Ok, now for the long-delayed fuller response to your comments & questions.
You wrote: “So far as I can see, the powers ontology leaves open a fully deterministic picture, according to which the set of powers in play at any given time is fully determined by the set in play at a previous time. … For if an action or event is brought about by the powers that are in play, then surely the full set of powers in play necessitates the outcome.”
It all hangs on the “surely.” As I see it, there are two lines of argument that would make it be that what you’ve said here would be true. There might be more than two. But there is also (I think) at least one line of thinking that makes it not be true. You can let me know what you think. Maybe Stephen and/or Rani can weigh in on the part that relates to them (Mumford & Anjum, Getting Causes from Powers, Oxford, 2011).
The first way to make your “surely” go through would be to stipulate that to cause just is to necessitate. Or, to put it differently, if we were to stipulate that causation is, by definition, deterministic. Then it would be that no matter what we take causation to be, if y was indeed caused by x, then, manifestly, x necessitated y. This view would disallow anything like probabilistic causation, of course, and – assuming as you mean to equate causation with real necessitation, real determinism – it would be flatly anti-Humean, since for Hume the terms “necessity” & “determinism” alike refer only to the expectation of regular but contingent order. Still, if we equate causation with determinism, then causal relations will be deterministic no matter how causation is otherwise conceptualized.
Second, it’s possible that the “surely” will go through on Stephen and Rani’s model (though they certainly deny that it does). On their model, it does seem as though there is a total number of powers in play at any given moment (represented by a field of vectors), all of which are “on” at all times, we might say. As I understand them, what they call “dispositional modality” refers to the fact that any given power or powers pragmatically designated as a cause may fail to reach the threshold level identified as “effect” — may fail to reach the finish-line, as it were. But given the picture as a whole, viz., one in which it starts to look as though there is just one total process (one total temporally extended display of powers), on-going throughout time — given this picture, I’m not sure that I can, actually, get them out of a charge that a diachronic slice of the process at time 2 is literally just an extension of the process at time 1. I don’t know how to get the needed punctuation in. It’s possible that their answer will be that there is no total number of powers in play at any given moment. They do say that to refer to such a powers-based “total cause” is to refuse to take the antecedent strengthening test. Perhaps one or both of them will want to comment, but I’m willing to grant you that it’s possible that on a picture such as theirs what happens at time 2 will be given by what’s happening at time 1 in the sense that what happens at time 2 is literally the same display of powers, just later.
But neither of these scenarios is obviously correct (though either could be). It’s easy to see that the second one is a substantive picture, such that one might disagree with it. But the first one is too. The first picture is one in which causation just is deterministic necessitation. You could hold this view directly, as various rationalists did, and as Kant did (more or less), or you might be able to get to it as an upshot. There’s nothing obvious about it, though, I don’t think.
Here, meanwhile, is a picture does not (to my mind) require causation to be necessitation: there are things (“things”) with powers to do x, y or zed. While it is true that the powers of things may or may not be expressed, some things are such that the display of their powers in response to the way other things are is fairly regular. If the world consisted purely of such things, and the processes in which they are involved, then the activity of things at time 2 might well follow necessarily from what things were doing at time 1.
But not all things are such that the display of their powers follows regularly upon some prior interaction with some other powerful particular. Some things react to other things only probabilistically. Still other things are able to initiate (or not initiate) activity voluntarily. And because some things are these ways, it is not the case that the totality of what things are doing at time 1 necessitates the totality of what things are doing at time 2. Sometimes I make tea that I don’t end up drinking. Other times I do drink it. The fact that I have the power to make tea necessitates nothing whatsoever in the way of tea-drinking. Rather than pointing to a necessitation relation, the fact that, if I do drink the tea, my doing so had a cause — this fact (if it is one) tells us only that there is something (viz., me, sentient, embodied being that I am) that has the power to bring about tea drinking.
You write: “I find Ekstrom’s point, which you discuss on p. 13, to be very strong. Surely it isn’t enough to say that an action was the exercise of the agent’s powers, when that simply raises the question of why the agent exercised that power to that degree at that time.”
Here too I think you are too quick with the “surely.” I am the cause of me drinking my tea (which is to say, via an arguably misleading locution, that I did indeed drink it). The reason that I drank it is that I wanted to experience the lovely taste & sensation of warmth. But I wasn’t caused to drink it by my desire; I had my desire. And even that isn’t quite right. I just am a kind of thing that can and does desire such a pleasure.
You write: “Without that determinism of powers, does the question have to go unanswered? (The point that the agent has the meta-power to choose when to exercise other powers simply pushes this issue back one step.)”
Here what I’ll say is that the part of the question that, from a substance-based powers perspective, is ill-conceived — that part does go unanswered, because it goes unasked. But there is no causal question that is thereby being ducked. What is the cause of my drinking the tea? Me. Why did I drink it? For the reasons I said. Does that mean that it was the reasons that caused the tea drinking, and not me? No. Reasons can’t pick things up. Or drink. Did the reasons cause me to cause it? No. Contra Tim O’Connor, that is not the right way to conceptualize the nature of a embodied subject.
You write: “So, I guess what I don’t get is why the powers ontology should reconfigure the debate in the way that you outline, rather than simply clarifying the metaphysics underlying determinism but leaving the question of the compatibility of free will with determinism untouched.”
From the perspective of a substance-based powers ontology, causation is production, not the fact of necessitation. At the level of a single phenomenon, there is causal determination but no determinism because the exercise of a cause’s power can be thwarted. At the level of the totality (in which the thwarting interferer can be taken into account), there is causal determination but no deterministic relation between time 1 and time 2 because not all “things” behave in the requisite way.
Does this help?