Much of the difficulty in analyzing the notion of power comes from the various limitations of creaturely power: our powers come and go, and they are not infallible (sometimes we have the power or ability to do something, and nevertheless fail to do it when we try). These are the sorts of cases which derailed conditional analyses of power. However, an omnipotent being would have none of these limitations. In our paper, Alexander Pruss and I exploited this fact to develop an analysis of omnipotence, or unlimited power, without the need for a prior analysis of power. This approach has the advantage of allowing us to understand omnipotence without first solving the puzzles about power. A disadvantage, however, is that it does answer all of the questions of the form “does God have the power to…” (which I take to be equivalent to “can God…” on the most usual meaning of the latter in these sorts of questions). Indeed, without an analysis of power, our account does not answer any questions of that form. What it does do is tell us enough about what an omnipotent being would be like that if we did have an analysis of power we would presumably be able to give the correct answer to each such question and explain why these are the correct answers.
One such question which is of particular interest is, “does God have the power to do evil?” According to the Pearce-Pruss theory, the claim that God is omnipotent entails the following two claims:
(1) If God were to will that he should do evil, he would intentionally bring it about that he does evil.
(2) God, being perfectly free, is free in choosing whether to do evil.
(Since we don’t have an analysis of God’s perfect freedom, we don’t have an official, precise formulation for condition (2), but surely perfect freedom implies freedom in all of one’s choices.)
Now we want to add the following:
(3) It is impossible that God should do evil.
(1) and (3) imply:
(4) It is impossible that God should will that he should do evil.
We argue that this is consistent with (2) if the impossibility follows from God’s character and/or choices in the (as yet unspecified) right way.
Now, the question I want to ask is, given this situation, can God do evil? That is, does God have the power or ability to do evil?
The following principle is extremely plausible:
(5) If any agent S has the power to perform an act A, then possibly S does A.
In other words, power implies possibility.
Erik Wielenberg denies this principle precisely because he holds that God would have power to do evil despite lacking the possibility of doing evil. I have argued that Thomas Reid rejected (5) on the same grounds.
As I said, (5) seems extremely plausible to me. In fact, until recently I thought that the case of God was the only conceivable grounds for questioning it. The other day, Abelard Podgorski, one of my colleagues at USC, suggested to me that the principle might also be violated by a being who was necessarily in a Frankfurt case.
Now, Frankfurt meant his case to show that an individual could be morally responsible (and, indeed, free) despite being unable (i.e. lacking the power) to do otherwise. However, an alternative interpretation of this case is that the individual has the power or ability to do otherwise, but it is not possible, in the present circumstance, that she successfully exercise it. One might think that one of the reasons it makes sense to ascribe the power to the individual in question is that there are other circumstances in which she might very well exercise it. But what if no circumstances in which the individual might exercise the power were possible? Could it still be ascribed to her?
Say ‘S is necessarily Frankfurted with respect to Aing’ (‘NF(S, A)’ for short) iff:
(i) S deliberates about whether to A
(ii) In every possible world where S is about to choose to A, some other agent intervenes to prevent S’s doing A
(iii) In (at least) some worlds the following counterpossible conditional holds: if, per impossibile S should choose A and no intervention should occur, then S would do A.
The intuition is supposed to be that such an S might count as having the power to A, despite the fact that, due to some other agent’s intervention in every possible world, it is not possible that S do A.
Now, my first thought was this. Suppose we simply accept the following conditional:
(6) If NF(S, A), then S has the power to do A despite lacking the possibility of doing A.
This would entail the falsity of (5) only if it were possible that NF(S, A). After all, it is true that if 2+2=5, then 2+2≠4, but it does not follow from this that possibly 2+2≠4. Furthermore, it surely seems implausible that anyone is necessarily Frankfurted.
But wait! Here is an argument that we ourselves are necessarily Frankfurted with respect to certain actions. Most theists agree that there are some conceivable states of affairs so bad that it is not possible that God permits them. But if God exists and is omnipotent necessarily, then, necessarily, nothing happens unless God permits it. Thus the states of affairs which are too bad to be permitted by God turn out to be impossible (despite being perfectly conceivable). For instance, perhaps it is impossible that every sentient being suffer excruciating pain beginning now and continuing through all eternity.
Now consider this: a being’s freely refraining from producing some great evil is a good state of affairs, not an evil one. But if Frankfurt cases are compatible with freedom, then perhaps it is possible that God brings it about that some creatures are in Frankfurt cases with respect to producing states of affairs so evil that God cannot allow them to be produced. For instance, perhaps there is a possible world in which, at this moment in history, I am standing in front of a button and I know that if I push the button every sentient being will suffer excruciating pain beginning now for the rest of eternity. It is not possible that God permit me to push the button (unless he prevents the button from having its effect), but if Frankfurt cases are compatible with freedom, then perhaps it is possible for me to be in that situation. If so, then I am necessarily Frankfurted with respect to causing every sentient being to suffer excruciating pain for the rest of eternity (and so, presumably, is everyone else).
I do have the intuition that in the worlds where God does not intervene I am free to push the button (and cause the resulting suffering), and that I have the power to push the button (and cause the suffering). Of course, all my powers are subject to overriding by God, so perhaps it makes no difference to my having a power if, necessarily, God always overrides it.
We can, of course, insert the ‘flicker of freedom’ strategy, and insist that there is some alternative course of action I am free to do and have the power to do, but deny that that action is causing the suffering. In fact, it seems plausible that possibly God allows me to push the button and then blocks its effect. It’s an important part of Frankfurt’s original case that the intervenor prevents the choice rather than the resulting action. Perhaps no possible being is necessarily Frankfurted with respect to choosing a course of action. At any rate, this idea about necessary Frankfurt cases seems to me to cast further doubt on (5).
(cross-posted at blog.kennypearce.net)