As previously noted, someone recently put some of my criticisms of the Leibnizian cosmological argument to William Lane Craig (some of the relevant posts where I present the criticisms can be found here, here, here, here, and here), and Craig offered a reply. I offer a rejoinder below, but first, a couple of initial comments and clarifications:
First, both the person who puts my arguments to Craig (alias "Midas") and Craig himself discuss several points I never asserted. For example, I don't deny that certain representative mental states can function as prima facie justification for possibility claims, and I don't take Kripke/Putnam a posteriori necessity cases to undermine the general reliability of modal seemings. Nor do I think occasional errors in modal inferences undermine the prima facie justification of possibility claims, any more than I think occasional errors in perceptual seemings undermine the justification of perceptual beliefs. My primary area of specialization is modal epistemology, and I can tell you right now that those are pretty bad criticisms of conceivability-possibility inferences. So I certainly don't want to be associated with such claims.
Second, whether intentionally or not (I'll assume the latter) Craig wastes a lot of space mischaracterizing my fourfold categorization of concrete beings (contingent dependent beings, contingent independent beings, necessary dependent beings, and necessary independent beings), and then subjecting them to criticism. But of course that's just to change the subject and attack straw men. (Variations of my own categorization can be found here, here, and here.)
So much for my initial concerns; on to the bigger points. Let me start by stating the version of the Leibnizian cosmological argument Craig defends, which is Stephen T. Davis' version of it:
1. Anything that exists has an explanation of its existence, either in terms of the necessity of its own nature or in terms of an external cause.
2. If the universe has an explanation of its existence, that explanation is God.
3. The universe exists.
4. Therefore, the universe has an explanation of its existence. (from 1, 3)
5. Therefore, the explanation of the existence of the universe is God. (from 2, 4)
The first thing to note is that premise (1) is ambiguous. Now Craig is certainly right in saying that the premise carves up logical space so that it represents two mutually exclusive and jointly exhaustive categories of beings; there is thus no third category of beings between the two categories. But of course it doesn't follow that there isn't a further sub-categorization within at least one of the two categories. And the epistemic possibility of the latter sort of sub-categorization is what I'm asserting. For consider the notion of "a being whose existence is explained in terms of the necessity of its own nature" in that premise. As Craig defines the notion in Reasonable Faith (p. 107 of the 3rd edition) such a being is one that exists of its own nature, and thus has no external cause. But this definition allows for at least two epistemically possible sorts of beings that could play such a role: metaphysically necessary beings and factually necessary beings. Metaphysically necessary beings are necessary in the sense that their inner nature causes them to exist in all possible worlds. By contrast, factually necessary beings are necessary in the sense that, while there are possible worlds at which they do not exist, their inner nature renders them eternal, indestructible, uncaused beings at all the possible worlds at which they do exist. Either sort of being stands in contrast to his notion of beings whose existence is explained in terms of an external cause (Call the latter sorts of beings 'dependent beings'). Because premise (1) admits of these two different interpretations, we should keep them separate. Thus, call the first version the Metaphysical Necessity version, and the second version the Factual Necessity version.
Now one of my main criticisms of Leibnizian cosmological arguments is that it's epistemically possible that all dependent beings are ultimately composed of factually necessary beings (e.g., perhaps matter-energy is a factually necessary being). In this scenario, then, the dependent beings (e.g., rocks, trees, planets, you and I, etc.) come into being when two or more factually necessary beings are combined, and they (i.e., the dependent beings) cease to exist when they decompose into their elements. However, the elements of which dependent beings are composed (i.e., the factually necessary beings) can't pass away, for while there are possible worlds at which they don’t exist, they are uncaused, eternal and indestructible at all possible worlds in which they do exist.
So that's the epistemically possible scenario that I take to be an undercutting defeater for the argument. The scenario functions as an undercutting defeater for different premises, depending on which version of premise (1) is being asserted. On the Metaphysical Necessity version, the scenario undercuts premise (1). For if my scenario is epistemically possible, then it needs ruling out before premise (1) can be accepted. Now Craig says he has defended the premise elsewhere with independent support (see the section on the argument in Reasonable Faith, 3rd edition), but that support appears to boil down to two primary considerations: (i) The principle stated in the premise is intuitively plausible, and (ii) it avoids criticisms of a stronger version of the principle (e.g., van Inwagen's argument that the principle entails that every truth is a necessary truth, and everything that obtains does so of necessity). What, then, about Craig's claim that the principle is intuitively plausible? Well, it's not clear to me that it is, but even if it were, that plausibility is undercut by my epistemically possible scenario stated above. And what about the consideration that it avoids the criticisms of a stronger version of the principle? The obvious answer is that that only shows that it's not prima facie false; it does nothing to show that it's prima facie true.
So my epistemically possible scenario undercuts the Metaphysical Necessity interpretation of premise (1). But does the argument fare any better on the Factual Necessity interpretation? No, for then the scenario functions as undercutting defeater for premise (2). For even if the dependent beings ultimately require explanation in terms of one or more factually necessary beings, the factually necessary beings could be the fundamental constituents or stuff of physical reality (matter-energy, say), and not God.
So that's my main criticism, suitably contextualized so as to apply to Craig's (and Davis') version of the argument. In light of that, we can see why Craig's main reply to my criticism is fundamentally misguided. He writes:
We generally trust our modal intuitions on other familiar matters (for example, our sense that the planet Earth exists contingently, not necessarily, even though we have no experience of its non-existence). If we are to do otherwise with respect to the universe’s contingency, then the non-theist needs to provide some reason for his skepticism other than his desire to avoid theism.
In reply, note that the criticism shows a fundamental misunderstanding of my main objection. For my criticism is not that the universe is metaphysically necesary. Rather, my criticism was that it's epistemically possible that the fundamental constituents of the material world are factually necessary. But of course that's compatible with saying that the universe does not exist at some possible worlds, and is thus contingent in that sense. And if so, then even if our modal intuitions are generally reliable, and even if they confer prima facie justification on the claim that there are possible worlds at which our universe does not exist, such evidence wouldn't favor the hypothesis that our universe is composed entirely of dependent beings over the hypothesis that our universe is partly composed of factually necessary beings.
Now Craig appeals to four other considerations in support the claim that the universe isn't factually or metaphysically necessary: it is supported by: (i) Richard Taylor's thought experiment involving the translucent sphere, (ii) Craig's thought experiments involving objects composed of quarks different from those in the actual world, (iii) his a priori arguments against the existence and traversability of an actually infinite series of events, and (iv) a posteriori evidence that our universe began to exist in a singularity (the Big Bang). Let's consider these in turn.
Now (i) and (ii) are thought experiments offered as evidence of the possible non-existence of our universe and its constituents. But that's just to offer evidence that the universe isn't a metaphysically necessary being. But since both dependent beings and factually necessary beings fail to exist at some possible worlds, such evidence doesn't favor the hypothesis that the universe is composed solely of dependent beings over the hypothesis that it's at least partly composed of factually necessary beings. (For further criticisms of Craig's use of these examples, see my post, "Modal Epistemology and the Cosmological Argument".)
What about (iii), i.e., the a priori arguments against the existence and traversability of actual infinites? Unfortunately, as I and others have argued, the a priori arguments against a beginningless past are undercut.
What about (iv), i.e., the a posteriori evidence? Unfortunately, we lack sufficient warrant to accept Craig's claim that the universe had an absolute beginning in a Big Bang singularity. For starters, unless one has adequate specialized training in the relevant scientific fields, one will not be competent to evaluate the evidence for this claim on one’s own. I take it that this applies to most of us; we must therefore defer to the consensus view on this matter among the relevant experts. But the problem is that there is significant divergence of opinion among the experts on this issue, as there is no consensus among them that our universe had an absolute beginning in a singularity. Craig is thus relying on a faulty appeal to authority.
Also, as (atheist defender of the in-principle possibility of intelligent design science) Bradley Monton has pointed out, Craig's case for an absolute beginning in a singularity is based on false physics. Now Craig often appeals to the Borde-Guth-Vilenkin theorem to argue that there must be an absolute beginning to the material world on any interpretation of the data, but James Sinclair (Craig's co-arguer on this issue) has admitted that this conclusion depends on the outcome of a number of technical arguments. And the problem is that these arguments cannot be competently assessed by non-specialists; the best we can do is sit tight in our agnosticism about the issue until there is a consensus of the experts to which we can defer.
 First example: Craig misconstrues my notion of a necessary dependent being as one that has an external cause, and yet also exists of its own inner nature. But of course that's a pretty egregious mischaracterization of the notion of a necessary dependent being I offered. My notion of a necessary dependent being is of one that exists at all possible worlds (hence, necessary), but in virtue of an external cause, and not its own inner nature (hence, dependent). Indeed, Craig seems to understand that this is the notion at issue, as he clearly understood Midas' example of the theistic conceptualist account of abstracta: if abstract objects are divine concepts, then they exist at all possible worlds, and yet they are dependent upon the mind of God. The illustration I originally offered was one I borrowed from leading philosopher of religion Richard Swinburne, viz., that of God the Son being willed by God the Father as an act of essence. For more on this latter example, see Swinburne's book, The Christian God.
Second example: Craig misconstrues my notion of a contingent independent being as a being that is self-caused, and then rightly points out that such a notion of a being seems incoherent, as it would appear to require that it is explanatorily prior to itself. But of course that's not the notion of a factually necessary being that I and a number of theistic philosophers of religion have in mind. Rather, the notion is of a being that does not exist at some possible worlds (hence, contingent), but is nonetheless uncaused (not self-caused), eternal (or everlasting), and indestructible at all possible worlds at which it does exist (hence, independent). The illustration I originally offered comes from Richard Swinburne as well, viz., his view of the first person of the trinity as a factually necessary being.
Perhaps, though, Craig might push the objection by saying that factually necessary beings are self-caused in some sense -- viz., it is their own internal nature that causes them to exist --, but if that is the notion of being self-caused at issue, then the same would seem to apply to metaphysically necessary beings in excelsis.
Third example: At one point Craig conflates my category of contingent independent beings with brute facts -- a conflation he earlier disambiguated by means of his illustration of God as a factually necessary being. It seems to me that there is a significant difference between saying that there is no reason at all why a being exists (brute fact) on the one hand, and saying that a being exists because it has the properties of being uncaused, eternal, and indestructible at all worlds at which it exists (factually necessary). Thus, suppose there is a world W at which a being x exists and has the properties of being uncaused and beginningless. Suppose further that, at W, there is no being y such that y that has what it takes to destroy or annihilate x. Then x has the world-indexed property of being-indestructible-at-W, and these facts about x at W ground the counterfactuals regarding x's indestructibility, so that (using Lewis-Stalnaker possible worlds semantics for counterfactuals) x is indestructible in all the worlds "closest" to W. X is thus a factually necessary being.
In short, it seems to me that my account of a factually necessary being is quite different from at least some notions of a brute fact. For unlike a being that has no explanation at all as to why it exists, a factually necessary being's existence is explained in terms of its own inner nature. In particular, its existence is explained in virtue of being beginningless and uncaused, and in virtue of the fact that nothing else that exists at W, or at any relevant world counterfactual to W, has what it takes to annihilate it. Now my question is, why in the world should we think that such a thing needs an explanation in terms of something beyond it? I think it's perfectly reasonable to say that it wouldn't be surprising in the least if it had no further explanation. But Craig will have none of that. No terminus of explanation will do besides a being that exists in all possible worlds. But surely this is explanatory overkill. For as my epistemically possible beinbg points out, there are less metaphysically extravagant, modally extreme candidates that can end the terminus of explanation in a satisfactory manner.
 Again the relevant epistemically possible beings here are uncaused, beings, and not self-caused beings, as Craig uncharitably attributes to the view I'm asserting at one point.
 Contrary to what Craig says, this scenario is a genuine threat to the plausibility of premise (1), and not a bizarre possibility on a par with, say, the possibility that our perceptual experiences are caused by Descartes' evil demon, and not the material world. For it's a genuinely live epistemic possibility (i.e., consistent with what we know or justifiedly believe) that dependent beings are explained in terms of metaphysically necessary beings or factually necessary beings. Indeed, Richard Swinburne is one of the leading Christian philosophers of religion alive, and he takes God to be a factually necessary being, as do many other Christian philosophers. So, contrary to what Craig says, it's not some far-fetched scenario trotted out solely by atheists and agnostics to avoid accepting the argument's conclusion. Rather, it's a genuinely live epistemic possibility. As such, it functions legitimately as an undercutting defeater for the premise.
 Indeed, it seems to me that Craig's restricted version of PSR is ad hoc. For it avoids van Inwagen's objection only by conceding that some states of affairs lack a sufficient reason for why they obtain. But if he allows this for states of affairs, he loses the grounds for denying it for objects: What principled basis could be offered for saying that objects require sufficient reasons, while states of affairs do not? In any case, it should be noted that even if Craig's restricted version of the principle avoids one of van Inwagen's criticisms, it falls prey to another of his criticisms. I made this point a few years ago, here.
 In his reply, Craig thinks the postulation of the fundamental stuff of material reality as necessary beings is really a concession to the theist. I'm not so sure, though, since such a reply seems to me to rely on the dubious assumptions of The Common Apologetic Strategy. But let us waive this objection. For it is a concession only if the necessary beings are taken to be metaphysically necessary, and not factually necessary. For of course matter-energy has been taken to be factually necessary among scientists ever since their discovery of the first law of thermodynamics.
 I should note that recent work in modal epistemology casts doubt on whether our modal judgments are reliable when they stray too far from the actual world. However, Craig thinks such modal skepticism is untenable:
We generally trust our modal intuitions on other familiar matters (for example, our sense that the planet Earth exists contingently, not necessarily, even though we have no experience of its non-existence). If we are to do otherwise with respect to the universe’s contingency, then the non-theist needs to provide some reason for his scepticism other than his desire to avoid theism.
Thus, Craig thinks that doubts about modal claims remote from ordinary experience relies on an arbitrarily selective form of modal skepticism, on the grounds that the demand for justification for exotic possibility claims should then apply to the humdrum possibility claims as well. And since we accept the latter without argument, we should, to be consistent, accept the former. But this objection is unpersuasive. For a number of plausible accounts of our knowledge of possibility have been proposed that allow for knowledge of humdrum metaphysical possibilities, while leaving exotic possibility claims unjustified. For example, it has been proposed that our knowledge of metaphysical possibilities is grounded in (i) our facility with counterfactual reasoning in ordinary contexts (Timothy Williamson), (ii) our folk theory of how the world works (Williamson, Rebecca Hanrahan, your truly), and (iii) arguments from analogy/relevant similarity with the actual world (Peter Hawke, yours truly). Such accounts can nicely explain the epistemic force of relatively uncontroversial thought experiments involving humdrum metaphysical possibilities (e.g., the Gettier cases), while leaving the more "far out" or exotic modal claims unjustified (e.g., "it is possible that I exist apart from my body", "it is possible that an Anselmian Being exists", etc.). Most saliently for our purposes, it appears that such accounts leave Craig's modal claim involving the possible non-existence of the fundamental constituents of reality unjustified: it's not clear how such a claim could be justified via relevant similarity with the actual world, our folk theory of how the actual world works, or our facility with counterfactual reasoning in ordinary contexts. In any case, my criticism doesn't rely on this point, as it allows that our modal judgements about our universe's contingency are reliable.
 A related concern is that Craig seems to assume a thing ontology (as opposed to a stuff ontology) of fundamental material reality in these thought experiments. Thus, consider Craig's thought experiment involving a universe composed of different quarks. The reasoning is supposed to be that the thought experiment provides evidence that the types of quarks composing our universe are contingent beings, and not (metaphysically) necessary beings. But the worry is that even if quarks are the most fundamental sorts of things in the universe, it's still an open question whether (i) these beings are, in turn, further composed of an even more fundamental stuff (matter-energy, say), and (ii) this fundamental stuff is metaphysically necessary.
Let me explain a bit. For simplicity's
sake, suppose there are just nine things (quarks, say), and these, in
turn, are composed of a more fundamental "stuff" (matter-energy, say).
Suppose further that the latter is a metaphysically necessary substance.
Finally, suppose the stuff is capable of an unlimited number of modes
of existing (e.g., as nine particles, as 18 smaller particles, as one
big particle, etc.). If so, then while the particles -- the things --
are contingent beings, matter-energy -- the "stuff" -- is not. But if
this scenario is epistemically possible, then Craig's reply here fails to rule out that he's conflating
the contingency of things within the universe with the contingency of
the stuff of which it's composed.
Now one might try to push Craig's point here by saying that we can adequately imagine the universe (or a
universe) as composed of different matter-energy, or of our universe
(or a universe) ceasing to exist (or of a possible world at which no
material objects exist), but at the very least, this isn't clear. For
even if one grants that imaginability can provide sufficient
justification for very many possibility claims, one might yet sensibly
worry that its justification-conferring ability does not extend to
states of affairs as remote from ordinary experience as the
non-existence of all matter-energy, or the existence of a different kind
of matter-energy. In this regard, she may find such claims to be on a
par with the controversial modal premise of (say) Plantinga’s modal
ontological argument (Possibly, an Anselmian Being exists), or of
conceivability arguments for dualism (Possibly, I exist apart from my
One might object that the previous criticism relies on an
arbitrarily selective form of modal skepticism, on the grounds that the
demand for justification for exotic possibility claims should then
apply to humdrum possibility claims as well. And since the sensible
non-theist accepts the latter without argument, she should, to be
consistent, accept the former. But this objection is less than
persuasive. For a number of plausible accounts of the epistemology of
modality have been proposed that provide a basis for distinguishing
between justified humdrum possibility claims and unjustified exotic
possibility claims. So, for example, it has been argued that our
knowledge of metaphysical possibility is grounded in (i) our
survival-conducive facility with counterfactual reasoning in ordinary
contexts (e.g., Williamson, Nichols) and (ii) arguments from
analogy/relevant similarity with the actual world (e.g., Hawke, yours
truly). Such theories receive confirmation in a number of ways, such as
their ability to explain the epistemic force of paradigm-case thought
experiments (e.g., Gettier cases), as well as the lack of such
conviction with respect to the more exotic modal thought experiments
(involving, e.g., the possible existence of Anselmian Beings and
disembodied existence). For the former thought experiment can be
grounded in such accounts, while the latter cannot.
saliently for our purposes, it’s not at all clear how such accounts of
our knowledge of metaphysical possibility could adequately support the
possible non-existence of all matter-energy, or of the possible
existence of a different kind of matter energy. Thus, it’s not clear how
the evolutionary pressures that gave rise to our competence with
counterfactual reasoning in daily life (e.g., reliable reasoning about
what would happen if one tried to cross a busy intersection) would make
us competent to determine something so remote from ordinary experience
as the possible non-existence of all matter-energy, or of a different
sort of matter-energy.
Nor are such possibility-candidates
sufficiently similar to our experience and knowledge of the actual world
so as to ground a solid analogical inference from the latter to the
former. For the relevant sorts of experiences here would involve
observations of the absolute origination and annihilation of
matter-energy. But in all of our actual experiences, what we observe is relevantly dissimilar to this, viz., the mere rearrangement of preexisting materials, as well as their decomposition into simpler elements.
It therefore appears that a non-theist could sensibly reject an
imaginability-possibility inference in support of such exotic claims
without thereby engaging in an unprincipled or arbitrarily selective
form of modal skepticism.
 There are several alternative models to the standard Big Bang model that don't involve the origin of the universe in a singularity, but here's one. According to M-theory (the theory that unifies the five versions of superstring theory), there are entities called 'branes', or n-dimensional membranes, and our universe is just one brane among many branes existing within a larger 11-dimensional space-time. Most saliently for our purposes, though, according to M-theory, the beginning of our universe is not the absolute beginning of concrete particulars, and the realm of concrete particulars may well have no beginning. On M-theory, then, the need for an absolute temporal beginning of concrete particulars doesn't arise. For a popular account of M-theory, see, e.g., Greene .
 A standard account of the conditions for proper appeal to authority is given in (e.g.) Trudy Govier's A Practical Study of Argument:
(i) The expert is reliable and credible in this context
(ii) Their claim falls within their area of specialization
(iii) The expert’s area of specialization is a genuine field of knowledge
(iv) The experts in that field agree about the claim in question
6th Edition (Wadsworth, 2005).
 It needs to be emphasized that while James Sinclair has degrees in the relevant sciences, he isn't a researcher in them. Indeed, he doesn't even have a Ph.D. in them. Rather, he's a war researcher for the U.S. Navy with a B.S. and an M.S. in physics.
 One final point about Craig's appeal to the Borde-Guth-Vilenkin theorem: In his dialogue with William Lane Craig on the kalam cosmological argument (a dialogue that occurred almost two years ago, and has not been released online, so far as I have been able to check), Wes Morriston notes that the Borde-Guth-Vilenkin theorem relies on the assumption that "the expansion rate of the universe never gets below some non-zero value, no matter how small. " (Alex Vilenkin, Many Worlds in One,, p. 175). But this assumption has recently been called into question by a number of scientists. So, for example, Aguirre and Gratton, and Carrol and Chen, have proposed models according to which the expansion rate falls below non-zero values.