What follows is a guest post by Peter Lupu with some additions and corrections by BV. 'CCB' abbreviates 'concrete contingent being.' The last post in this series is here. Thanks again to Vlastimil Vohamka for pointing us to Maitzen's article, which has proven to be stimulating indeed.
1. As a general rule, dummy sortals such as ‘thing’, ‘object’, ‘CCB’, etc., are not referential terms, unless there is an explicit or implicit background presupposition as to which sortal term is intended as a replacement. This presupposition, if satisfied, fixes the referent of the dummy sortal. In the absence of the satisfaction of such a presupposition, sentences in which they are used (not mentioned) have no truth-conditions and questions in which they are used (not mentioned) have no answer-conditions.
2. Examples such as ‘Cats are CCBs’ are no exception. Either this sentence has no truth-conditions because the term ‘CCB’ is merely a place holder for an unspecified sortal or it should be understood along the lines of: ‘Cats are animals’, etc., where ‘animal’ is (one possible) substitution term for the dummy sortal ‘CCB.'
BV adds: Right here I think a very simple objection can be brought against the semantic thesis. We know that cats exist, we know that they are concrete, and we know that they are contingent. So we know that 'Cats are concrete contingent beings' is true. Now whatever is true is meaningful (though not vice versa). Therefore, 'Cats are concrete contingent beings' is meaningful. Now if a sentence is meaningful, then its constituent terms are meaningful. Hence 'CCB' is meaningful despite its being a dummy sortal. I would also underscore a point I have made several times before. The immediate inference from the admittedly true (a) to (b) below is invalid:
a. The question 'How many CCBs are there?' is unanswerable, hence senseless
b. The question 'Why are there any CCBs?' is unanswerable, hence senseless.
3. The semantic thesis is the driving force behind Steve M’s view. It is the fallback position in all of his responses to challenges by Bill, Steven, and others. So far as I can tell, Steve M. did not defend the general form of the semantic thesis in his original paper. It is, therefore, surprising that it has been ignored by almost everyone in these discussions and that neither Bill nor Steven challenged the semantic thesis. I have written an extensive comment on this thesis and challenged it on several grounds.
B. Explanatory Thesis
1. As a general rule, Why-Questions are answered by giving an explanation. ‘Why are there any CCBs?’ is a [explanation-seeking] Why-Question. [It is worth noting that the grammatically interrogative form of words 'Why is there anything at all?' could be used simply to express wonder that anything at all should exist, and not as a demand for an explanation.] Therefore, it invites an explanation. What sort of explanation? Steve M. holds two theses about this last question:
(MI) The Adequacy Thesis: empirical explanations typical in science offer (at least in principle) adequate explanations for the Why-CCBs question, provided the Why-CCB questions are meaningful at all (and their meaningfulness is a function of satisfying the semantic thesis);
(MII) The Completeness Thesis: Once an empirical explanation is given to Why CCBs?, there is nothing left to explain. And in any case there are no suitable forms of explanation beyond empirical explanations that could be even relevant to explain Why-CCBs?
2. Bill and Steven certainly deny (MII). They may also have some reservations about MI. What is the basis on which Bill and Steven challenge MII? They maintain that even if we assume that an adequate empirical explanation is offered (i.e., MI is satisfied) to each and every CCB, there is something else left over to explain. What is that “something else” that is left over that needs explaining (Steve M. asks)?
3. It is at this juncture that the discussion either reverts back to the semantic thesis or it needs to be advanced into a new metaphysical realm.
C. Metaphysical Thesis
Dummy sortals do not pick out any properties or universals (monadic or relational) except via the mediation of genuine sortals. i.e., there are no properties over and beyond those picked out by genuine sortals.
1. Steven attempted to answer the challenge posed by the question at the end of B2 in one of his posts. His answer is this: what is left over after all empirical explanations favored by Steve M. are assumed to have been given is a very general property, feature, or aspect that all CCBs, and only CCBs, have in common. So why shouldn't ‘Why-CCBs’ questions be understood as inquiring into an explanation of this general feature that all and only CCBs share? Call this alleged general feature ‘X’.
2. The dispute has turned to whether X has any content, i.e., Steve M. challenged the contention that there is any phenomenon described by X that was not already accounted for by his favorite empirical explanations. Bill and Steven tried to articulate the content of X without (apparently) noticing that every such effort was rebutted by Steve M. either by appealing to the semantic thesis or to the explanatory thesis or both.
3. So what could X be? I suggest the following: X is the (second-order) property such that the property of *is a contingent being* is instantiated (or something along these lines). [I would put it this way: X is the being-instantiated of the property of being a contingent being.]
4. Since the universal/property *is a contingent being* need not be instantiated, the fact that it is in fact instantiated in the actual world (i.e., that X holds) needs explaining (So claim Bill and Steven). And whatever is the explanation (including a “brute-fact” explanation) for this fact, it cannot take the form of an empirical explanation.
5. The Metaphysical Thesis I am attributing to Steve M. of course rules out that there is a property such as X. Why? Two reasons: first, the property *is a contingent being* is not a sortal property; second, the predicate ‘is a contingent being’ (or any of its variants) contains a dummy sortal and therefore it does not pick out a property (nor does it have an extension) in the absence of a specific background presupposition of a specific sortal substituend.
Unless these three theses are clearly separated, the discussion will be going in circles. As one can see, the driving force behind the explanatory and metaphysical theses is ultimately the semantic thesis. No one challenged this thesis directly (except me in a comment that was ignored by everyone with the exception of Bill).
It is the thesis of Stephen Maitzen's Stop Asking Why There's Anything that the Leibniz question, 'Why is there anything, rather than nothing at all?' is ill-posed as it stands and unanswerable. Maitzen's point is intended to apply not only to the 'wide-open' formulation just mentioned but also to such other formulations as 'Why are there any concrete contingent beings at all?' I will discuss only the latter formulation. It is defensible in ways that the wide-open question is not. Call it the modified Leibniz question. For Maitzen it is a pseudo-question. For me it is a genuine question. On my classificatory scheme, Maitzen is a rejectionist concerning the modified Leibniz question. The question is not to be answered but rejected as senseless, because of an internal semantic defect that renders it necessarily unanswerable and therefore illegitimate as a question.
My defense of the meaningfulness of the modified Leibniz question does not commit me to any particular answer to the question such as the theistic answer. For there are several possible types of answer, one of them being the 'brutal' answer: it is simply a brute fact that concrete contingent beings (CCBs) exist. When Russell, in his famous BBC debate with the Jesuit Copleston, said that the the universe is just there and that is all, he was answering the question, not rejecting it. His answer presupposed the meaningfulness of the question.
1. Getting a Sense of What the Dispute is About
Maitzen's paper is in the context of a defense of naturalism and an attack on theism. So I have to be careful not to assume theism or anything that entails or presupposes theism. Defining 'naturalism' is a tricky business but it suffices for present purposes to say that naturalism entails the nonexistence of God as classically conceived, and the nonexistence of immortal souls, but does not entail the nonexistence of abstracta, many of which are necessary beings.
To make things hard on theists let us assume (contrary to current cosmology) that the universe has an actually infinite past. Hence it always existed. Let us also assume that the each total state of the universe at a time is (deterministically) caused to exist by an earlier such state of the universe. A third assumption is that the universe is nothing over and above the sum of its states. The third assumption implies that if each state has a causal explanation in terms of earlier states (in accordance with the laws of nature), then all of the states have an explanation, in which case the universe itself has a causal explanation. This in turn implies that there is no need to posit anything external to the universe, such as God, to explain why the universe exists. The idea, then, is that the universe exists because it causes itself to exist in that later states are caused to exist by earlier states, there being no earliest, uncaused, state. We thereby explain why the universe exists via an infinite regress of universe-immanent causes and in so doing obviate the need for a transcendent cause.
If this could be made to work, then we would have a nice neat self-contained universe whose existence was not a brute fact but also not dependent on anything external to the universe. We would also have an answer to the modified Leibniz question. Why are there any CCBs given the (broadly logical) possibility that there not be any? Because each is caused to exist by other CCBs.
The five or so assumptions behind this reasoning can all be questioned. But even if they are all true, the argument is still no good for a fairly obvious (to me!) reason. The whole collection of states, despite its being beginningless and endless, is (modally) contingent: it might not have existed at all. So, despite every state's having a cause, we can still ask why there are any states in the first place.
The fact that U always existed, if it is a fact, does not entail that U must exist. If I want to know why this universe of ours exists as opposed to there being some other universe or no universe at all, it does no good to tell me that it always existed. For what I want to know is why it exists at all, or 'in the first place.' I am not asking about its temporal duration but about its very existence. Why it exists at all is a legitimate question since there is no necessity that there be a universe in the first place. There might have been no universe, where 'universe' stands for the sum-total of concrete contingent beings all of which, on the assumption of naturalism, are physical or material beings. And it seems obvious that the fact, if it is a fact, that every state has a cause in earlier states does not explain why there is the whole system of states.
The dispute between Maitzen and me can now be formulated.
BV: The question 'Why are there any CCBs at all? is a legitimate question ( a meaningful question) that cannot be answered in a universe-immanent or naturalistic way as above where every CCB is causally explained by other CCBs.
SM: The question 'Why are there any CCBs at all?' is not a legitimate question (not a meaningful question) except insofar as it can be reformulated as a question whose answer can take a universe-immanent or naturalistic form.
2. Maitzen's Argument For the Meaninglessness of the Modified Leibniz Question
The argument begins with considerations about counting. Maitzen arrives at a result that I do not question. We can counts pens, plums and penguins, but we cannot count things, entities, or concrete contingent beings. Or at least we cannot count them under those heads. The reason is quite simple. The first trio of terms is a trio of sortals, the second of dummy sortals. Sortals encapsulate individuative criteria that make possible the counting of the items to which the sortals apply. Thus it makes sense to ask how many cats are on my desk. The answer at the moment is two. But it makes no sense to ask how many CCBs are on my desk at the moment. For to answer the question I would have to be able to count the CCBs, and that is something I cannot do because of the semantic indeterminacy of 'CCB.' When one counts cats one does not count the proper parts of cats for the simple reason that the proper parts of cats are not cats. (Pre-born babies inside a mother are not proper parts of the mother.) In fact, it occurs to me now that a necessary condition of a term's being a sortal is that it be such that, if it applies to a thing, then it does not apply to the proper parts of the thing. When I set out to count CCBs, however, I get no guidance from the term: I don't know whether to count the proper parts of the cat as CCBs or not. It is not that I or we contingently lack the ability to count them, but that the semantic indeterminateness of 'CCB' makes it impossible to count them. Things get even hairier -- you will forgive the pun -- when we ask about undetached arbitrary parts (e.g., Manny minus his tail) and mereological sums (e.g., Manny + the cigar in the ashtray).
All of this was discussed in greater detail in earlier posts. For now the point is simply that the question 'How many CCBs are there?' cannot be answered due to the semantic indeterminateness of 'CCB.' And since it cannot be answered for this semantic reason, the question is senseless, a pseudo-question.
So far, so good. But then on p. 56 of Maitzen's paper we find the following sudden but crucial move: "These considerations, I believe, also show that the question ‘Why is there anything?' (i.e., ‘Why is there any thing?’) confuses grammatical and logical function and hence necessarily lacks an answer . . . . " The main weakness of Maitzen's paper, as I see it, is that he doesn't adequately explain the inferential connection between the counting question and the explanation question, between the 'How many?' question and the 'Why any?' question. I cheerfully concede that it is senseless to ask how many CCBs there are if all we have to go on is 'CCB' as it is commonly understood. (Of course there is a difference between 'thing,' say, and 'concrete contingent being.' The first is a bit of ordinary English while the second is a term of art (terminus technicus). But this difference does not make a difference for present purposes.) But why should the fact that 'CCB' is a dummy sortal also make the 'Why any?' question senseless? For that is precisely what Maitzen is claiming. 'Why is there anything?' is senseless because "the question's reliance on the dummy sortal 'thing' leaves it indeterminate what's being asked." (p. 56)
But wait a minute. What is being asked about CCBs in the second question is not how many, but why they exist at all. Why should the fact that we cannot assign a precise number to them render the second question senseless? I know that there are at least two CCBs. Here is one cat, here is another (he said Mooreanly). Each is a concrete contingent being. So there are at least two. If there are at least two, then there are some. If there are some, then 'CCBs exist' is true. Since it is true, it is meaningful. (Not every meaningful proposition is true, but every true proposition is meaningful.)
To put it another way, 'CCBs exist' is a (closed) sentence. It expresses a complete thought, a proposition. It is not an open sentence like 'Xs exist.' The latter is no more a sentence than a dummy sortal is a sortal. Unlike 'CCBs exist,' it cannot be evaluated as either true or false. So, while 'CCB' lacks the semantic determinacy of a sortal, it is not wholly semantically indeterminate like the variable 'X.' It makes a semantic contribution to the sentence 'CCBs exist.'
Now if it is meaningful to assert that CCBs exist, despite their number being indeterminate, then it is also meaningful to ask why CCBs exist, despite their number being indeterminate. Now it is meaningful to assert that CCBs exist. Therefore, it is meaningful to ask why they exist, despite their number being indeterminate.
Although the uncountability of CCBs is a good reason to think that 'How many CCBs are there?' is senseless, it is not a good reason to think that 'Why are there any CCBs?' is also senseless.
My point is that it is a non sequitur for Maitzen to move from
a. 'How many CCBs are there?' is a senseless question
b. 'Why are there any CCBs?' is a senseless question.
(a) is true. But one can hold (a) consistently with holding the negation of (b).
How might Maitzen respond?
3. 'Concrete Contingent Being' as a Mere Covering Term
For Maitzen, 'CCB' is "only a covering term for pens, plums, penguins . . . ." (p. 57) and other instances of sorts. It doesn't refer to anything distinct from pens, plums, penguins, cats, human births, explosions, and so on. In other words, 'CCB' does not pick out a special sort -- an uber-sort, if you will -- the instances of which are distinct from the instances of genuine sorts. And so 'CCB' does not pick out a sort whose instances elude natural-scientific explanation and therefore EITHER require some special explanation by God or some other entity transcendent of the physical universe OR are such that their existence is a brute fact. As Maitzen puts it, "there aren't any contingent things whose explanations outstrip the explanations available for the individuals covered by the covering term 'contingent things.'" (p. 58) The 'Why any?' question "has no content until we replace referentially indeterminate words with genuine sortals." (p. 59)
If Maitzen is telling us that CCBs are not a sort of thing distinct from ordinary sorts, then he is right, and I agree. Suppose we we have a complete list of all the sorts of thing in the universe: pens, plums, pussycats, penguins, and so on. It would be absurd if someone were to object: "But you forgot to list the concrete contingent beings!" That would be absurd since each pen, plum etc. is a CCB, and there is no CCB that is not either a pen or a plum or, etc. But it doesn't follow that a sentence in which 'CCB' occurs is without content.
It is simply false to say that the 'Why any?' question "has no content until we replace referentially indeterminate words with genuine sortals." (p. 59) Right here is where Maitzen makes the mistake that invalidates the move from (a) to (b). He conflates the partial semantic indeterminacy of dummy sortals with the total semantic indeterminacy of variables. Compare:
Why are there any penguins?
Why are there any concrete contingent beings?
Why are there any Xs?
The first two questions are genuine, despte the fact we can count only penguins. The third question is pseudo since it has no definite sense.
Note finally that we cannot replace the second question with a long disjunctive question like 'Why are there either penguins or plums or pussycats or pens, or . . . ?' For suppose you had a complete naturalistic answer to the latter question. You could still meaningfully ask why there are any CCBs at all as opposed to none at all, and why these rather than some other possible set.
There is more to say, but tomorrow's another day, and brevity is the soul of blog.
Readers who have stuck with me over the years will remember commenter 'Spur' whose comments were the best I received at the old Powerblogs site. Safely ensconced in an academic position, he now enters the blogosphere under his real name, Stephen Puryear. His weblog is entitled Second Thoughts.
Pedant and quibbler that I am, it annoys me when I hear professional philosophers use the phrase 'Leibniz's Law.' My reason is that it is used by said philosophers in three mutually incompatible ways. That makes it a junk phrase, a wastebasket expression, one to be avoided. Some use it as Dale Tuggy does, here, to refer to the Indiscernibility of Identicals, a principle than which no more luminous can be conceived. (Roughly, if a = b, then whatever is true of a is true of b, and vice versa.) Fred Sommers, referencing Benson Mates, also uses it in this way. (See The Logic of Natural Language, p. 127)
Others, such as the distinguished Australian philosopher Peter Forrest, use it to refer to the Identity of Indiscernibles, a principle rather less luminous to the intellect and, in my humble opinion, false. (Roughly, if whatever is true of a is true of b and vice versa, then a = b.) And there are those who use it as to refer to the conjunction of the Indiscernibility of Identicals and the Identity of Indiscernibles.
So 'Leibniz's Law' has no standardly accepted usage and is insofar forth useless. And unnecessary. You mean 'Indiscernibility of Identicals'? Then say that. If you mean its converse, say that. Ditto for their conjunction.
There is also the problem of using a great philosopher's name to label a principle that the philosopher may not even have held. Analytic philosophers are notorious for being lousy historians. Not all of them, of course, but the run-of-the-mill. If Sommers is right, Leibniz was a traditional logician who did not think of identity as a relation as Frege and Russell do. (p. 127) Accordingly, 'a = b' as this formula is understood in modern predicate logic does not occur in Leibniz.
No doubt you have heard of Hume's Fork. 'Fork,' presumably from the Latin furca, suggests a bifurcation, a division; in this case of meaningful statements into two mutually exclusive and jointly exhaustive classes, the one consisting of relations of ideas, the other of matters of fact. In the Enquiry, Hume writes:
Propositions of this kind [relations of ideas] can be discovered purely by thinking, with no need to attend to anything that actually exists anywhere in the universe. . . . Matters of fact . . . are not established in the same way; and we cannot have such strong grounds for thinking them true. The contrary of every matter of fact is still possible, because it doesn't imply a contradiction and is conceived by the mind as easily and clearly as if it conformed perfectly to reality. That the sun will not rise tomorrow is just as intelligible as - and no more contradictory than - the proposition that the sun will rise tomorrow.
One question that arises is whether Hume's Fork was anticipated by any earlier philosopher. Leibniz of course makes a distinction between truths of reason and truths of fact that is very similar to Hume's distinction between relations of ideas and matters of fact. See, for example, Monadology #33. In a very astute comment from the old blog, 'Spur' details the similarities and concludes:
Leibniz and Hume have the same basic distinction in mind, between those truths which are necessary and can be known a priori, and those which are contingent and can only be known a posteriori. The two philosophers use slightly different terminology, and Leibniz would balk at Hume's use of 'relations between ideas' in connection with truths of reason only, but the basic distinction seems to me to be the same.
I deny that the basic distinction is the same and I base my denial on a fact that Spur will admit, namely, that for Leibniz, every proposition is analytic in that every (true) proposition is such that the predicate is contained in the subject: Praedicatum inesse subjecto verae propositionis. I argue as follows. Since for Leibniz every truth is analytic, while for Hume some truths are analytic and some are not, the two distinctions cannot be the same. To this, the Spurian (I do not say Spurious) response is:
The [Leibnizian] distinction is between two kinds of analytic truths: those that can be finitely analyzed, and those that can't. This is an absolute distinction and there are no truths that belong to both classes. Even from God's point of view there is presumably an absolute distinction between necessary and contingent truths, though perhaps he wouldn't view this as a distinction between finitely and non-finitely analyzable truths, because his knowledge of truths is intuitive and never involves analysis.
I grant that the two kinds of Leibnizian analytic truths form mutually exclusive and jointly exhaustive classes. But I deny that this suffices to show that "the same basic distinction" is to be found in both Leibniz and Hume.
One consideration is that they do not form the same mutually exclusive and jointly exhaustive classes. Though every Humean relation of ideas is a Leibnizian truth of reason, the converse does not hold. I think Spur will agree to this. But if he does, then surely this shows that the two distinctions are not the same. I should think that extensional sameness is necessary, though not sufficient, for sameness.
But even if the two distinctions were extensionally the same, they are not 'intensionally' the same distinction.
Consider Judas is Judas and Judas betrays Christ. For both philosophers, the first proposition is necessary and the second is contingent. But Leibniz and Hume cannot mean the same by 'contingent.' If you negate the first, the result is a contradiction, and both philosophers would agree that it is, and that it doesn't matter whether the proposition is viewed from a divine or a human point of view. The negation of the second, however, is, from God's point of view a contradiction for Leibniz, but not for Hume. For Leibniz, the betrayal of Christ is included within the complete individual concept of Judas that God has before his mind. So if God entertains the proposition Judas does not betray Christ, he sees immediately that it is self-contradictory in the same way that I see immediately that The meanest man in Fargo, North Dakota is not mean is self-contradictory.
Of course, for Leibniz, it is contingent that Judas exists: there are possible worlds in which Judas does not exist. But given that Judas does exist, he has all his properties essentially. Thus Judas betrays Christ is contingent only in an epistemic sense: we finite intellects see no contradiction when we entertain the negation of the proposition in question. Given our finitude, our concepts of individuals cannot be complete: they cannot include every property, monadic and relational, of individuals. But if, per impossibile, we could ascend to the divine standpoint, and if every truth is analytic (as Leibniz in effect holds via his predicate-in-subject principle), then we would see that Judas betrays Christ is conditionally necessary: necessary given the existence of Judas.
'Contingent' therefore means different things for Leibniz and Hume. Contingency in Hume cuts deeper. Not only is the existence of Judas contingent, it is also contingent that he has the properties he has. This is a contingency rooted in reality and not merely in our ignorance.
Perhaps my point could be put as follows. The Leibnizian distinction is not absolute in the sense that, relative to the absolute point of view, God's point of view, the distinction collapses. For God, both of the Judas propositions cited above are analytic, both are necessarily true (given the existence of Judas), and both are knowable a priori. But for Hume, the distinction is absolute in that there is no point of view relative to which the distinction collapses.
I'm stretching now, but I think one could say that, even if Hume admitted God into his system, he would say that not even for God is a matter of fact knowable a priori. For the empiricist Hume the world is radically contingent in a way it could not be for Leibniz the rationalist.
Leibniz's Theodicy consists of two parts, the first on faith and reason, the second on the freedom of man in the origin of evil. I am trying to understand paragraph #37 (p. 144 of the Huggard translation):
. . it follows not that what is foreseen is necessary, for necessary truth is that whereof the contrary is impossible or implies contradiction. Now this truth which states that I shall write tomorrow is not of that nature, it is not necessary. Yet supposing that God foresees it, it is necessary that it come to pass; that is, the consequence is necessary, namely, that it exist, since it has been foreseen; for God is infallible. This is what is termed a hypothetical necessity. But our concern is not this necessity: it is an absolute necessity that is required, to be able to say that an action is necessary, that it is not contingent, that it is not the effect of a free choice.
Clearly, the proposition P expressed by 'BV writes on 13 June 2009' is logically contingent. There is no logical necessity that I write tomorrow or on any day. Both my writing tomorrow and my not writing tomorrow are logically possible. But given that God foreknows that P, P must be true. That is,
1. Necessarily (if God foreknows that P, then P is true).
We note that the necessity in (1) attaches to the conditional, not to its consequent. This is a case, then, of necessitas consequentiae, not of necessitas consequentiis. In Leibniz's jargon, (1) is a case of hypothetical necessity as opposed to absolute necessity. The consequence is necessary, not the consequent. From (1) one cannot infer
2. If God foreknows that P, then necessarily P is true.
So far, so good. If a proposition is known, by God or by anyone, then it must be true; but that is consistent with saying that the proposition known is contingently true. Given that I know that I am blogging, then I must be blogging; but that is not to say that I am necessarily blogging: I might not have been blogging now.
What I don't understand, though, is the last sentence in the passage quoted. The last sentence strikes me as false. Leibniz seems not to appreciate that if a contingent state of affairs is necessitated by something other than the agent, then there is a prima facie difficulty about reconciling it with freedom of choice. The source of necessitation might be divine foreknowledge (theological fatalism), or the laws of logic (logical fatalism), or the past and the laws of nature (causal determinism). No matter what the source of necessitation, one cannot dissolve the problem of reconciling free will and the necessitation of the act willed simply by pointing out the difference between hypothetical and absolute necessity.
In other words, Leibniz appears to be taxing the fatalist and the determinist with a sophomoric error, namely, that of confusing (1) and (2) above. But no sophisticated fatalist or determinist need make that error. It is clear that my blogging now is a logically contingent state of affairs. But if determinism is true, then it is not nomologically possible that I be doing anything other than blogging now: past events under the aegis of the laws of nature necessitate my blogging now. How then can my blogging now be free? What Leibniz fails to see is that simply distinguishing the necessity of the consequence from the necessity of the consequent does nothing to answer the question.
I have been searching the 'Net and various databases such as JSTOR without success for a good article on deus ex machina objections in philosophy. What exactly is a deus ex machina (DEM)? When one taxes a theory or an explanatory posit with DEM, what exactly is one alleging? How does a DEM differ from a legitimate philosophical explanation that invokes divine or some other nonnaturalistic agency? Since it is presumably the case that not every recourse to divine agency in philosophical theories is a DEM, what exactly distinguishes legitimate recourse to divine agency from DEM? Does anyone have any references for me? Herewith, some preliminary exploratory notes on deus ex machina.