Prominent Christian philosopher, apologist, and analytic theologian William Lane Craig is in the process of releasing his magnum opus, a 5-volume Systematic Philosophical Theology (Wiley Blackwell, 2025–2026). Volume 2a (released in 2025) focuses on God’s attributes and includes a 100-page chapter on omniscience. Now aged 76, Craig presumably intends this chapter to be the crowning summary of his nearly 40-year publication record on the topic of divine omniscience and foreknowledge.
In this 4-part blog series I review and critique Craig’s omniscience chapter. The chapter is divided into four main sections—§1 Biblical data, §2 Omniscience, §3 Foreknowledge, and §4 Middle Knowledge—which correspond to the four parts of this blog series. This post is on §4. For my previous posts on §1, §2, and §3, see here, here, and here.
Part 4. Molinism (pp. 251–284)
The final section of Craig’s omniscience chapter is all about his favorite theory of divine providence, i.e., Molinism. This section is divided into three main parts: (1) explicating Molinism, (2) arguing for Molinism, and (3) rebutting arguments against Molinism.
(1) Explicating Molinism
(1.a) How Molinism works
Craig’s description of Molinism is straightforward. He goes into the historical background, the driving motivations, and the basic framework. There’s nothing new or controversial here, so to make things easy on myself and also to introduce greater precision than Craig himself does, I’m going to borrow from a post I wrote previously:
The term “Molinism” comes from Jesuit theologian Luis de Molina (1535–1600), who sought to understand how God could exercise meticulous providence without compromising human libertarian free will (LFW). Meticulous providence is the idea that every detail of creation, across its entire history (past, present, and future), has been specifically selected and ordained by God. As for LFW, this is the idea that at least some of the time some of God’s creatures, humans especially, can exercise in an original way an intrinsic power to choose among multiple open, live options. Because this power is exercised in an original way, it is incompatible with determination by outside causes, whether those be secondary (i.e., creature-to-creature) or primary (i.e., God-to-creature). The obvious problem is that these two views seem to be in direct conflict with each other. If God providentially specifies everything creatures do, then how can there be any leeway left over for creatures to choose their own course of action? (Analogy: If the scriptwriter writes all the lines, spells out all the stage directions in minute detail, and the play has to go exactly as he specifies, then how can the actors still have any freedom with respect to how they play their parts?)
Molina’s proposed solution to this dilemma centers around the novel idea that God has middle knowledge (MK). Let’s break this down. First, MK is non-natural because it deals with logically contingent truths that are independent of God’s absolutely necessary nature. Second, MK is pre-volitional because it is explanatorily prior to any act of God’s will, including God’s free decision to create. Third, because MK is explanatorily prior to God’s will, it constrains God’s options with respect to creation. God cannot bring about, directly or indirectly, anything that conflicts with His MK.
As to the content of God’s MK, it assigns determinate truth values to all settled conditional future contingents (CFCs). A settled CFC specifies for some internally consistent and fully causally specified indeterministic scenario (S) and some outcome (O) that is causally contingent with respect to that scenario, either that O would occur if S or that O would not occur if S. For example, suppose S involves Bob’s being about to make a LFW choice between chocolate and vanilla ice cream. For S to be “fully causally specified” its antecedent has to specific all causally relevant information including Bob’s motives and deliberations up the very moment before Bob’s choice. If we assume that chocolate and vanilla are Bob’s only options (he can’t abstain and he can’t have both), then Molinism says that either <If S were realized then Bob would freely choose chocolate> or <If S were realized then Bob would freely choose vanilla> is true. Furthermore, there are true conditionals of this sort for any internally consistent and fully causally specified indeterministic scenario and any partition of causally contingent outcomes given that scenario. So, for example, we could replace Bob in the ice cream scenario with any other hypothetical individual (Mary, Peter, etc.), changing only what needs changing to reflect that individual’s causal history, and there would be (according to Molinism) a determinate truth as to what that individual would freely do in the revised scenario.
Note 1: Many discussions of Molinism focus on so-called “counterfactuals of creaturely freedom” or CCFs. But focusing exclusively on creaturely freedom is too narrow. To support God’s meticulous providence, MK needs to provide God determinate information about the outcomes of every conceivable indeterministic scenario, even those that do not involve free will, such as, say, quantum-level events. That’s why I explicate Molinism in terms of CFCs rather than CCFs.
Note 2: In the previous post in this series I distinguished between settled future contingent propositions (SFCPs) and open future contingent propositions (OFCPs). A similar distinction can be made here. Thus, the Molinist’s CFCs are really settled CFCs (SCFCs). They present the consequent as following determinately (albeit contingently) from the antecedent. This is why they are expressed in terms of what “would” happen given some FCSI scenario. Molinists generally ignore open CFCs (OCFCs). These are standardly expressed in terms of what “might” (and might not) happen given some scenario.
Finally, God uses MK to inform His decision about what sort of world, if any, to create. And because (according to Molinism) God has exhaustive MK of all true SCFCs, He can exercise meticulous control over every detail of creation throughout its entire history subject only to the constraints that MK places on God’s options. God achieves this meticulous control by considering various initial conditions, causal laws, and interaction policies. Together these define a “fully causally indeterministic specified” (FCSI) scenario. By His MK God knows exactly what “would freely” happen where He to insert one or more specific free creatures (say, Adam and Eve) into that scenario. That “would freely” outcome transforms the initial scenario into a new FCSI scenario in which, again, by His MK God knows exactly what “would freely” happen. And so on. By using His MK and imaginatively chaining SCFCs together, with the “would freely” consequent of one becoming the FCSI antecedent for the next, God can meticulously map out an entire timeline of creation history. After comparing various such timelines, God decides which one, if any, He wants to implement. And that, according to Molinism, is how God obtains exhaustively definite foreknowledge (EDF) and exercises meticulous providence without causally determining what creatures “would freely” do in FCSI scenarios.
(1.b) Problems with Craig’s and Molina’s explications of Molinism
While reading this subsection of Craig’s omniscience chapter, I noticed a couple problems with both Craig’s and Molina’s explications of Molinism. Craig quotes from Molina extensively and rather uncritically, and so tends to copy Molina’s mistakes.
The first problem is their tendency to think of middle knowledge as innate or intrinsic to God. This is a problem because, as MK is non-natural, contingent, and pre-volitional it seems there is nothing in God to which MK can correspond. What in God specifies for a given FCSI scenario S with causally contingent outcome O whether O “would freely” result or “would freely” not result were S to obtain? As non-natural it can’t be anything in God’s essence. As contingent it can’t be anything in God’s natural or non-contingent energies. As pre-volitional it can’t be anything in God’s free will or contingent energies. (See this post for explanation of the essence–energies distinction.) I’ll call this Problem 1 (P1).
The second problem is that Craig and Molina repeatedly talk about merely possible creatures as if they were actual creatures with an actual ability to exercise LFW. This is a problem because merely possible creatures are just ideas in God’s mind. They have no actual will of their own by which they can specify for themselves whether outcome O “would freely” result or “would freely” not result where S to obtain. I’ll call this Problem 2 (P2).
P1 and P2 are closely related. Indeed, together they form the basis of the infamous grounding objection against Molinism to be discussed below. All I’d like to note for now is that both Craig and Molina walk right into P1 and P2, seemingly without realizing that there is anything problematic about them. In some passages both problems occur side by side. Thus, in Molina’s words, God has MK because
in virtue of the most profound and inscrutable comprehension of each free will [P2], he saw in his own essence [P1] what each such will would do with its innate freedom [P2] were it to be placed in this or that or, indeed in infinitely many orders of things—even though it would really be able [P2], if it so willed, to do the opposite. (p. 255)
In his gloss on this and another Molina passage Craig says that “the content of divine middle knowledge … depends on what the creatures themselves [P2] would do” (p. 256), that “such knowledge depends on a decision of free will [P2]” (p. 257) by a creature, and that God by His MK takes into account “the free decisions of creatures [P2] in his planning” (p. 258).
In another passage from Molina quoted by Craig, Molina again speaks of “God’s comprehension in his essence [P1] of each created free will [P2] through his natural [i.e., middle] knowledge” (p. 258). Note that Molina himself seemingly confuses MK with natural knowledge at this point, hence the editorial correction by Craig.
At this point Craig poses a very sensible question: How is it that “in knowing his own essence alone God is able to have middle knowledge” (p. 258)? He then approvingly cites Molina’s answer: God has supercomprehension. More specifically, God “comprehends in the deepest and most eminent way whatever falls under his omnipotence [P1], to penetrate created free choice [P2] in such a way as to discern and intuit with certainty which part it is going to turn itself to by its own innate freedom [P2]” (p. 259) As Craig explains,
Because his [God’s] intellect is infinite, whereas a free creature is finite, God’s insight into the will of a free creature [P2] is of such surpassing quality that God knows exactly what the free creature [P2] would do were God to place him in a certain set of circumstances. (p. 259)
In a footnote, Craig further suggests that we “think of Molina’s doctrine of supercomprehension in terms of individual essences’ having contingent counterfactual properties discerned by God” (p. 259). As to how an individual essence could even have contingent properties, Craig doesn’t say.
In parallel with Molina’s supercomprehension idea, Craig also approvingly references Francisco Suárez’s (1548–1617) suggestion that God knows MK simply because He essentially knows all truths and equates this with his own “conceptualist model” of divine omniscience (p. 259). (See §2.e of the previous post in this series for my discussion of this so-called “model.”)
What are we to make of all this? Molina and Craig need both P1 and P2 to make Molinism sound halfway plausible. Without P1 it follows that contingent MK information comes from outside God and constrains God’s own omnipotence. For any standard creation ex nihilo (CEN)-affirming monotheist, this is absurd. According to CEN, apart from creation there is nothing but God and God alone, and so there can be no contingent information external to God. Since Molina and Craig both want to be orthodox Christian monotheists, they therefore somehow have to locate MK information in God. Hence, P1. But, as I have already explained, P1 is incredibly problematic because there doesn’t seem to be any locus in God from which that information could be sourced.
As for P2, Molinists need this to avoid their view collapsing into determinism. If we don’t reify merely possible creatures or individual creaturely essences into quasi-actual free agents, then the contingent specification of MK toward a “would freely” as opposed to a “would freely not” (or vice-versa) is not a specification that comes from that creaturely individual or any creaturely individual. It’s simply God imagining things as if so-and-so “would freely” do such-and-such in an imagined FCSI scenario. This is similar to J. K. Rowling’s imagining Harry Potter’s “freely” choosing to do such-and-such in some scenario. But imagined freedom is obviously not real freedom. Hence, P2. But, as I have already explained, P2 is incredibly problematic because God’s innate ideas of possible creatures are not creatures, much less free creaturely agents.
(2) Arguments for Molinism
Craig offers three lines of argument for thinking that God has MK. The first is biblical. The second is philosophical. And the third is “theological” or, more accurately, pragmatic.
(2.a) A biblical argument for MK
Craig’s biblical argument for MK is brief and inconclusive. He refers back to his biblical survey (discussed in part one of this series) but then quickly concedes that “this does not settle the matter” of whether God has MK. This is because “the scriptural passages show only that God possesses counterfactual knowledge” (p. 260) and are silent on whether this knowledge is pre-volitional, as Molinism requires. Consequently, “no amount of proof-texting” can prove that God’s counterfactual knowledge “is possessed logically prior to God’s creative decree” (p. 261).
So far, the cumulative case for MK looks very weak.
(2.b) A philosophical argument for MK
Craig’s main argument for Molinism is philosophical. Here’s the core argument (adapted from Craig, p. 262):
(1) If there are true SCFCs, then God knows them.
(2) There are true SCFCs.
(3) If God knows a true SCFC, then He knows it either pre-volitionally or post-volitionally.
(4) It is not the case that God knows true SCFCs post-volitionally.
Therefore,
(5) God knows true SCFCs pre-volitionally. (from 1–4)
I’ll grant Craig the validity of this argument. If premises (1)–(4) are all true, then (5) logically follows and with it the core thesis of Molinism. The question, then, is whether (1)–(4) are all true.
As for (1), I will not contest it. It follows from God’s propositional omniscience (though see my second post in this series for discussion of the limits of propositional omniscience).
Premise (2) is just plain false. There are no true SCFCs. I’ll give my argument against (2) in the course of reviewing how Craig tries to support it. He proposes three arguments. I’ll take them in reverse order.
First, Craig says that “Scripture itself gives examples of … true counterfactuals,” such as 1 Cor. 2:8 (p. 263). But this is cheating on Craig’s part. He shifts the target from a specific type of counterfactual (i.e., an SCFC) to counterfactuals in general. If Scripture is to provide support for Molinism, it needs to provide examples of true SCFCs, not merely true counterfactuals. And as Craig himself has already conceded in his discussion of the Biblical case for MK (see §2.a above), there are no Biblical prooftexts for Molinism. So, Craig’s first supporting reason for (2) is irrelevant.
Second, Craig argues that “we ourselves often know the truth of counterfactuals about how people would act or react under particular circumstances” (p. 263). I grant the claim but deny its relevance. Notice that Craig again speaks generically here of “counterfactuals” rather than specifically of SCFCs. As before, this is cheating on his part, for it doesn’t matter if we know the truth of some counterfactuals if none of the counterfactuals we know are SCFCs. Indeed, it is far from obvious that we do know any SCFCs for, in general, the more confident we are that some “would” counterfactual is true, the less confident we should be that its “would freely” correlate is true. The reason for this will soon become apparent.
Side note: Craig’s “cheating” with these first two reasons is so blatant that it calls out for explanation. One theory is that it’s deliberate deception on his part. He appeals to counterfactuals in general rather than Molinist-style SCFCs in hopes that readers won’t notice the switch. A more charitable theory is that he genuinely believes there is no additional difficulty when it comes to SCFCs being true and knowable than there is for counterfactuals in general. I suspect the latter is closer to the truth.
Third, Craig says “it is plausible that” there are true SCFCs because the antecedents of these conditionals are fully causally specified (p. 263). This is a pretty weak claim on Craig’s part. He is correct that a “fully specified” antecedent removes one way in which a counterfactual can fail to have a determinate truth value. In a footnote he references the famous pair of examples from David Lewis: <If Verdi and Bizet were compatriots, they would both be French> and <If Verdi and Bizet were compatriots, they would both be Italian>. Lewis’s examples seem to have no determinate truth values because we have no reason for thinking either conditional is more likely to be true (or false) than the other. But we can easily change this perception by further specifying the antecedent. For example, if the antecedent said “If Verdi and Bizet were compatriots and Bizet was born in Paris” then we would be much more inclined to conclude that they would both be French. So antecedent strengthening, as it is called, can resolve counterfactual ambiguity. But this is not the problem with SCFCs. The problem with SCFCs is not the specificity of the antecedent but the contingency of the connection between antecedent (A) and consequent (C). This contingency is essential to Molinism. Given any FCSI scenario it must be really possible for a causally contingent outcome (relative to that scenario) both to occur and not to occur. Hence, as far as the FCSI antecedent is concerned, it remains an open question whether C “would freely” occur or “would freely not” occur given A. That’s the problem. The problem in other words, is that the Molinist’s SCFCs don’t cohere with the requite contingent relation between antecedent and consequent. If it’s an open question whether C “would freely” occur given A, then it can’t also be a settled fact that C “would freely” occur given A. Hence, there can be no true SCFCs.
In sum, then, Craig’s support for premise (2) is not only largely irrelevant and weak but, once we appreciate the contingency of the connection between antecedent and consequent in a CFC, it becomes readily clear that premise (2) is necessarily false. SCFCs by their very nature are incoherent since, again, it can’t both be an open question whether C would obtain given A and a settled fact that C would obtain given A.
Turning to premise (3), Craig thinks it must be true because pre-volitional and post-volitional are “logically exhaustive alternatives” (p. 263). This doesn’t seem quite right, however, for there is, in theory at least, a third possibility. Perhaps God’s knowledge of some SCFC is neither pre-volitional nor post-volitional but merely coincident with God’s volitional decree. On this suggestion, God comes to know an SCFC while issuing the decree and not because of the decree. If this is a coherent possibility, then Craig’s got to do a little more work to establish premise (3). That said, I won’t press the point further. I’ll concede the premise.
We come, finally, to premise (4). It is not the case, says Craig, that God’s knowledge of SCFCs is post-volitional. His argument for this is that, if (4) were false, then by premise (3) God’s knowledge of SCFCs would be post-volitional in which case “it is God who determined what every creature would do in every circumstance” (p. 263). But then, says Craig, “there really are no counterfactuals of creaturely freedom” (p. 264), which conflicts with premise (2). In short, Craig’s defense of (4) presupposes his prior defense of premise (2). Since I’ve already argued that premise (2) is false and that, at any rate, Craig’s defense of it is weak, I conclude that premise (4) has little to recommend it.
Overall, then, Craig’s philosophical argument for MK is a failure. Premise (2) is the lynchpin, and it has not only not been adequately defended, but there is very good reason to think the premise is not only false but necessarily so on grounds of incoherence.
(2.c) A theological argument for MK
By Craig’s estimation this is “[p]erhaps the strongest” argument for MK (p. 264). It’s actually a collection of arguments the overall gist of which is that MK “is astonishing in its subtlety and power” and “one of the most fruitful theological concepts ever conceived” (p. 264). Somewhat crudely put, the argument is basically “The thesis that God has MK must be true because it’s just so dang useful!” In this vein Craig refences several publications (mostly his own and Thomas Flint’s) in which MK is applied to various topics in theology, including not only divine foreknowledge and providence, but also biblical inspiration, soteriological particularism, perseverance of the saints, Christology, and papal infallibility.
Craig’s most developed argument of this sort focuses on divine providence. He references a couple biblical passages:
[T]his Jesus, delivered up according to the definite plan and foreknowledge of God, you crucified and killed by the hands of lawless men. (Acts 2:23, ESV)
[F]or truly in this city there were gathered together against your holy servant Jesus, whom you anointed, both Herod and Pontius Pilate, along with the Gentiles and the peoples of Israel, to do whatever your hand and your plan had predestined to take place. (Acts 4:27–28, ESV)
He then offers a Molinist reading of these passages which, to his mind, present a “staggering assertion of divine sovereignty over the affairs of men” (p. 265). As Craig puts it,
When one reflects that the existence of the various circumstances and persons involved was itself the result of myriads of prior free choices on the part of these and other agents, and these in turn of yet other prior contingencies, and so on, then we see that only an omniscient mind could providentially direct a world of free creatures toward his sovereignly established ends. (p. 266)
In response to Craig, there is a major hole in his reasoning, for the biblical passages he references say nothing about when “the definite plan and foreknowledge of God” concerning the crucifixion was established nor how detailed the plan was. Craig seems to just assume that this plan and all of the details as events actually transpired must have been part of God’s settled plan from the outset of creation. Molinism requires that assumption, but the biblical text is compatible with a much more proximate and much less detailed plan. In short, we don’t need Molinism to make sense of these passages. All we need is a God who can engineer the requisite details when necessary. Perhaps all that God had originally “predestined to take place” is that the Son would need to become incarnate. After the Fall the plan is amended to require that He eventually die. As history unfolds, the plan gradually becomes more precise until eventually it requires that Christ be born of Jewish decent and die by crucifixion in Jerusalem at the hand of the Romans by the instigation of the Jewish leaders. Maybe there’s a bit more to it than that, but my point is merely that the plan didn’t need to be fully settled in meticulous detail at the moment of creation. And so, these passages do not give us a strong argument for Molinism. The argument is in fact quite weak.
In sum, the theological argument for MK is easily overblown. While there are many facets of it that I have not looked at and that Craig himself merely footnotes, that MK helps us in various theological debates and offers “stunning” interpretations of certain biblical passages is not a strong argument in its favor. Other ideas, such as open futurism, are arguably just as fruitful. Moreover, the alleged theological benefits of Molinism are simply irrelevant if Molinism proves to be incoherent, something I have already argued is the case. (And which I have argued in detail elsewhere.)
(3) Answering objections to Molinism
Craig’s final subsection is a response to two common philosophical objections against Molinism. Both are variations on the so-called grounding objection. The first, which occupies about 75% of this subsection, contends that there are no true SCFCs. The second, which occupies the remaining 25%, contends that SCFCs can’t be prevolitionally true. Obviously, if the first objection succeeds then the second one is moot. But potentially the second could succeed even if the first one fails. So, they are different objections, despite their overlap.
Craig does make clear that these are not the only philosophical objections that have been leveled against Molinism, and he says that he will deal with additional objections in a later volume of his systematic (p. 266).
Side note: I sketch seven objections to Molinism here. The first five are philosophical objections. The sixth is biblical. The seventh is theological.
(3.a) Are there true SCFCs?
This first objection is what we might call the metaphysical grounding objection (MGO). It contends that there are no MK truths because there is nothing in reality that suffices to underwrite or ground those truths. Such truths cannot be grounded in God’s metaphysically necessary nature because they are logically contingent. They cannot be grounded in God’s will because they are pre-volitional. And they cannot be grounded in creation because they are explanatorily prior to God’s creative decree. What, then, could possibly ground MK truths? By the nature of such truths it would have to be something intrinsic to God that is both metaphysically contingent and pre-volitional. But what in a monotheistic framework could possibly fill that role? To my knowledge Molinists have never given even a remotely plausible answer to this question. Indeed, I don’t think they can answer it.
Rather than meeting the objection head-on, Craig’s strategy is dismissive. He first contends that the MGO “has scarcely ever been articulated or defended in any depth by its advocates” (p. 267), as though the objection is inherently confusing or counterintuitive, leaving the objector with the burden of first stating it clearly. I don’t think this is a serious reply on Craig’s part. As I’ve just described it above, the MGO is quite intuitive. Fortunately, Craig doesn’t stop with bare dismissal but aims to provide an undercutting defeater by challenging the intuition that truths need to be grounded in anything more than whatever abstract “facts” or “states of affairs” are disclosed by the disquotation principle. Let me explain what this means.
The disquotation principle (i.e., “p” is true iff p) takes the linguistic predicate “is true” to be a device of metalinguistic ascent. That is, we can use it replace a first-order or object language statement p with a second-order or metalanguage statement about the object language statement, namely, that “p” is true. For example, instead of saying “The cat is on the mat” (first-order expression) we can say “‘The cat is on the mat’ is true” (second-order expression). The disquotation principle simply says that we can use the truth predicate to switch back-and-forth between the first- and second-order expressions as needed. Now, when Craig talks about the “facts” disclosed by this principle, all he’s saying is that if we need to find a “ground” for a second-order expression like “‘p‘ is true” then we can just dis-quote it by removing the quotations and the “is true” predicate to bring us back to the first-order expression p. The “fact” disclosed by dis-quoting “‘p‘ is true” is simply p. In sum, Craig thinks that insofar as truths need grounds at all, finding grounds is metaphysically trivial. Basically, we just restate the proposition we want to find grounds for, call that restatement a “fact” or “state of affairs,” and say “there’s your grounds.” Thus, what grounds or makes <the cat is on the mat> true is simply the “fact” that the cat is on the mat. If that sounds trivial to the point of absurdity, it’s because it is. As I said in the previous installment of this series, while Craig pays lip service to the correspondence theory of truth, he actually holds to a deflationary theory of truth. This is the view that the disquotation principle tells us all we need to know about truth. Truth is merely a device of metalinguistic ascent and therefore has no metaphysical implications whatsoever. That’s why “grounding” is trivial for Craig.
Most critics of Molinism are anti-deflationists about truth, however. They aren’t satisfied with vacuous “grounding” of the sort that Craig suggests. They want metaphysically robust grounding for MK truths. But that’s something Molinism seems unable to provide. Hence, the MGO. Craig’s main strategy against the MGO, therefore, is to challenge this robust grounding intuition. He does this by listing several statements that, he thinks, cannot plausibly be grounded in a metaphysically robust way. Here they are:
(6) No physical objects exist.
(7) Dinosaurs are extinct today.
(8) All ravens are black.
(9) Torturing a child is wrong.
(10) Napoleon lost the Battle of Waterloo.
(11) The U.S. President in 2070 will be a woman.
(12) If a rigid rod were placed in uniform motion through the aether, it would suffer a FitzGerald–Lorentz contraction. (pp. 269–270)
I’ll work through Craig’s treatment of (6)–(12) shortly. First, I need to introduce the concept of a truthmaker. This is something the existence of which “makes” a true proposition true. In other words, a truthmaker explains or grounds the truth of a proposition. I bring this concept up because Craig frames his discussion of (6)–(12) in terms of truthmaking. But he misunderstands the truthmaking principle. As he puts it, a truthmaker is “[a]n entity a [that] makes a proposition p true if and only if that a exists entails that p” (p. 268, my underlines). There are two problems with this statement. First, “that a exists” represents a proposition. But truthmaking is not (in general) a relation between a proposition p and another proposition <a exists>. It is not, as Craig states in a footnote, “essentially a logical relation” (p. 268). No. Truthmaking is an explanatory (not logical) relation between concrete reality (what exists) and a proposition. Second, truthmaking is not best understood as an entailment relation. This is because entailment is typically understood as a relation between propositions, whereas truthmakers (in general) are not propositions. Take, for example, the proposition <Tat the cat is on the mat>. If we suppose that this proposition is true, then—according to the truthmaker idea—there exists something that suffices to make that proposition true. Most plausibly that “something” is not a proposition or any abstract “fact” but consists of an actual cat named Tat, a physical mat, and the concrete state of affairs that is Tat’s being on said mat. So, to correct Craig’s statement of what a truthmaker is we should instead say something like this:
truthmaker (TM) =def. An entity a the existence of which provides a sufficient metaphysical ground or explanation for the truth of some proposition.
And if we suppose additionally that all truths have truthmakers—this is in fact a corollary of the correspondence theory—then we arrive at a principle called truthmaker maximalism:
truthmaker maximalism (TMmax) = def. For every true proposition p, there exists a truthmaker, i.e., an entity a the existence of which provides a sufficient metaphysical ground or explanation for the truth of p.
Craig’s main goal in listing (6)–(12) is to argue that either (a) TMmax is false or (b) it’s true but trivial, since we can easily obtain truthmakers by disquotation, as explained above.
Let’s now consider statements (6)–(12). (6), says Craig, “could be true” and (7) is true “yet they preclude truthmakers which are or imply the existence of the relevant concrete objects (such as dinosaurs)” (p. 270). But why should we accept this? It seems to me that if (6) were true, then it would be true in virtue of the concrete state that is God’s having a physical object-free experience of reality. Likewise, (7) is true in virtue of the concrete state that is God’s present memory of dinosaurs. Craig seems to think this doesn’t work because memories aren’t the “relevant” sort of concrete object. I disagree. God has the memories He does because He was there when dinosaurs existed and observed first-hand when they became extinct. So long as God cannot have faulty memories, why isn’t that relevant?
As for (8), Craig says that this “is a universally quantified statement which as such does not carry existential implications” and so cannot be made true by “existing ravens’ being black” (p. 271). (8), however, is ambiguous. There are two different propositions it could be expressing. One is <All actual ravens are black>. The other is <Ravens are essentially black>. The first is straightforwardly existential and would be made true by all existing ravens concretely being black. The second is not existential because it could still be true even if all ravens went extinct or never existed in the first place. But even if ravens don’t exist, raven nature still does because it’s an eternal idea in God’s mind. The existence of that concrete idea makes the proposition true. Something similar can be said for (9). Even if no children ever exist, still the idea of children exists in God’s mind. And it is plausibly in virtue of that idea and the consequent value that God would place on children were they to exist that (9) is true.
Moving on, Craig says that (10) and (11) “are tensed statements about persons who no longer or do not yet exist … and so cannot have such persons among their truthmakers” (p. 271). But the right thing to say here, at least for an open futurist, is that (10) is probably false because it is plausibly a future contingent and so, while it is possible that a woman is President in 2070, it is not the case (at least not yet) that a woman will be President in 2070. But suppose this isn’t a future contingent. They we can say that the truthmaker for (10) is the (locally) deterministic causal structure of concrete reality, which structure ensures that a woman is President in 2070. Craig’s position on (11) is more puzzling because, as a professed Christian who presumably believes in the general resurrection of the dead, he should not affirm that Napolean no longer exists. He could say that Napoleon no longer exists in the land of the living, but not that he doesn’t exist period. In any case, that Napoleon lost the battle of Waterloo has a concrete truthmaker in the same way that (7) does, namely, God concretely remembers Napoleon’s loss.
Finally, (12) is a true counterfactual about an apparently non-existent entity (the luminiferous aether) that was hypothesized by 19th-century physicists but eventually rejected because of failed experimental tests based on propositions like (12). Craig thinks it’s hard to find a plausible truthmaker for (12). I disagree. Just as with (8) and (9), even if the aether doesn’t exist, the idea of the aether concretely does in God’s mind.
In sum, Craig’s examples don’t pose any problems whatsoever for TMmax—at least not if we assume Christian theism, which Craig and I both affirm. That he doesn’t see how there could be truthmakers for (6)–(12) is more a reflection of his poor understanding of truthmaking and a lack of metaphysical imagination than a problem with TMmax. (For my full discussion and defense of TMmax, see this post.)
But Craig obviously doesn’t see matters that way. He thinks he’s successfully shown that there are statements “which either are true without having truthmakers or else have as their truthmakers abstractions like facts or states of affairs,” with the latter being “disclosed” by disquotation. He concludes that the idea that truthmakers are concrete entities is “untenable” and “naïve.” Applied to Molinism, the idea that the truthmakers of true SCFCs must be “any sort of concrete object” is, he says, “extraordinary” (p. 272).
At this point, since Craig thinks he’s adequately made his case against any metaphysically robust interpretation TMmax, he turns his attention to rebutting some of the more forceful proponents of the MGO, particularly Tim O’Connor and William Hasker.
(3.a.i) Craig contra O’Connor
In a 1992 article “The Impossibility of Middle Knowledge,” O’Connor forcefully presses the MGO against Molinists Tom Flint and Alfred Freddoso. I think O’Connor does a bang up job. Craig disagrees. He begins his rebuttal to O’Connor by endorsing Freddoso’s suggestion that SCFCs (settled conditional future contingents) be understood in a manner analogous to SFCPs (settled future contingent propositions):
[T]here are now adequate metaphysical grounds for the truth of a future-tense proposition Fp [i.e., It will be the case that p] just in case there will be at some future time adequate metaphysical grounds for the truth of its present-tense counterpart p … But if this is so, then it seems reasonable to claim that there are now adequate metaphysical grounds for the truth of a conditional future contingent F t(p) on H [i.e., If H were the case then it would be the case that p obtains at time t] just in case there would be adequate metaphysical grounds at t for the truth of the present-tense proposition p on the condition that H should obtain at t. (p. 273, Craig quoting Freddoso)
In reply to Freddoso, O’Connor argues that Freddoso’s suggestion is “just wrong” because if the actual grounding only obtains in the future or in some hypothetical scenario then there is not now anything “there” in reality which is its grounds. He argues that Freddoso is only proposing truth-conditions, not providing truthmakers, and that it would be more accurate to say not that Fp has grounds now but that p will have grounds at a future date. Similarly, F t(p) on H doesn’t have grounds now but perhaps F t(p) would have grounds were H to obtain.
Side note: The truth-condition / truthmaker distinction is important. The truth-conditions of a proposition are its logical entailments. They are always other propositions. In contrast, truthmakers (pace Craig) are generally not propositions but concrete realities of one sort or another. One truth-condition for <Tat the cat is on the mat> is the proposition <Tat the cat exists>. The truthmaker for that proposition, however, is a concrete state of affairs that includes, among perhaps other things, a physical mat and an actual living cat complete with bones, blood, and fur.
In reply to O’Connor, Craig objects: “What is wrong with the facts or states of affairs proposed by Freddoso as the truthmakers of such propositions?” (p. 275) And he complains that O’Connor is presupposing the “naïve” view that truthmakers are (in general) concrete (p. 275). I’ve already explained above, however, why Craig’s charge of naïveté is bogus. It’s not hard at all—at least not for theists—to describe concrete truthmakers for all sorts of propositions that Craig thinks are hard cases for TMmax.
In a long footnote on p. 274 Craig levels more silly criticisms of O’Connor and in the process endorses some truly absurd ideas about the metaphysics of truth. In contrast to O’Connor who holds (quite correctly, I wish to add) that “God’s infallible knowledge of a genuinely contingent proposition p involves or just consists of an immediate acquaintance with the grounds for p” and that “One cannot … discern the truth of a contingent proposition by having a specially penetrating insight into the nature of the proposition itself” (p. 274), Craig replies (in essence) “Nuh-uh! If that were right, then open futurism would be true and God would be ignorant!” Setting aside the straw man about divine ignorance, which I have already addressed, Craig insists again (cf. p. 218) that truth is an intrinsic property of propositions, even logically contingent propositions! And so (pace O’Connor), “God most certainly can by an immediate inspection of the proposition itself discern whether it bears the property truth or not” (p. 274). I’ve already commented on how nutty this intrinsicality claim is—see §1.b.1 and §2.e of my previous post for discussion. But Craig doesn’t stop there. He argues that any present-tense proposition attributing truth to an SFCPs or SCFCs (or any truth-ascribing proposition for that matter) has an “evident truthmaker,” namely, “the inherence of the property of truth” in the very propositions that it says are true (p. 274). If we follow this to its logical conclusion we get a truly bizarre result. Since every proposition is implicitly truth-ascribing (just apply the disquotation principle in reverse), it follows that every true proposition is its own truthmaker for as long as it is true! If truth is intrinsic, this is exactly what we should expect. Nevertheless, it’s incredibly nutty. Indeed, it’s a prescription for alethic chaos because it means that all contingent truth is brute. Some logically contingent propositions just make themselves true (for a time) and then, perhaps, they don’t. Now, I doubt Craig actually endorses this idea, but it’s implicit in what he says, and it goes to show that he hasn’t thought through the metaphysics of truth nearly well enough.
Moving on, O’Connor has a second objection against Freddoso that Craig takes issue with. This is that it “spawns a vicious infinite regress of truth” (p. 275). O’Connor argues that if Freddoso is right that the SFCP proposition Fp has grounds just in case “there will be at some future time adequate metaphysical grounds for the truth of its present-tense counterpart p” then because that too is an SFCP proposition, it also needs grounds and therefore presupposes another SFCP proposition needing grounds, and so on. Craig replies that “O’Connor has conflated the truthmaker of Fp with the truth-conditions of the statement that Fp has a truthmaker.” There is no regress, says Craig, because the truthmaker of Fp is “the fact that there will be a truthmaker of p” and “[f]acts do not themselves have truthmakers” (p. 275).
Craig’s reply to O’Connor is mistaken. First, recall from a few paragraphs back O’Connor’s charge that Freddoso is offering truth-conditions not truthmakers. This shows that O’Connor is alert to the truthmaker/truth-condition distinction. His regress objection, moreover, is stated entirely in terms of truth-conditions (see O’Connor, p. 156). As for Craig’s claim that facts do not have truthmakers, this depends on what a “fact” is, an issue Craig never addresses. On one common philosophical usage of the word, a “fact” just is a true proposition. That usage is consistent with how Craig uses the term. Indeed, every time he uses the phrase “the fact that” one can coherently replace it with “the true proposition that” without noticeably changing his meaning. Obviously if Craig’s facts are just true propositions then it is quite reasonable to think that facts have truthmakers, at which point the prospect of a vicious regress reemerges. And if Craig’s facts are not true propositions then it’s incumbent upon him to tell us what they are. Label aside, what’s the actual difference between a proposition and a fact?
Recall that the disquotation principle Craig uses to “disclose” facts is a biconditional metalinguistic principle. It tells us how to convert a first-order object language statement into a second-order metalanguage statement and vice-versa. Both the second-order statements and the first-order facts disclosed by the principle are linguistic entities of the same basic kind, namely, statements. Perhaps Craig’s idea is that once we arrive at a first-order statement we don’t have a way to disquote it further to disclose a zero-order statement. I conjecture that this is what, in his mind, prevents the regress. But if those first-order “facts” are statements then they express propositions of which we can then ask, are they true or false? Once we pose that question we metalinguistically ascend back to the second-order level, at which point disquotation again becomes possible. Thus, even if we can somehow avoid a vicious regress, we wind up in a vicious circle. To spell out the problem, “It is true that p” leads by disquotation to the fact that p. We can take that fact and metalinguistically ascend by asking whether p is a “true fact”? Is it really a fact or not? If not, then p is useless as a truthmaker. If we admit p as a true fact, however, then we’re right back to “It is true that p.” In short, there’s no genuine explanation for truth to be had here. We’re just going round-and-round a semantic wheel!
Craig closes out his discussion of Freddoso and O’Connor by stating his overall position on truthmaking as applied to Molinist counterfactuals (i.e., SCFCs). He says that “if true counterfactuals of creaturely freedom have truthmakers, then the most obvious and plausible candidates are the facts or states of affairs disclosed by the disquotation principle” (p. 277). This is clearly his view of truthmaking, not just for SCFCs, but propositions generally. He says that these “disclosed” facts are “unobjectionable truthmakers” (p. 277). I’m sorry but, no. I object. Here’s why: When we use disquotation to reduce a truth-ascribing proposition <p is true> to a corresponding fact p, we aren’t giving any useful information about what that fact consists in. All “p” gives us is a label for whatever it is in reality that makes the proposition true. This is all Craig’s “fact” talk does. He explains the truth of p by pointing to the fact that p. To allude to Molière’s famous satirical example, identifying truthmakers in this way is analogous to answering “How does opium produce drowsiness?” with “Because opium has ‘dormitive power,'” that is, the power to produce drowsiness. The problem here is that the explanation offered, while superficially obvious, is utterly vacuous and uninformative because it does nothing more than restate what we wanted to explain. If this is all Molinists can say about truthmakers for SCFCs—and it’s certainly all that Craig is prepared to say—then I concur with O’Connor that the more honest position would be to admit that SCFCs don’t have truthmakers. Better to just admit SCFCs as brute and thumb one’s nose at TMmax than to pretend that utterly uninformative “facts” can play the truthmaking role.
(3.a.ii) Craig contra Hasker
Having supposedly dispatched O’Connor’s version of the MGO, Craig shifts his focus to William Hasker’s contention that “truths about ‘what would be the case … if‘ must be grounded in truths about what is in fact the case” (p. 277). Hasker unfortunately misstates the problem by suggesting that conditional truths need to be grounded in categorical truths. Craig calls Hasker out on the misstatement (p. 278). What Hasker should have said is that “truths about ‘what would be the case … if‘ must be grounded in what is in fact the case.” (I deleted the second instance of “truths about.”)
At this point Craig proceeds to talk about dispositional properties like the fragility of glass. He points out, quite correctly, that we express dispositional properties using counterfactual conditionals. For example, the glass’s being fragile means (among other things) that if it were dropped onto a hard surface it would (probably) shatter. And he notes, again quite correctly, that dispositional properties typically have a “causal basis” in categorical properties of the things involved, such as the molecular structure of glass (p. 278). Craig then argues that “it is a non sequitur to conclude that the [categorical] causal basis of a disposition is the truthmaker of the counterfactual at issue” because “if there were different laws of nature” then that same causal basis might yield a very different result (p. 278). And, moreover, the laws of nature, says Craig, “are equivalent to various counterfactual propositions” (p. 278). The upshot is that there is no way to reduce dispositional properties to purely categorical properties, and so, concludes Craig, Hasker is wrong to think that the counterfactual must ultimately be grounded in the purely categorical.
What should we make of this? Well, for starters Craig is just wrong to say that laws of nature are “equivalent to” propositions of any sort, much less counterfactual propositions. To be sure, we express laws of nature using propositions, but those expressions are not the laws themselves. They are merely expressions of the laws. As to what natural laws are most fundamentally, I propose that they are God’s categorical volitions concerning how the created world should behave. For simplicity’s sake, let’s set relativity aside and suppose that Newton’s F=ma is a basic law of nature. Why does the world obey and continue to obey that law? Theists who believe that God created and sustains the world should say that, ultimately, the world behaves in accordance with F=ma because God categorically wants it to, at least for now. I conclude, therefore, that Craig has not successfully rebutted Hasker. Very plausibly, especially for theists, what would be the case is ultimately grounded in what is the case.
Craig has one more arrow to fire at Hasker. He says that if Hasker is right about the conditional needing to be grounded in the categorical then “the grounding objection seems implicitly to reject libertarian freedom” (p. 279). Why? Because “[i]t is simply a fact” that a free agent “would freely choose to act” in a certain way under specified circumstances (p. 279). Obviously, if the grounding requirement were accepted then this couldn’t be “simply a fact” but would have to have a categorical causal basis.
The problem with Craig’s argument here is that it is blatantly question-begging. Craig just assumes that there are brute facts to the effect that creatures “would freely” choose in specific ways under specified circumstances. While this is a core assumption of Molinism, when your theory is under direct challenge you can’t just reassert your theory. That’s begging the question. And Hasker is not implicitly rejecting libertarian freedom. He’s merely challenging Molinism. Craig simply assumes that one can only have libertarian freedom if there are brute “would freely” facts of the sort Molinism posits. That’s far from obvious. Open theists and other non-Molinist free-will theists would beg to differ. So, again, Craig has not successfully rebutted Hasker. A better strategy would have been to play pure defense and challenge Hasker to prove that the conditional must ground out in the categorical.
(3.b) Are SCFCs true prior to God’s decree?
In the final subsection of his chapter section on Molinism, Craig replies to a pair of objections from Robert Adams.
Adams’ first objection is that Molinist CFCs cannot have any determinate truth value in the moment explanatorily prior to God’s decree. The reason for this turns on the standard (Lewisian) semantics for counterfactuals. According to this semantics, the truth conditions for an arbitrary SCFC counterfactual A ☐→ C (i.e., if antecedent A were the case then consequent C would be the case) entail that the counterfactual is true if and only if all of the A-worlds relevantly similar to the actual world are also C-worlds. The obvious problem for Molinism is that which world is actual is only settled once God issues His divine decree, and so (argues Adams) there can be no true Molinist CFCs until God’s issues that decree.
In reply to this objection Craig argues that middle knowledge CFCs don’t require a complete actual world to have determinate truth values. This is because the actual world is a function of God’s natural knowledge, middle knowledge, and creative decree, in that order. Middle knowledge CFCs depend for their truth on God’s natural knowledge but otherwise just happen to be true. That is, they happen to correspond to abstract “facts” disclosed via disquotation. These facts are actual, albeit brute. They just obtain, with no explanation for why they obtain as opposed to other facts that would yield different middle knowledge for God. In short, then, according to Craig, middle knowledge CFCs don’t depend for their truth on which world is actual because they constrain and partly constitute the actual world.
As I see matters, neither Adams’ first objection nor Craig’s reply are any good. Both fail (in part at least) because they mistake the underlying, commonsense rationale for the standard counterfactual semantics with a distinctively philosophical (and non-commonsensical) perspective on possible worlds. The reason why the standard semantics links counterfactual truth conditions to “the actual world” is because (a) most counterfactuals come up in contexts that presuppose a large amount of background information not stated in the counterfactual, (b) this background information is normally anchored in the real world, and (c) because contexts are so variable, it would be impossible to articulate a general semantics for counterfactuals unless we had a generic way of referencing that background information. This is what reference to “the actual world” does in the standard semantics.
For example, the counterfactual <If the Nazi’s had won WWII, then the U.S. would now have German as an official language> presupposes much of the actual course of world history up until at least the beginning of WWII. It also presumably presupposes the actual laws of nature and normal human psychology and psychology. The counterfactual itself doesn’t explicitly state any of that information, of course, but anyone hypothesizing about Nazis, WWII, the U.S., and the German language would be naturally understood as hypothesizing about things in actual history. If they meant something completely different by those terms than how we understand them in the real world, then we’d have a hard time making any sense of the counterfactual, much less assessing it for truth or falsity. Not everything in actual history is relevant, however. It probably doesn’t matter one bit whether some caveman scratched a stick figure on a cave wall 10,000 years ago or not. And it almost certainly doesn’t matter whether some random star went nova in the Andromeda galaxy. And so on. Not all details matter, and not all that do matter do so to the same degree. Hence, to assess the counterfactual we want to consider scenarios that are relevantly similar to, or relevantly close to, the actual world, allowing non-relevant details to vary and changing relevant details only as needed to generate smooth deviations from actual history that satisfy the antecedent. Notice, though, that by “the actual world” in this semantic approach we do not mean what philosophers like Adams and Craig apparently take it to mean, namely, a complete timeline fully specifying all of creation from beginning to end. That philosophical conception of “the actual world” is way more detailed than any commonsense semantics could possibly require.
This is ultimately why Adams’ first objection fails. Normal counterfactuals don’t need to be assessed in relation to that logically maximal state of affairs that God’s creative decree (on Molinism) designates as “the actual world.” Normal counterfactuals just need to be grounded somewhat in our concrete actual world so that we have a context for assessing their truth values. Counterfactuals, being counterfactual, necessarily hypothesize about scenarios that differ somewhat from reality as we know it. To assess a counterfactual, therefore, we just need to know how that scenario is relevantly similar to and different from reality as we know it.
Craig’s reply to Adams also fails, however. This is because Molinist counterfactuals are not normal counterfactuals. They have fully causally specified antecedents and so explicitly contain their full relevant context within themselves. Accordingly, they don’t need to reference “the actual world” to supply an implicit background context because all the necessary information is already there, in the antecedent.
Consider, then, an arbitrary fully causally specified indeterministic (FCSI) scenario, A. Let A ☐→ C stand for <If A were to obtain, then it would be the case that C obtains> and let A ☐→ ~C stand for <If A were to obtain, then it would be the case that C does not obtain>. To evaluate whether A ☐→ C is true or A ☐→ ~C is true (or whether neither is true) we focus on the A–C relation. We posit the actuality of A and ask if either C or ~C is what we would expect to find in all, none, or only some A-scenarios, that is, all scenarios relevantly similar to A in virtue of sharing the same full causal specification. Now if, on the one hand, it would be that C obtains in all (or no) A-scenarios, then even though A is an FCSI scenario, C is not causally contingent in relation to A. So, we get a determinately true (or false) counterfactual, but not an SCFC. If, on the other hand, C obtains in only some A-scenarios and ~C obtains in the rest, then C is causally contingent in relation to A, but neither A ☐→ C nor A ☐→ ~C comes out as true. Since all Molinist SCFCs correspond to this latter case, it follows that no Molinist counterfactuals are true. They all incoherently posit a determinately settled outcome given an FCSI scenario that, by definition, is open-ended toward multiple causally possible outcomes.
To his credit, Craig is aware of this problem. In a footnote he says,
It might be said that since libertarian freedom requires the ability of creatures to choose differently in identical circumstances, equally similar worlds could not support true counterfactuals of creaturely freedom [i.e., Molinist SCFCs]. For in equally similar worlds free creatures will choose differently, so that the consequents of such counterfactuals would not be uniformly true in all antecedent-permitting worlds. (p. 281)
In response to this worry, Craig endorses the suggestion by Alvin Plantinga that “shared counterfactuals are themselves a measure of similarity between worlds” (p. 281). In other words, Craig proposes that we pack MK information into our hypothetically actual FCSI scenarios on the grounds that it’s relevant to the assessment of MK counterfactuals. But this is both illegitimate and question-begging. It’s illegitimate because the truth or falsity of Molinist SCFCs cannot be added to the antecedent represented by A without immediately nullifying the contingency of C in relation to A. The resultant conditionals would no longer be CFCs and so would be useless for Molinism. Craig’s suggesting is also question-begging because we can’t simply presuppose that some MK information is true in response to an objection that contests that very point.
Turning to Adams’ second objection, this charges that Molinism is stuck in an explanatory loop. If creatures are sometimes free, then it is up to them whether some of the middle knowledge counterfactuals that describe them are true. That is, their actual free choices are explanatorily prior to the truth of those counterfactuals. But, as prevolitional, the truth of those counterfactuals is also explanatorily prior to God’s creative decree and thus explanatorily prior to the existence of the creatures in question and thus explanatorily prior to any actual free choices those creatures make. I think this is a devastating objection.
Craig’s reply follows that of Thomas Flint, who argues that the relation of explanatory priority employed throughout the objection is either equivocal or, so far as we can tell, non-transitive. Accordingly, the explanatory loop cannot be closed (p. 282).
I believe Craig’s and Flint’s replies miss the fundamental point. The objector doesn’t need to close an explanatory loop. He can simply argue (consequence argument-style) that given what’s fixed at each explanatory stage, there can be no future contingency on Molinism. I lay out just this sort of argument here. Very briefly, given prevolitionality, God’s middle knowledge is fixed in relation to God’s decree. God’s decree can’t overwrite it. Every feasible world has to be compatible with God’s middle knowledge. But then that information is also fixed in relation to actual creaturely choices. You and I can’t overwrite God’s middle knowledge concerning us any more than we can overwrite God’s creative decree. If we could, then middle knowledge would become providentially useless. Because we can’t overwrite it, God’s middle knowledge constrains our freedom as much as any other fixed fact, i.e,. any fact that we have no say about. Thus, we are not free to do otherwise than what God’s middle knowledge says we “would” do. (Craig’s appeal to “counterfactual dependence” does not undermine this result. See §2.d of my previous post for elaboration on this point.)
Concluding thoughts
Overall, I find Craig’s discussion of Molinism unimpressive. His explanation of Molinism lacks important precision, such as the distinction between SCFCs and OCFCs. His positive case for Molinism is weak because he provides no relevant proof of premise (2), namely, that there are true SCFCs. And, finally, his responses to the general MGO and the more specific objections of O’Connor, Hasker, and Adams are all quite weak. In particular, Craig’s metaphysics of truth is explanatorily vacuous and as nutty as it gets. Truth is not an intrinsic property of any logically contingent proposition. The disquotation principle does not “disclose” any truthmaking “facts.” At most it gives us a formulaic way to name a proposition’s truthmaker without giving us any insight into what that truthmaker consists of. Moreover, these so-called “facts” are not discernably different from true propositions and so seem to stand in need of truthmakers themselves.