Some old maps put monsters in the blank spaces. The sea serpents did not mark the edge of the world. They marked the edge of the survey.
Something similar happens in classical theism. Across nine centuries, three religious traditions, and several philosophical languages, thinkers who agree about almost nothing else arrive at the same joint in their systems and place something there.
I call the device a modal firewall: a selective restriction, reclassification, or termination of explanatory inquiry, placed at the point where continuing the explanation threatens to turn a contingent world into a necessary one.
The term is diagnostic, not accusatory, and is meant as a working constraint rather than a courtesy. A firewall may be independently justified. It may mark a real distinction rather than an improvised defense. But its location matters. When a boundary appears precisely where an explanatory system approaches an unwanted modal consequence, the boundary acquires a burden. It must be defended by more than the fact that it preserves the result the theory wants. Some of the cases below discharge that burden. One of them discharges it almost entirely.
One unexpected conclusion is worth stating at the outset. In the standard presentation, contingency is the premise and the necessary being is the conclusion: things need not have been, so they point beyond themselves to what must be. The record of these devices suggests something closer to the reverse. Contingency is not the comfortable starting point of classical theism. It is the thing classical theism has to work hardest to keep, and the architecture at the joint is the record of that work.
The predicament
The trouble is structural. Classical theism holds three things at once. First, the source of everything is necessary: not merely everlasting, but unable not to exist. Classical theists combine this with divine simplicity, according to which God is not composed of parts and possesses no intrinsic features that are distinct from the divine essence. Second, what the source produces is contingent: creation is a gift, not an emanation. This world did not have to exist. Third, the ascent from creatures to their source is a real explanation, not a dignified shrug. Natural theology depends on refusing to say that reality simply stops making sense at the summit.
Any two of these commitments sit together comfortably. A necessary source can produce a necessary order. A contingent world can remain brute. An explanation can proceed from one contingent fact to another. The strain appears when all three are combined. If a contingent world depends explanatorily on a necessary being, what stops necessity from running back down the chain the explanation has just climbed?
That question does not by itself establish a contradiction. Much depends on what kind of dependence is involved, whether the explanans necessitates the explanandum, and whether divine willing can be necessary as an act without necessitating its object. But those are not side issues. They are the point at which the theory must explain how the modal difference survives the explanatory connection.
Something has to happen at the joint. What’s fascinating is not that theists place something there. It is that they place very different things there, and that the location itself migrates across the tradition.
The test
Before the cases, the criteria.
Not every explanatory stopping point is suspect. Mathematics stops where it stops for reasons that have nothing to do with protecting a desired modal result. A theory may show that a relation is asymmetric, that a contrast has been misdescribed, that an explanatory demand is illegitimate, or that a terminus follows from the nature of the subject matter. Any diagnostic worth having has to be able to acquit such cases, and has to be able to say in advance what acquittal would look like.
Five questions do that work.
- Priority. Was the stopping point motivated before the modal danger appeared, or does it enter the system at the moment of pressure?
- Independent yield. Does it solve any problem other than preserving contingency?
- A real alternative. Does the theory identify a genuinely open opposite, or does it redescribe a necessary outcome in more accommodating vocabulary?
- Transmission. Does the explanatory relation carry necessity from ground to consequence, and if not, what specifically blocks the carriage?
- Counterfactual acceptance. Would the boundary still be accepted if it produced the opposite modal result?
A boundary is earned when these have answers. It is suspect when its only defense is the very contingency it was introduced to save. After the survey, the questions are applied to two cases (one that passes and one that does not).
Eight ways to hold the line
Avicenna refuses to protect alternativity. The First is necessary in itself, and everything else proceeds from it by necessity of nature. He does not simply erase the distinction between creator and creature. He redraws it. Every created thing is possible in itself and necessary through another. Considered in itself, its essence does not include existence. Considered in relation to its cause, its existence cannot fail.
This is an elegant reclassification. It preserves asymmetric dependence, aseity, and the difference between having existence in oneself and receiving it from another. What it does not preserve is an actual alternative. Nothing in the complete causal order could have been otherwise. Contingency has been moved from the order of events into the internal constitution of dependent natures.
Al-Ghazālī’s critique of the philosophers is aimed, among other things, at exactly this point. A possible-in-itself world that necessarily follows from its source is not the free creation the theologian wanted. The alternative must be real, not merely abstracted from the cause that excludes it. This is the one place in the sequence where the historical through-line is well documented: a specific placement was found inadequate, in print, by a reader who named it.
Maimonides answers in two layers. His first move is particularization. Necessity of nature tends toward uniformity, yet the heavens exhibit this number of spheres, these motions, these directions. The arrangement appears particular in a way that impersonal emanation cannot explain. A purposive agent is therefore required. What a rationalist may count as an unpaid explanatory debt, Maimonides converts into evidence for will. The residue is not an embarrassment. It is the argument.
His second move limits what may be inferred once will has been introduced. Maimonidean negative theology sharply restricts affirmative claims about the divine essence, even while allowing us to speak of divine actions. The result is not absolute silence, but it narrows the terms on which the divine decision can be inspected. The boundary is drawn, and then the conditions for examining it are tightened.
Maimonides is unusually candid about the stakes. Creation is preferable to eternity because eternity has not been demonstrated and because creation better accommodates will, prophecy, and law. Were eternity demonstrated, scriptural interpretation would have to change radically. He posts a warning sign beside his own boundary. The cost is that the stopping point becomes difficult to audit: the less that may be said about what stands behind it, the harder it is to ask whether the limit is independently earned or simply protective.
Aquinas makes a cleaner cut. God’s goodness is complete without any created order. No creature is therefore necessary as a means to divine perfection. God wills creatures, but does not need them. Explanation runs at full strength on both sides of the cut. Divine willing is a positive explanatory principle in the system, not a refusal to explain.
The elegance is also the exposure. Divine simplicity says that God’s act of willing is not a separable state added to the divine essence. If the divine essence is necessary, how can the willing of this world fail to be necessary? Aquinas’s account of relations helps hold the line. The creature is really related to God because it depends on God. The corresponding relation to the creature is not a real addition in God, but a relation of reason. The relation does not modify the necessary source.
That is more precise than saying that God is not related to creation at all. Aquinas is not cutting every connection. He is denying that creation introduces a new intrinsic feature into God. Whether that asymmetry explains contingency or mainly preserves divine immutability is the question the survey has to answer.
Scotus is the hinge, and the most direct answer anyone in this tradition gives to the question the reclassification strategy leaves open: is the opposite genuinely available?
The received picture he inherited treats possibility as spread across time. Something is possible if it obtains at some time or other; what is actual now is, in the relevant sense, necessary now, and the only openness available is the openness of a later moment. On that model, a will is free because it may will otherwise afterward. It is not free at the moment it wills. Scotus rejects the model. In the Lectura I 39 he argues that the will retains a real power for the opposite at the very instant at which it elicits its act. Not a power to change later. A power for not-A at the same instant at which A is willed. Later readers call this synchronic contingency, and the name marks a genuine break: possibility stops being a function of the temporal series and becomes a matter of the logical compatibility of alternatives, available to be assessed at an instant.
The structural consequence is the one that matters here. Scotus argues that contingency in the world cannot be produced by a first cause that causes necessarily; if the first cause acts of necessity, nothing downstream is contingent, however many intermediate causes are inserted. Nor can the divine intellect be the source, since the intellect apprehends what is intelligible rather than settling what obtains. The source of contingency must therefore be the divine will, and the divine will must be contingent in the synchronic sense: at the instant it wills this order, it retains the power for another.
This is not a restriction on inquiry. It is a positive theory of what possibility is, and its rationale runs well beyond the modal question at the joint: it reorganizes the analysis of freedom, of future contingents, and of divine knowledge. That is exactly the profile of a boundary with independent motivation, and Scotus deserves the credit.
The cost shows up one level in. A power for the opposite that is never exercised has to be something, and Scotus’s answer is that it is grounded in the logical compatibility (or, in Scotus’s own language, the non-repugnance of the terms), which he treats as prior to and independent of any act of will. That relocates the question rather than closing it: what accounts for the modal facts about non-repugnance, and what is their relation to a simple God? Scotus’s own answer requires the formal distinction, which lets him say that the divine attributes are distinct in a way weaker than a real distinction but stronger than a merely conceptual one. Thomists have argued ever since that this is simplicity purchased at a discount. Whether or not they are right, the observation for present purposes is structural: to make the divine will contingent, Scotus has to introduce distinctions inside God. The wall has not been dismantled. It has been moved indoors.
Molina moves the installation outside the divine decree. Between God’s natural knowledge of what creatures could do and God’s free knowledge of what they actually will do, Molina inserts middle knowledge: prevolitional knowledge of what any possible free creature would freely do in any possible circumstance. These counterfactual truths are available to guide the decree, but their truth is not fixed by the decree.
It matters that Molina did not arrive here by reading Aquinas on simplicity and finding the modal transmission unblocked. His context was the de auxiliis controversy: the fiercest dispute in early modern scholasticism over how efficacious grace and exhaustive providence could be reconciled with genuinely free consent. Báñez and the Dominicans held that God moves the will to its act by a predetermining concursus, and that freedom survives because God moves each thing according to its mode. Molina held that a will moved antecedently to its act is not free in the sense that matters: freedom is the power, with everything required for acting already in place, to act or not to act. The Jesuits answered that predetermination leaves the alternative no genuine work to perform; the Dominicans answered that without it, providence goes blind and grace waits on the creature. The Congregatio de Auxiliis heard the matter for nine years and condemned neither side. Rome declined to place the stop.
The honest claim is not that Molina relocated the boundary because Aquinas’s placement had failed as a piece of modal engineering. It is that the same pressure resurfaced in a different dispute, and that the machinery built to relieve it has the profile the diagnostic seeks. The specification of the tier is where this becomes visible. If the counterfactual truths are determined by God, creaturely freedom seems threatened. If they are determined by creaturely essences, the choices look settled before the chooser acts. If they depend on actual creatures, they arrive too late to guide creation. Middle knowledge therefore occupies a remarkably exact position: prior to the decree, about free acts, true independently of the decree, and still known infallibly by God.
This is not evidence of bad faith. Middle knowledge performs serious theological work, and Molinists offer arguments for each feature. But no case in the tradition displays its function more plainly. The theory needs a body of truths that can inform providence without being authored by providence. The installation stands outside God because placing it within God would threaten either contingency or freedom.
Descartes puts the boundary where nobody else in this sequence does: outside the modal order altogether.
On the doctrine of the creation of the eternal truths, announced to Mersenne in the spring of 1630 and never retracted, the truths of mathematics, logic, and metaphysics are effects of the divine will in exactly the sense that creatures are. God was as free to make the radii of a circle unequal, Descartes says, as he was free not to create a world at all. In the Sixth Replies the point is generalized: every order, every law, and every ground of anything’s being true or good depends on the divine will. And in the letters he is willing to say the thing the position requires and most readers resist: God cannot have been determined to make contradictories incompatible, and could therefore have done otherwise.
Observe what this does to the predicament. The threat was that necessity would travel from a necessary source down a dependence relation to a contingent effect. Descartes cuts the wire at the source. There is no antecedent necessity to transmit, because necessity is itself downstream of the will. Nothing in the essences makes one preferable to another before the choice; the choice therefore cannot be constrained by them, and divine indifference is total. Where the tradition insulates the effect, Descartes dissolves the modal status of the ground.
Descartes treats this opacity not as a defect of the theory but as one of its consequences. Asked how we are to conceive a free choice among contradictories, he replies that we cannot and should not try: our minds are finite and were made able to conceive as possible only what God in fact willed to be possible.
Whether the wire is actually cut is a further question, and it is not clear that it is. Something modal appears to stand before the decree even on this account: that God exists, that God can create, that God cannot make it the case that there is no God. If those hold prior to the willing, the firewall has not reached all the way out, and the transmission question returns at the new location. Amy Karofsky presses the standard version of this objection, adding a second: if God can do anything, then anything is possible, which would make even the impossible possible.
A further argument of hers cuts closer to what is distinctive in Descartes. Against contingentarian theism generally, she poses a Euthyphro-shaped dilemma. Either God chooses among alternatives for a reason, in which case the reason determines the choice and the outcome is not contingent after all; or God chooses for no reason, in which case the choice is arbitrary, anyt;hing may follow, and nothing is necessary. Either nothing is contingent or nothing is necessary. Most of the tradition treats this as a dilemma to be escaped. Descartes takes the second horn deliberately: supreme indifference is not an awkward consequence of the doctrine but its content. The argument functions against him less as a trap than as an accounting. He has purchased immunity from modal transmission at the price of necessity itself, and the position is stable only if one is willing to say so.
Leibniz stops blocking and starts grading. He refuses the earlier appeal to unexplained choice. Nothing occurs without a sufficient reason, not even a divine decision. Contingency must therefore survive inside a system that grants explanation everywhere.
His instruments include the distinction between metaphysical and moral necessity, the claim that possible alternatives remain possible in themselves, and the doctrine of infinite analysis. On one prominent reading, a necessary truth can be reduced by finite analysis to an identity, while a contingent truth requires reasons for reasons without a terminating demonstration available to a finite mind. God grasps the whole infinite order, but we do not.
The wall has become a customs house. Necessity reaches the border but must declare under which description it travels. God is morally necessitated to choose the best, Leibniz says, but not metaphysically necessitated in the sense that alternatives contain contradictions. Critics have long wondered whether the lower grade preserves a genuinely open opposite or merely gives necessity a more accommodating name. The infinite-analysis doctrine faces a parallel question. If the complete reason is there for God, does our inability to finish the proof make the truth contingent, or only make its necessity inaccessible to us?
Spinoza is the outlier who builds nothing. Every junction the preceding tradition guarded, he argues the guard away: no volitional cutoff, no middle tier, no synchronic power for the opposite, no lower grade of necessity that leaves reality otherwise, no refuge in an unfinished analysis. From the necessity of the divine nature, things follow as they do. The distinction between nature as productive and nature as produced survives, but it is causal and aspectual, not a border between necessity and contingency.
Spinoza pays for this in daylight. Freedom must be reconstructed within necessity; contingency is reassigned to the imperfection of our knowledge. Yet his refusal is philosophically clarifying. He functions as the control case. The other systems can be measured against a thinker who accepts the consequence they are designed to avoid, and the fact that the protective machinery simply does not appear in him is evidence of the same kind as its presence elsewhere.
Running the test
Aquinas, largely acquitted
Take the strongest version of the charge: that divine simplicity, or the doctrine of relations that accompanies it, is a device for keeping necessity from travelling down the creative relation.
On priority, the charge fails. Simplicity is not a Thomistic invention and does not enter at the point of modal pressure. It arrives as an inheritance: Plotinus, Augustine, Boethius, Anselm, with a lineage centuries older than the specific worry about modal transmission from a necessary source. Its stated motivation is compositional and has nothing to do with modality: whatever has parts requires something to compose them, and so cannot be first; whatever has an essence distinct from its existence cannot be unique. The doctrine is a conclusion of the argument that gets Aquinas to God at all, not a repair applied afterward.
On independent yield, it fails as decisively. Simplicity does an enormous amount of non-modal work: aseity, uniqueness, immutability, the identity of God with his own goodness, the whole account of analogical predication. Strip out the modal question entirely and the doctrine still has to be there.
On a real alternative, Aquinas is in the clear. The alternatives are not merely admitted, they are catalogued: the divine ideas include orders God did not make, and the text distinguishes what God wills necessarily (his own goodness) from what he does not (everything else). There is a genuine foil, affirmed in the theory’s own voice.
On counterfactual acceptance, the doctrine passes too. Simplicity is held by thinkers who draw opposite modal conclusions from it, and weakened by others (Scotus) without the modal result driving the revision. It is not tuned to an output.
That is four out of five, and the four are not close. Divine simplicity is not a modal firewall. Anyone who reads this diagnostic as an indictment of the doctrine has misread it.
Where the burden actually sits is question four, transmission. Aquinas’s instruments against the collapse are scope devices: the distinction between suppositional and absolute necessity, and the reply on foreknowledge that distinguishes the necessity of a conditional from the necessity of what it is about. Both have real logical content, and both are answers to arguments that transmit necessity along an inference. The simplicity problem is not of that form. If the act of willing this world just is the necessary divine essence, there is no consequence relation to disambiguate: the argument runs through an identity, and a scope distinction has nothing to operate on. The Thomist’s remaining options are structural: either the difference between willing this order and willing another is no intrinsic difference in God at all, so that worlds differ while God does not; or simplicity is qualified; or some restricted form of the collapse is accepted.
The verdict is differentiated. The doctrine is acquitted. The narrower claim that the divine will’s relation to creatures adds nothing real to God is not a firewall either, since the theory of relations that supports it is independently developed and generally applied. What remains is an unpaid explanatory debt at a precisely identifiable node, and a live contemporary literature arguing about whether it can be paid.[1]
Molina, pressed
Middle knowledge comes out differently, and the difference is instructive because it is not a difference in the sincerity of the author.
On independent yield, Molina does very well. The tier buys exhaustive providence without determination, an unmysterious account of foreknowledge, congruous grace, and (four centuries later) the machinery of the free-will defense. Nobody should call this a decorative patch.
On a real alternative, he does better than anyone else in the survey. The modal field is the widest in the tradition and it is layered: the possible worlds of natural knowledge, the feasible subset the true counterfactuals leave available, and the actual world selected from among those. An omnipotent will bounded by contingent truths it neither authored nor can alter is a striking piece of metaphysics, and the alternatives in it are load-bearing rather than decorative.
On priority, the answer splits, and the split is the interesting part. The tier was built for a real, prior, independent problem (grace and consent) and that counts genuinely in its favor. But the problem it was built for does not specify it. Nothing in the de auxiliis dispute requires that the relevant truths be prevolitional, or contingent, or ungrounded in essences, or infallibly known. Those four properties are fixed by a different requirement: that the truths inform providence without being authored by it. The motivation is prior; the specification is not.
On transmission, the answer is a bedrock. What blocks necessity from running through is that the counterfactuals are contingent, prior to every act of will, and — on the developed view — grounded in nothing. The grounding objection was raised at once, in the Thomist form that truth follows being and that prior to the decree there is no being of the relevant creatures for these truths to follow from, and it has been pressed in the same shape ever since. Grounded in the divine nature, the counterfactuals inherit necessity and determine the creature. Grounded in the creature’s essence, they determine the choice. Grounded in the creature’s actual choices, they arrive too late. Grounded in nothing, they are brute: determinate contingent truths about wills that no will, no nature, and no fact makes true. Sophisticated Molinists now accept the last horn openly. That is candor, and candor is not the same as an answer.
On counterfactual acceptance, the case is at its weakest. Suppose the tier turned out, on analysis, to necessitate the creaturely acts it reports. No Molinist would keep it. The properties are specified to exactly the tolerance the problem requires, and the position has no independent attachment to any of them.
The result: the strongest firewall candidate in the sequence, but not an unmotivated one, and not a fraud. That is a real finding, and it is a different finding from the one Aquinas gets. A diagnostic that returned the same answer for both would not be worth running.
The awkward question
Amy Karofsky has recently defended necessitarianism with an argument she titles the uncontainable contagion of contingency: if a contingent thing could have differed, then facts and relations involving it could have differed, and mere membership in a world is enough to constitute a relation. Any supposedly necessary item connected to the contingent one therefore appears caught in a network that might have been otherwise. If anything is contingent, everything is; and since radical contingentarianism is false, nothing is contingent.
Two further arguments of hers bear directly on the architecture surveyed above, and both are sharper against the firewall structure than the contagion thesis itself. (A third, the dilemma about reasons for divine choice, has already been used against Descartes.)
The first is a dilemma about criteria. Suppose some criterion distinguishes the contingent from the necessary. That criterion must itself be either contingent or necessary, and falls on one side of the very line it was introduced to draw. This is precisely the question a firewall cannot avoid. If the barrier and all its relevant operations are necessary, why is the outcome on the far side not necessary? If the barrier is contingent, has contingency entered the source, or the structure surrounding it?
The second is an argument against degrees of necessity, and it is aimed, whether or not by design, at the customs house. If metaphysical and moral necessity genuinely differ, some criterion must distinguish them; and that criterion can be neither metaphysically nor naturally necessary without circularity. Leibniz’s grading strategy is the obvious target. Scotus’s distinction between logical possibility and what a cause can actually produce is a further one.
Karofsky’s argument is not the same as the firewall argument. It is stronger and more controversial: it aims to show that contingency cannot be contained at all. The firewall diagnosis makes a narrower claim: that when a theory introduces a boundary exactly where explanatory continuation threatens modal collapse, that boundary deserves scrutiny of a specific kind. Nor is her argument uncontested; Stephen Maitzen has pressed hard on the modal logic and on her treatment of infinite regresses, among other things. Take her contagion argument as a stress test rather than a definition of the phenomenon. If it succeeds, it supplies an independent reason to doubt that contingency can be stably partitioned. If it fails, the differentiated record above still stands on its own evidence.
Run the stress test against the survey and four recurring strategies come into view. They overlap; a single thinker may use more than one, and Spinoza rejects all four.
The first places the barrier in reality. Divine will, asymmetric relations, synchronic power for the opposite, and middle knowledge are all meant to make a real difference to what follows from what. These cases face the modal question directly. The answer may be that explanation need not necessitate, that willing can be free without being an intrinsic alteration, that a power for the opposite is real at the instant of acting, or that counterfactual truths have a special status. But one of those answers must do the work. Merely repeating that God chose freely does not yet explain how a necessary source supports a nonnecessary effect.
The second places the limit in cognition or description. Maimonidean apophasis restricts what we can say. Cartesian incomprehensibility restricts what we can conceive. Leibnizian infinite analysis restricts what finite reasoners can complete. Such limits may be independently justified, and they may explain why no proof of necessity is available to us. What they cannot do by themselves is establish that the world is metaphysically contingent. A boundary on demonstration is not automatically a boundary on reality. It may stop us from watching necessity spread without stopping the spread.
The third works by reclassification. Avicenna’s possible-in-itself beings, Leibniz’s moral necessity, and other graded modalities preserve a contrast while revising what the contrast means. The decisive question is whether the opposite remains genuinely open. Scotus is the tradition’s most serious attempt to answer it in the affirmative, and the attempt succeeds or fails on whether unexercised logical possibility is a real feature of the world or a shadow cast by our descriptions.
The fourth relocates the modal order itself. Descartes alone makes the necessary truths products of divine will, so that there is no antecedent necessity available to travel anywhere. This is the most radical placement in the sequence, and it purchases immunity from transmission by surrendering the very stability the other positions were trying to protect.
None of these questions has a one-line answer. That is precisely why the pattern matters. The same pressure recurs under different terminologies, and the boundary moves: from essence to will, from will to relation, from relation to a power for the opposite at an instant, from that power to counterfactual truth, from counterfactual truth to the modal order as such, from metaphysics to analysis. The architecture changes because each location carries a different cost.
The inversion
Natural theology has often presented the necessary being as the explanation of contingency. Things need not have been, so they point beyond themselves to what must be.
The firewall record suggests something closer to the reverse, and this is the claim I would most like a reader to carry away. Dependence on a necessary source creates the most direct route along which necessity threatens to travel. Classical theism is therefore not an easy home of contingency. It is one of the most demanding settings in which contingency must be preserved and explained. The theory makes the world absolutely dependent on a metaphysically necessary source, then must explain why the modal profile of the effect does not simply inherit that of its ground.
Humean mosaics and brute-fact naturalisms have an easier story about preserving contingency for an unglamorous reason: their dependence relations are weaker, or they stop earlier. That does not make them explanatorily superior. They pay in bruteness what theism pays in insulation. A world that declines to explain its basic arrangement will not suffer modal transmission from an explanation it never offered. The price is that the arrangement remains unpaid for.
The interesting comparison is therefore not between theories with costs and theories without them. There are no theories without them. It is between different ways of paying. The theist buys explanation and inherits the problem of modal transmission. The brute-fact theorist blocks transmission by refusing the explanatory ascent. Spinoza purchases complete intelligibility by surrendering alternativity. Leibniz tries to keep both explanation and contingency by grading necessity. Descartes keeps the ascent and the alternatives by making necessity itself a creature. Molina locates the alternative in truths prior to the decree. Scotus locates it in a power for the opposite that survives the act of willing. Aquinas locates it in a will whose relation to creatures adds nothing real to God.
Once the ledger is visible, the familiar dialectic changes. Contingency is no longer the free starting point from which the necessary being is inferred. It is an achievement for which the completed system must account.
What follows
Identifying a firewall locates a burden; it does not discharge it, and as the Aquinas case shows, running the test can also lift one. Several of these boundaries have serious independent arguments behind them. Aquinas’s account of divine goodness is not an evasion. Molina’s middle knowledge is not a decorative patch. Scotus’s synchronic contingency is a genuine advance in modal theory, not a defensive maneuver dressed as one. Maimonidean negative theology does not exist only to conceal a modal problem. Leibniz’s distinction between kinds of necessity belongs to a much larger account of reason, value, and freedom. Even Descartes’s universal possibilism, which looks from outside like the most desperate placement in the survey, follows from a considered view about what omnipotence requires.
What the survey does show is narrower, and harder to dismiss. It shows that a specific pressure recurs across systems with almost nothing else in common, that something is built at the same joint each time, and that the something is different each time, which is what one would expect if the pressure were real and the solutions were attempts rather than inheritances. It also shows that the diagnostic can distinguish among cases: applied to Aquinas it mostly acquits, applied to Molina it mostly does not, and the difference tracks features of the theories rather than sympathies.
The five questions are worth restating now that they have been used. Was the stopping point motivated before the modal danger appeared? Does it solve a problem other than preserving contingency? Does the theory identify a genuinely open alternative, or only redescribe a necessary outcome? Does the explanatory relation transmit necessity, and if not, what blocks the transmission? Would the boundary still be accepted if it produced the opposite modal result?
A boundary is earned when those questions have answers. It is suspect when its only defense is the very contingency it was introduced to save.
Which returns us to the map. The monsters and the wall are the same fact from two sides: from outside, a survey that stops; from inside, something built at the place where it stopped. The map may be right. There may really be dragons. But the serpents are drawn at the edge of the survey, and someone should ask whether the surveyor stopped because the land ended or because he did.
Sources and further reading
Avicenna, The Metaphysics of The Healing, Books VIII–IX; and the overview in Avicenna’s Metaphysics.
Al-Ghazālī, The Incoherence of the Philosophers, especially Discussion 1.
Maimonides, The Guide of the Perplexed, I.50–60 and II.13–25; see also the Stanford Encyclopedia overview.
Thomas Aquinas, Summa Theologiae I, q. 13, a. 7; q. 14, a. 13; q. 19, aa. 3 and 8; q. 28, a. 1; q. 45, a. 3. Online text.
John Duns Scotus, Lectura I, d. 39, and Ordinatio I, dd. 38–39; translated with commentary in A. Vos et al., Contingency and Freedom: Lectura I 39 (Kluwer, 1994). For the background against which the break is visible, Simo Knuuttila, Modalities in Medieval Philosophy (Routledge, 1993). See also Duns Scotus.
Luis de Molina, On Divine Foreknowledge: Part IV of the Concordia, trans. Alfred J. Freddoso (Cornell, 1988); Thomas P. Flint, Divine Providence: The Molinist Account (Cornell, 1998); on the grounding objection, Robert M. Adams, “Middle Knowledge and the Problem of Evil,” American Philosophical Quarterly 14 (1977): 109–17, and William Hasker, God, Time, and Knowledge (Cornell, 1989). See also Middle Knowledge.
René Descartes, letters to Mersenne of 15 April, 6 May, and 27 May 1630 (AT I 145–53; CSMK 23–26); Sixth Set of Replies (AT VII 431–36; CSM II 291–94); letter to Mesland of 2 May 1644; letter to Arnauld of 29 July 1648. See also Descartes’ Modal Metaphysics, and Harry Frankfurt, “Descartes on the Creation of the Eternal Truths,” Philosophical Review 86 (1977): 36–57.
G. W. Leibniz, “On Contingency” and Theodicy; see also Leibniz’s Modal Metaphysics.
Baruch Spinoza, Ethics, Part I, especially propositions 16, 29, and 33. Online text.
Amy Karofsky, A Case for Necessitarianism (Routledge, 2021), ch. 3; the contagion argument is §3.4, “The Uncontainable Contagion of Contingency,” where the criterion dilemma also appears; the dilemma about reasons for divine choice is at §2.18; the argument against degrees of necessity is in ch. 4; the objections to Aquinas and Descartes are in the Introduction, “Some Medieval and Modern Philosophers,” pp. 10–12. See also her “Suárez’ Influence on Descartes’ Theory of Eternal Truths,” Medieval Philosophy and Theology 10 (2001): 241–62. Publisher page. For criticism, Stephen Maitzen, “Necessitarianism Under Fire.”
[1] The contemporary modal collapse literature is arguing about precisely this trilemma. See R. T. Mullins, The End of the Timeless God (Oxford, 2016), ch. 4, and “Classical Theism” in the T&T Clark Handbook of Analytic Theology (2021); Christopher Tomaszewski, “Collapsing the Modal Collapse Argument: On an Invalid Argument against Divine Simplicity,” Analysis 79 (2019): 275–84; Joseph C. Schmid, “The Fruitful Death of Modal Collapse Arguments,” International Journal for Philosophy of Religion 91 (2022): 3–22, and “From Modal Collapse to Providential Collapse,” Philosophia 50 (2022): 1413–35; Schmid and Mullins, “The Aloneness Argument against Classical Theism,” Religious Studies 58 (2022): 401–19; and, on the other side, W. Matthews Grant, Free Will and God’s Universal Causality (Bloomsbury, 2019). Tomaszewski’s objection is that the standard derivation is invalid, trading on an illicit substitution into a modal context; Schmid’s response is that the classical theist escapes the collapse only by accepting an indeterministic link between God and his effects, which then costs something at providence. The dispute is a live instance of question four above.
