Why Contingency Won’t Stay Put

Most philosophers treat contingency as metaphysical common sense.

The cup on my desk might not have existed. The meeting might have been cancelled. The tree might have grown two feet to the left. The universe itself might have been otherwise. We begin with these familiar thoughts and then ask how much necessity we can responsibly add.

Amy Karofsky reverses the pressure.

In A Case for Necessitarianism, she argues that necessitarianism is the view that absolutely nothing about the universe could have been otherwise in any way whatsoever, while “contingentarianism” is the opposing view that at least something could have been different. Her challenge is not merely that contingentarianism lacks proof. It is that contingency is harder to contain than its defenders suppose.

Call this the contagion of contingency.

The ordinary contingentarian wants a modest position. Not everything is up for grabs. Some things are necessary, some things are contingent, and the two can live together peacefully. Perhaps the laws of nature are necessary but the initial conditions are contingent. Perhaps God is necessary but the created world is contingent. Perhaps essences are necessary but accidents are contingent. Perhaps the modal structure is fixed while ordinary facts vary.

Karofsky’s reply is that this picture is unstable. Contingency does not sit politely inside its assigned compartment. If one thing could have been otherwise, then its relations to other things could have been otherwise. And if those relations could have been otherwise, then the other things related to it could have been otherwise in at least that respect. Contingency spreads by relation.

Suppose C is contingent. C might not have existed, or might have been different. But C is related to M. If C had been absent or different, then M would have stood in a different relation to C. So M, too, would have been otherwise: not necessarily in its intrinsic nature, but in its relational profile.

Here, a careful reader will object. The intrinsic/extrinsic distinction exists precisely to block this kind of inference. A necessitarian about intrinsic natures can grant that relational properties shift across counterfactuals while denying that anything important about the underlying entities is contingent. To say M’s relational profile could have been otherwise is not yet to say M itself could have been otherwise. Not, at any rate, in the sense that matters for whether the world had to be this way.

The argument cannot be dismissed by this move, but it does need to meet it. Karofsky’s reply, in effect, is that being related to one’s actual surroundings in the actual way is not a thin or detachable feature. M’s relational profile is one of the ways M actually is. To say M could have stood in different relations is to say M could have been an entity with a different profile. And that, on her view, is enough. The disagreement at this joint is real, and the contagion argument carries only as much weight as one’s willingness to count relational variation as genuine variation. But the contingentarian who resists here owes an account of why intrinsic natures are the only modally serious feature of an entity, and that account is harder to give than it looks.

Once the step is granted, the spread is hard to stop. Everything that is related to a contingent entity inherits at least relational contingency. The surrounding modal structure loosens.

This is why contingency can’t be quarantined.

The contingentarian wants a firewall. The firewall says: contingency goes this far, but no farther. The world may vary, but God does not. Accidents may vary, but essences do not. Initial conditions may vary, but laws do not. Created reality may vary, but the divine nature does not. Ordinary facts may vary, but modal principles do not.

But the firewall has to stand somewhere. It must distinguish the necessary from the contingent. It must mark one side as fixed and the other as variable. And that very distinction now becomes part of the modal machinery in need of explanation. Why is the line here rather than there? Why are these features allowed to vary but not those? Why does contingency stop at exactly this border?

Karofsky’s diagnosis is that such borders are not neutral. They are doing metaphysical work. And if they are doing metaphysical work, they cannot simply be asserted. They must be grounded. But once the grounding relation is introduced, the old pressure returns: is the ground necessary or contingent?

If the ground is necessary, then what follows from it seems necessary as well. If the ground is contingent, then the basis of the modal system has itself become contingent. Either necessity flows downward, or contingency leaks upward.

The “seems necessary as well” hides a controversy worth naming. In contemporary grounding theory, Fine, Rosen, and others have distinguished grounding from modal entailment precisely so that one can hold X grounds Y without thereby holding necessarily, if X then Y. On these views, grounding is hyperintensional and need not transmit modal status downward. The argument therefore depends on a fairly strong, modally-loaded conception of grounding: one on which what is wholly grounded in a necessary basis is itself necessary. That conception is defensible, and arguably the natural one for the explanatory work the firewall is being asked to do, but it is not free. A contingentarian has the option of retreating to a thinner notion of grounding on which the downward flow does not occur. Whether that retreat saves contingency or only relabels the problem is itself a substantive question. The question has to be fought, not assumed.

That is where the firewall fails, where its survival depends on commitments most contingentarians have not made explicit.

This gives us a useful way to distinguish two modal pressures.

Contingency is viral; it spreads horizontally, through relation. Necessity is gravitational; it pulls vertically, through grounding. If one entity could have been otherwise, then anything related to it could have been otherwise at least relationally. If something is wholly grounded in a necessary basis, and the grounding relation transmits modal status, then what is grounded cannot be otherwise either. What follows from necessity, follows necessarily.

The gravitational pressure is the Spinozistic worry behind many modal-collapse arguments. Classical theists know it well. If God exists necessarily, and if the world follows from God’s nature, then the world seems necessary. To preserve contingency, the theist often appeals to divine will: God necessarily exists, but freely chooses which world to create. God’s nature is necessary; God’s creative act is contingent.

That is a modal firewall.

Karofsky’s argument suggests that the firewall merely relocates the problem. A sophisticated version of the theistic reply, due to Leibniz and his heirs, holds that God acts on inclining but not determining reasons. There is sufficient reason for the choice without metaphysical necessitation. The hope is to thread between necessity and arbitrariness: the choice is rational, grounded in reasons rooted in the divine nature, yet not strictly forced by them. Whether this position is ultimately coherent is a question careful philosophers have disagreed about for three centuries; this is not a knockdown matter. But the pressure Karofsky identifies survives even against the Leibnizian. If the inclining reasons fully explain the choice, the choice looks necessary after all. If they only partially explain it, the unexplained residue is just the contingency the firewall was supposed to keep out of the foundation. The Leibnizian threading is delicate, and the question is whether anything ultimately occupies the narrow space it requires.

The same pattern appears outside theology.

The metaphysician who says that laws are necessary but initial conditions are contingent faces the question: why those initial conditions rather than others? If the answer is grounded in the necessary structure of reality, the initial conditions are not contingent after all. If there is no answer, then brute contingency has entered at the base. And if brute contingency is allowed at the base, then it becomes unclear why anything is necessary at all.

The modal-firewall project is caught between two dangers. On one side is necessity flowing downward: whatever is rooted in necessity becomes necessary. On the other side is contingency leaking upward: whatever depends on contingency becomes contaminated by it.

The contingentarian wants to preserve a mixed world: some necessity, some contingency. But a mixed world requires principled boundaries. And principled boundaries require explanation. Once explanation is demanded, the boundary either hardens into necessity or dissolves into contingency.

This is the deepest force of Karofsky’s argument. She is not merely saying that contingency is mysterious. She is saying that contingency is structurally unstable. It cannot be introduced as a harmless local fact. It changes the modal status of whatever it touches.

Here the argument leans on a further premise that should be made visible: that everything in a world is connected to everything else at least by co-belonging to that world. World-membership is a thin relation, and whether contingency propagates through such a thin relation is precisely the kind of question the contingentarian will want to press. If world-membership is too thin to carry modal weight, then the contagion has natural barriers after all, and the firewall can stand at the edges of any region rich enough in genuine relation. Karofsky’s argument, fully stated, requires that no relation a contingent thing bears to anything else is so thin as to be modally inert. That is a strong claim, and it is where the contingentarian’s last redoubt is built.

Even granting that the redoubt eventually falls, the larger structure remains:

If we want contingency, we need firewalls. If we want explanation, the firewalls need grounds. If the grounds are necessary, contingency collapses. If the grounds are contingent, necessity is infected.

Karofsky’s necessitarianism is one way of accepting the result rather than resisting it. Instead of trying to preserve contingency while preventing modal collapse, she denies that there was any real contingency to preserve. The world is not one selection from a space of alternatives. It is not a contingent actualization of one among many unrealized possibilities. It is simply what is.

That conclusion will feel too strong to many readers. It may even feel existentially suffocating. The serious resistance to Karofsky will not come from the intuition that things obviously could have been otherwise (that intuition is what her argument puts under pressure) but from one of the technical joints: the insistence that intrinsic natures are the only modally serious feature, the appeal to a hyperintensional conception of grounding, the Leibnizian threading between necessity and arbitrariness, the claim that no relation is too thin to transmit modal status. Each is a place where a careful contingentarian can plant a flag. None is obviously indefensible.

But the power of Karofsky’s argument is that it makes the ordinary view look less innocent. The thought that things “could have been otherwise” seems easy only until we ask what, exactly, makes that otherwise possible. Once we ask that question, we find ourselves committed to one of those technical positions whether we wanted to be or not.

Contingency stops looking like common sense.

It starts looking like a leak.