No Optimal Illusion
March 5, 2015 — 12:04

Author: Michael Almeida  Category: Uncategorized  Tags: ,   Comments: 17

Some theists say that, if there is a best world, then God must actualize it; but if there is no best possible world, then God can actualize a bettered world w. A bettered world w is a world for which there is a better actualizable world w’. The thought that God cannot actualize a bettered world when there is no best world is what we might call an optimal illusion.

Let’s show it’s not an illusion. God cannot actualize a suboptimal world, even assuming there is no best world. Suppose first that there is a best world, wb, in the series S.

S. w0 < w1 < w2 < . . .< wb – 1 < wb

Now suppose there is no best world in series S’.

S’. w0 < w1 < w2 < . . .< wb – 1 < wb  < . . .< w∞.

Now consider the claim that God might actualize w10 in S’ and not fail to be morally perfect, despite the fact that he might have actualized wb in S’. This is allegedly because there is no best world in S’. By parity of reasoning, it follows that God might actualize w10 in S, and not fail to be morally perfect, despite the fact that he might have actualized the best world wb in S. God in S’ is indiscernible from God in S; either they’re both perfect or they’re both not.

Conclusion (1): Either (i) God must actualize a best world, whether or not there is a best world, or (ii) God need not actualize a best world, whether or not there is a best world.

Conclusion (2): It is false that God must actualize a best world, if there is one, and that God might not actualize a best world, if there isn’t one.

Conclusion (3): There is no good argument for (ii). Therefore (i) God must actualize a best world, whether or not there is a best world

• Heath White

Mike,

All you need to defuse this paradox is ‘ought’ implies ‘can’. We can say, if there is a best possible world, God ought to create it. But we cannot say that God ought to create a best possible world if there is no such world, since then even God can’t do it. I don’t know exactly what God ought to do in that case, but it won’t be something impossible.

March 5, 2015 — 12:53
• Michael Almeida

Hi Heath,

(1) No where in the puzzle do I say that God ought to do something. Nor do I assume that God ought to do something. I assume that God is a perfect being. So, ought implies can is not relevant here.

(2) You don’t address the puzzle, which is this: if God actualizes w10 in S’ and is perfect, then if God actualizes w10 in S, he is also perfect. After all, God in S’ is indiscernible from God in S; in virtue of what would he be perfect in S’ and not in S’? They are intrinsically exactly the same.

But suppose you insist that God in S must actualize wb in order to be perfect. Then it must also be true that God in S’ must actualize wb in order to be perfect. In that case they are indiscernible. But wait! The puzzle simply arises all over again as follows. Consider the series S”:

S”. w0 < w1 < w2 < . . .< wb – 1 < wb < wb + 1

S'' also has a best world. It is wb + 1. In order to God in S'' to be perfect, he must actualize wb + 1. But then it follows that in order for God in S’ to be perfect he must actualize not wb, but wb + 1. In that way, the God in S’ is indiscernible from the God in S'', and they're both perfect. But wait! There is yet another series that ends with a best world, wb + 2, and yet another that ends in wb + 3….. And the problem iterates upward. So, there is no world wn in S’ such that God actualizes wn and is perfect. QED

(3) I should mention that it is not uncommon to introduce 'oughts' into this sort of discussion. But what is claimed here has nothing to do with what God ought to do. It has to do with the concept of moral perfection. Suppose that a perfect, concrete equilateral triangle has three exactly equal sides. You might observe that it is impossible that any concrete triangle have three exactly equal sides. I agree. I conclude that perfect, concrete, equilateral triangles are impossible. I say the same thing about moral perfection. A being is morally perfect just in case it never performs a suboptimal action. You might point out that it is impossible for anything never to perform a suboptimal action. If that's true, I conclude that there are no morally perfect beings.

March 5, 2015 — 13:37
• Heath White

Mike,

Re (1) and (3): You do not use ‘ought’ but you do use ‘must’. I took that to be a moral ‘must’, which implies ‘ought’. And ‘ought’ implies ‘can’, so ‘morally must’ implies ‘can’ also. The rest follows.

But maybe what you meant was something different, a metaphysical must. I.e. “Necessarily, if an omnipotent being is morally perfect, it actualizes an optimal world.” From this, it follows that: necessarily, if an omnipotent being does not actualize an optimal world, it is not morally perfect. And that is your thesis, I take it.

So my first-line argument would simply be the Moorean switcheroo: Moral perfection is possible. Therefore, if there is an omnipotent being, and actualizing an optimal world is not possible, then actualizing an optimal world is not a necessary condition of moral perfection for an omnipotent being.

Maybe this is more persuasive if I claim that being beyond criticism is possible, and so if actualizing an optimal world is not possible, then an omnipotent being is not criticizable for not actualizing one. After all, it hardly makes sense to criticize God, or anyone else, for not doing the impossible.

Re (2): Your argument is that if creating w10 is consistent with moral perfection in S’, then it is consistent with moral perfection in S. This claim requires that the moral status of an action does not depend on what alternatives are available. But that is implausible. Suppose I give my kid thin gruel for dinner. It makes a great deal of difference, to the moral status of my action, what else I could have given him for dinner. Likewise, it makes a difference to the moral status of God’s action of actualizing w10, whether there was an optimal alternative available.

March 5, 2015 — 14:54
• Michael Almeida

I’m going to go right to what matters, and try to keep that in focus instead of addressing related, but minor issues. You say this,

Re (2): Your argument is that if creating w10 is consistent with moral perfection in S’, then it is consistent with moral perfection in S. This claim requires that the moral status of an action does not depend on what alternatives are available.

I don’t follow the objection. I gave you the alternative to hold that only bringing about wb is sufficient for perfection in S. Right? If that’s true, then wb is sufficient for perfection in S’. Do you deny that? If so, I’m happy, that is what I deny too. But then suppose you decide to say, instead, that bringing about wb is, after all, sufficient for perfection in S’. I will then deny that, since there is another series in which wb + 1 is best (this argument is laid out above). So actualizing wb + 1 must also be required for perfection in S’. But then I will deny that, since there is yet another series (again, the argument is detailed above). At some point you will have to say either:

(i) There is no world wn in the series S’ such that actualizing wn in S’ is consistent with moral perfection, or

(ii) There is a world wn such that actualizing wn in S* is not consistent with moral perfection, but actualizing wn in S’ is consistent with moral perfection.

(ii) is indefensible. Both agents are intrinsically and morally indiscernible. Each could actualize a better world; each actualized the same world. If wn is good enough for perfection in S’, it is good enough for perfection in S*. It has the same properties in both cases. There is no reason why I have to do a bit more for perfection in S* than I have to do in S’.

March 5, 2015 — 15:34
• I don’t see the parity. Assume God aims at the best (which, if he’s perfect, why wouldn’t he?). If there is a best, there are no countervailing considerations, and God chooses a sub-optimal world, then God’s choice is incoherent with God’s aim. This incoherence results in practical irrationality, and so God is not perfect after all.

On the other hand, if there is a lack of a best, then God has a countervailing consideration (no best), and the common intuition among those in practical rationality literature is that the countervailing consideration either restores coherence or prevents the incoherence from resulting in irrationality. (I refer to aiming at the best and choosing the suboptimal “motivated submaximization”. One relevant paper is here: http://philpapers.org/rec/TUCSAM)

If there is a best, then God is irrational in choosing the suboptimal. If there is no best, then God is not irrational in choosing the suboptimal. So no parity.

March 12, 2015 — 19:00
• Oops. Motivated submaximization is aiming at the best and then choosing the suboptimal *because of a countervailing consideration.* i accidentally left out the starred part.

March 12, 2015 — 19:14
• Michael Almeida

Hi Chris,

In the infinite case, it must be assumed that there is some world whose intrinsic value is sufficiently high for God to actualize it. So, let v be the value of world w10 in S’, and let v be such that, it is not necessary that a world has value v and God does not exist. Now in both S and S’ it is true that God can actualize w11 instead of w10, and w11 have v + 1 value in each. But, we are assuming that God need not actualize w11 in S’, despite the fact that he can and it’s better. The parity is as follows: it is also true in S that God can actualize w11, w11 has value v + 1. To continue, S includes w10, and w10 has value v.

If the intrinsic value of w10 is sufficiently good in S’ that, God might be perfect and not actualize w11, then it is also true in S that w10 is sufficiently good and God might be perfect and not actualize w11. If that’s not parity, I don’t know what is. The question is whether God must actualize the next best world in the series.

Now you might insist that God must actualize wb in S, but not actualize wb in S’. I think that’s incoherent, but no matter. You’re in the same boat with Heath. I said this to Heath.

But suppose you insist that God in S must actualize wb in order to be perfect. Then it must also be true that God in S’ must actualize wb in order to be perfect. In that case they are indiscernible. But wait! The puzzle simply arises all over again as follows. Consider the series S”, which, for all we know, obtains:

S”. w0 < w1 < w2 < . . .< wb – 1 < wb < wb + 1

S'' also has a best world. It is wb + 1. In order for God in S'' to be perfect, he must actualize wb + 1. But then it follows that in order for God in S’ to be perfect he must actualize not wb, but wb + 1. In that way, the God in S’ is indiscernible from the God in S'', and they're both perfect. But wait! There is yet another series that ends with a best world, wb + 2, and yet another that ends in wb + 3….. And the problem iterates upward. So, there is no world wn in S’ such that God actualizes wn and is perfect. QED

March 12, 2015 — 19:22
• Hi Mike,
I appreciate how quickly you always respond to comments. You say, “If the intrinsic value of w10 is sufficiently good in S’ that, God might be perfect and not actualize w11, then it is also true in S that w10 is sufficiently good and God might be perfect and not actualize w11.” This isn’t correct. Countervailing considerations have justificatory strength. The stronger the consideration, the more sacrifice it justifies an agent in making. What’s “sufficiently good” when a countervailing consideration is present will generally be lower than what’s “sufficiently good” when one isn’t present.

March 12, 2015 — 19:44
• Michael Almeida

It is difficult to follow your claim, since you run together points about the intrinsic properties of worlds and points about their extrinsic properties. You seem to want to say this, which I think is incoherent.
1. In S if God actualize w10, and could actualize w11, then he is imperfect (because there is a best world)
2. In S’, if God actualizes w10 and could actualize w11, then he is not imperfect (because there is no best world).

That view is not coherent. God in S is indiscernible from God in S’. There is literally no intrinsic difference between them. One could actualize a world better, so could the other. If one need not, neither must the other. The perfection of God supervenes on his intrinsic properties. If one is perfect, so is the other. But wait!

As I said, it doesn’t matter if you insist on saying that God must actualize the best world in the finite case, the problem for the view that God must actualize different worlds depending on the extrinsic properties of worlds is worse. See the argument above which concedes this claim. I repeat it again. —-

Now you might insist that God must actualize wb in S, but not actualize wb in S’. I think that’s incoherent, but no matter. You’re in the same boat with Heath. I said this to Heath.

But suppose you insist that God in S must actualize wb in order to be perfect. Then it must also be true that God in S’ must actualize wb in order to be perfect. In that case they are indiscernible. But wait! The puzzle simply arises all over again as follows. Consider the series S”, which, for all we know, obtains:

S”. w0 < w1 < w2 < . . .< wb – 1 < wb < wb + 1

S'' also has a best world. It is wb + 1. In order for God in S'' to be perfect, he must actualize wb + 1. But then it follows that in order for God in S’ to be perfect he must actualize not wb, but wb + 1. In that way, the God in S’ is indiscernible from the God in S'', and they're both perfect. But wait! There is yet another series that ends with a best world, wb + 2, and yet another that ends in wb + 3….. And the problem iterates upward. So, there is no world wn in S’ such that God actualizes wn and is perfect. QED

March 12, 2015 — 20:17
• Hi Mike,
I don’t understand what you think I’m conflating. You say: “God in S is indiscernible from God in S’. There is literally no intrinsic difference between them.” This is false, at least if I’m understanding what “intrinsic difference” means. Since God has different options in those worlds, God knows these differences. Differences in knowledge, i would think, count as intrinsic differences. The argument you repeated fails, because the rationality/morality of a choice depends on what other options an agent had. When you change the options, you can change which option God must/can choose if his choice is to be coherent with his aim at the best. Note that the orthodox view in ethics and practical rationality is that what alternatives one has affects what it is moral/rational for the agent to do.

March 13, 2015 — 6:18
• Michael Almeida

Hi Chris,

It’s obvious to me that you are not tracking the argument when you say this:

The argument you repeated fails, because the rationality/morality of a choice depends on what other options an agent had. When you change the options, you can change which option God must/can choose if his choice is to be coherent with his aim at the best

This misunderstanding is down to me; I’ve posted hastily and not clearly. Let me be clearer. I never assume that God must aim at the best. My argument aims to show that God cannot actualize wn when he could have actualize wn+1 in both series (or, needn’t do that in either series). It’s an argument for symmetry. That’s the conclusion of the argument. Now suppose you believe that God can actualize wb – 1 in series S’.

S’. w0 < w1 < w2 < . . .< wb – 1 < wb < . . .< w∞.

I say that God can actualize wb – 1 in series S, too.

S. w0 < w1 < w2 < . . .< wb – 1 < wb.

You want to say that God must actualize the next world, wb, in S, but you also say (I think) that God might actualize wb – 1 in S'. But consider the features of two worlds in S’:

1. It is true in S' that God actualizes wb – 1 and he can do better.
2. It is true in S' that God can actualize wb instead of wb – 1.
3. wb – 1 has intrinsic value v in S'
4. It is true in S’ that God need not actualize the next world wb, since wb – 1 is intrinsically valuable enough.

All of (1) – (3) are true in S as well!

1′. It is true in S that God actualizes wb – 1 and he can do better.
2′. It is true in S that God can actualize wb instead of wb – 1.
3′. wb – 1 has intrinsic value v in S.

But you deny (4) for S.

4′. It is true in S that God need not actualize the next world wb, since wb – 1 is intrinsically valuable enough.

Why is wb – 1 good enough for God in S' and not good enough for God in S? Is it because God could do better in S? Well, God could do better in S', too. Is it because God could have actualized wb in S? Well, God could have actualize wb in S', too. The question is whether God must actualize wb in S if he need not do so in S' The fact that God cannot actualize a best world in S’ (which I of course grant) is irrelevant to the question is whether he must actualize wb in S’; it’s irrelevant to the question of whether only wb (or better) is intrinsically good enough for God to actualize in S’. He certainly can actualize wb or better.

Now take your assumption that God must actualize the best in the finite series S. So, God must actualize wb in S. I say, if that's true–if nothing less than wb is good enough for an absolutely perfect being in S–then God must actualize (at least) wb in S' too. The fact that God cannot actualize a best world in S' is not relevant to whether he must actualize wb in S'. No one is assuming that God must do his best in S'. I'm saying only that, if only wb is good enough in S, then only wb (or better) is good enough in S'. And that begins the argument repeated above. But that's enough for now.

I hope that's clearer.

March 13, 2015 — 8:04
• Hi Mike,

I think i agree with you that this presentation of the argument doesn’t fall prey to my charge you quoted in the most recent comment. I never meant to claim that you introduced the assumption about God’s aims; I did. I’ll continue to work with that assumption until i have some reason to think it is inadequate (though, really, the assumption could be substituted with claims about God’s pro tanto reasons, etc). After all, you were claiming to give an argument that there is no relevant asymmetry between S and S’. Why shouldn’t i be allowed to rely on plausible assumptions about God’s aims in replying to that argument?

I suspect that we’ll never get closer to agreement going back and forth on the blog. I’ll restate the outline of the asymmetry, and if you like (which i certainly understand if you don’t!), you can get more details from the papers in which I defend these views in detail. In S and S’, the options are different. The difference with respect to options leads to the following asymmetry: in S’, God has a countervailing consideration that he doesn’t have in S. This countervailing consideration justifies him in selecting a suboptimal option. This countervailing consideration has finite strength, so it justifies some sacrifice of the good but not all of it (so God can’t choose a world with negative value, for example). In S, he has no countervailing consideration, so God’s aim at the best rationally requires him to choose the best (however good it is). The difference with respect to this countervailing consideration is the asymmetry which explains why God is rational for choosing less than the best in S’ but irrational in S.

Perhaps its worth mentioning that I think it’s very misleading (though this is common in the mainstream literature) to say that, in EverBetter cases, an agent is choosing a world that is good enough. What’s doing the explanatory work is not some intrinsic cut-off regarding good–the good enough–but the limits of the countervailing consideration’s justificatory power. If you want the details on this issue, I’ll need to send you a separate paper that addresses the mainstream literature rather than the one i linked above. I’m happy to send either of the papers upon request.

March 13, 2015 — 11:00
• Michael Almeida

Hi Chris,

This is a little baffling. No kidding, I’m prepared not to disagree (for the sake of argument) with anything you say here.

. . . in S’, God has a countervailing consideration that he doesn’t have in S. This countervailing consideration justifies him in selecting a suboptimal option.

Ok, fine. But I would not call that a ‘justification’. It’s closer to an excuse for not doing better.

This countervailing consideration has finite strength, so it justifies some sacrifice of the good but not all of it (so God can’t choose a world with negative value, for example).

I doubt it, but I concede the point.

In S, he has no countervailing consideration, so God’s aim at the best rationally requires him to choose the best (however good it is).

If you say so, fine. I’ve granted it for the sake of discussion.

The difference with respect to this countervailing consideration is the asymmetry which explains why God is rational for choosing less than the best in S’ but irrational in S.

Again, fine with me.

I actually don’t believe much of this, but I’m perfectly happy not to argue about it. It is orthogonal to the claim that I argue for above. Everything I say is consistent with all of this. Recall, my argument is for the symmetry in the choice of worlds to actualize. I don’t deny that there are reasons that an absolutely perfect being could adduce to excuse (I wouldn’t say justify) his failing to actualize the best world in S’. I don’t deny that there are no such reasons in S. What I argue is that, if God must actualize wb in S, then he must actualize wb (or better) in S’. This is not inconsistent with anything you say here, as far as I can tell.

What you have not provided is a reason to believe that God must actualize wb in S, but can actualize wb – 1 in S’. The alleged ‘countervailing reasons’ do not give us any reason to believe that. At most they give us reason to believe that God might actualize a suboptimal world in S’.

March 13, 2015 — 12:33
• Here’s the connection, which i took to be obvious:

If it is irrational to actualize a world, God can’t actualize it. In S, then, God can’t actualize a suboptimal world, since they are all irrational. He can actualize wb in S since it is the unique best. Hence, in S, he must actualize wb.

For reasons already explained, in S’, there are suboptimal worlds that God is rational/excused/whatever in making. So God can actualize them.

Hence, God must actualize wb in S but not S’.

March 13, 2015 — 13:32
• Michael Almeida

in S’, there are suboptimal worlds that God is rational/excused/whatever in making. So God can actualize them. Hence, God must actualize wb in S but not S’.

Hi Chris,

Well, I didn’t say God must actualize wb in S’. I said God must actualize wb or better. The negation of the latter claim simply does not follow from the asymmetry in countervailing reasons. The alleged ‘countervailing reasons’–which I’ve granted for the sake of argument—give us no reason to believe that God can actualize a world worse than wb in S’. They give us reason to believe that God can actualize a suboptimal world, that’s it. I conceded this for the sake of argument, too. All of this is independent of my conclusion.

March 13, 2015 — 13:51
• Ok, I think i see some of where what I was saying was confusing you. Right, it doesn’t follow that God can create a suboptimal world worse than wb in S’ from what I’ve said. But I’m arguing against this conclusion: “God cannot actualize a suboptimal world, even assuming there is no best world” (original post). You grant for the sake of argument that countervailing considerations “give us reason to believe that God can actualize a suboptimal world.” Granting this sounds to me like you are granting that I’ve given reason to doubt your conclusion. But you don’t. Can you help me understand your conclusion?

March 13, 2015 — 18:16
• Michael Almeida

Hi Chris,

The conclusion you quote above is the one I’m aiming at ultimately. On the way to that conclusion, I had to show that, for some finite S, if God must actualize wb in S, then God must actualize wb (or better) in infinite S’. It was a long discussion to get there. The argument for the more interesting conclusion goes this way. We start with finite series S, as above.

S. w0 < w1 < w2 < . . .< wb – 1 < wb.

And argue that if God must actualize wb in S, then God must actualize wb or better in S'

S’. w0 < w1 < w2 < . . .< wb – 1 < wb < . . .< w∞.

We observe that there exists another finite set of worlds S1 that is exactly like S and S' up to wb. S1 differs from S in having a best world wb + 1.

S1. w0 < w1 < w2 < . . .< wb – 1 < wb < wb + 1.

In S1, God must actualize the best world, wb + 1. But if God must actualize wb + 1 in S1, then God must actualize wb + 1 (or better) in S'. The reasoning to arrive at this conclusion is the same as the argument from S that God must actualize wb or better in S'. But next we observe that there exists another finite set of worlds S2 that is exactly like S' and S1 up to wb + 1. S2 differs from S1 in having a best world wb + 2.

S2. w0 < w1 < w2 < . . .< wb – 1 < wb < wb + 1 < wb + 2.

In S2, God must actualize the best world, wb + 2. But if God must actualize wb + 2 in S2, then God must actualize wb + 2 (or better) in S'. The reasoning to arrive at this conclusion is the same as the argument from S that God must actualize wb or better in S' and the argument from S1 that must actualize wb + 1 or better in S'.

The argument iterates upward in the obvious ways. Since there are infinitely many finite series' of worlds, we arrive at the conclusion that, for every finite world wn, God must actualize a world wn + 1 or better in S'. Since S' contains nothing but finite worlds (I'm assuming the worlds are countable) there is no world wn in S' that God can actualize.

There's probably a simpler presentation of the argument as a mathematical induction, but I haven't worked it up. Anyway, that's more or less how it goes.

March 14, 2015 — 8:17