning their universe and given that most time periods are longer than the fifteen or so billion years that this universe has taken to generate us having these thoughts (because almost all periods are longer than that), it is not very unlikely on the maximal multiverse hypothesis that creatures having these sorts of thoughts will find themselves in a universe in which induction has worked and continues to work for periods a lot longer than fifteen or so billion years. Someone might argue that on the maximal multiverse hypothesis we may consider the issue as analogous to one where a barman with an infinite amount of time ahead of him intends, over that time, to stop serving repeated Singapore slings to all but one of the infinite number of customers to whom he starts off serving them. This being so, he will of course disappoint an infinite number of his customers each time he does his rounds around the Long Bar (it is a very long bar). But – in order to leave himself enough people to disappoint later by failing to serve them the Singapore slings they've come to expect – the frequency with which these disappointed customers are distributed around the Long Bar is infinitely low. So, were you a customer, in seeing the bar man coming towards you on one particular round, you could be almost sure that he wouldn't disappoint you on that round. (This is not, it will be noted – an inadequacy of the analogy raises itself here – as a result of your performing an induction along the following lines: ‘Well, he's never disappointed me in the past, so …’. Rather, it is the result of your performing a calculation of this sort: ‘the people he disappoints on any round of the bar have to be infinitely thinly distributed and thus the chances of my being one are infinitely small.’) So, the continuing tractability of our universe to induction as we find ourselves utilizing it is indeed a feature which needs explanation, someone pushing this line might concede, but it gets an adequate explanation in terms of the maximal multiverse hypothesis, an explanation no better indeed than the one in terms of God, but no worse either. However, this counterargument does not work. Just as the ways in which the universe could diverge from the present in the future and surprise us are more frequent in logical space than are those in which it could continue more or less according to our expectations, so the ways in which the universe might converge on the present and not be anything like we suppose our history to have been are more frequent than those in which it does so more or less via the processes we suppose it to have followed. The issue here then may be put as one of whether we have any reason to think that induction has worked in the past. In a maximal multiverse most people like us (except for some of their relational properties) believing induction has worked in the past are in a temporal ‘oasis’; in other words, they are mistaken in thinking it has worked in the past. Even amongst that relatively rare group who find themselves with what they take to be good evidence that it has worked in the past (it doesn't seem to them as if they're in a temporal oasis), the majority are mistaken. In essence then, the force of the argument in the main text may perhaps better be put like this: either abandon the view that one knows induction has worked in the past (absent supernatural intervention) and that it will continue to work in the future (absent supernatural intervention) or abandon the maximal multiverse hypothesis and believe instead in the God hypothesis.
27 There is a variant of the Problem of Evil to be addressed at this stage then: given that the universe does sometimes do unpredictable things, how is this to be reconciled with its being created by the classical theistic God? But this is a topic for another paper.
The sort of inductively tractable universe that God would have good reason to create creatures of our sort in would only require of course inductive tractability at the level of the sorts of objects and actions which had moral salience for such creatures; our easy understanding of sub-microscopic properties, e.g. the more recondite recesses of Quantum Mechanics, would not have its probability appreciably raised by the hypothesis that the universe was created by God; nor would the discovery that Quantum Mechanics is easily understandable by the majority of people exposed to it increase the probability of the God hypothesis. This would be or is (depending on how easily one believes Quantum Mechanics is in fact understood) another bit of ‘over design’.
28 I am grateful for having discussed another solution – although not one open to someone holding the maximal multiverse view – with Walker, who is now more sympathetic to it than he was when he first wrote on this topic.
We may posit that there is a metaphysical distinction between what we might call ‘real’ properties (green and blue, for example), on the one hand, and what we might call ‘non-real’ ones (grue and bleen, for example) on the other; and we may posit that there is a metaphysical principle dictating what we might call ‘an ontological preference’ for universes which instantiate real, rather than non-real, properties. These posits, taken together, would then explain (to a greater or lesser extent, depending on the strength of the preference posited) why we find ourselves in a universe where things are green and blue, rather than grue and bleen. One might suggest that we could get away with positing an ontological preference for real simplicity and letting the rest take care of itself: given that green and blue are, one may contend, really simpler than grue and bleen, and given that it would be really simpler for creatures to be directly aware of the really simple properties, rather as Russell thought we are directly aware of universals, from this preference for real simplicity alone we could explain why it is that we find ourselves in a universe where things are green and blue and where we have managed to discover this feature and innumerable other similar features. We cannot of course posit an ontological preference for what would be really simplest: ours is not the really simplest world (for there is stuff – a whole universe at least – in it). So we would need to posit a preference for ‘relative real simplicity’ and work from there. In that this preference is being designed to do the job God's choice in creation does on the God hypothesis and may, in principle, be reworked and reworked without limit until it does so, it would be futile to maintain that strategies of this type will inevitably leave unexplained something the God hypothesis explains. But there is something rather unsatisfactory about them nonetheless, indeed precisely because they seem so ad hoc. Going down this road, it seems, we may in the end be left with a Parfitian ‘selector’ that is less plausible as a necessity than the God hypothesis. Some of these issues come up in my ‘Why is there anything at all?’, in Nagasawa, Y. and Wielenberg, E. (eds.), New Waves in Philosophy of Religion (London: Palgrave Macmillan, 2008), but I do not feel I have plumbed the depths and hope to be able to return to it at a later date.
29 I am grateful to Sophie Allen, Rodney Holder, Dennis Lehmkuhl, Richard Swinburne, and Ralph Walker for their comments on a draft of this paper. I am also grateful for the chairmanship of Anthony O'Hear at the meeting of the Royal Institute of Philosophy on 10th October 2008 where this paper was delivered as a lecture and for the questions and comments which he and those attending the lecture provided on that occasion.
Email your librarian or administrator to recommend adding this journal to your organisation's collection.
* Views captured on Cambridge Core between <date>. This data will be updated every 24 hours.
Usage data cannot currently be displayed