Does it violate free will if the person was never given the desire to do something in the first place?

[Mijin]

@Half_Man_Half_Wit

First of all, you can’t possibly know that; the set might well be finite, indeed, it might have only one element, for all we know.

You’re the one that made a probability claim.
We’re either reasoning on the basis only of what we know for a fact: in which case it’s infinity over infinity and all bets are off. Or we’re limiting the sets based on nothing more than personal gut feels, in which case I’ll do the same and say the set of universes with omnimax god and suffering is empty.
Pick your poison.

@Half_Man_Half_Wit

Second, there’s no problem at all with defining probability measures on infinite sets.

It depends what we mean by “problem” here, but many statements that can be trivially applied to finite sets yield “undefined” with infinite sets. Anyway, if you believe that you can derive a finite, greater than zero probability from an uncountably infinite set divided by an uncountably infinite set, then colour me impressed. Let’s see your working.

@Half_Man_Half_Wit

Also, if it were possible at all for God to instantiate a world with zero evil (that met other constraints like logical possibility or nontrivial free agency), then by omnibenevolence that’s what they’d do. So in fact it can’t be possible.

I didn’t understand this, but it smells a bit like assuming the conclusion.

@Half_Man_Half_Wit

Even if Godzilla were a consistent being, the expected amount of Godzilla would probably be way, way below one

But are you conceding the point? Even though godzilla is compatible with our universe* we don’t have good grounds to expect to see 1 or more godzillas, and furthermore that “compatible” does not mean anything like “plausible” or “likely”?

* Yes we could argue that the physics of our universe make this or that property of particular godzilla versions impossible, but of course this is irrelevant. We could change the hypothetical to make it physically possible without affecting the point.

[Half_Man_Half_Wit]

@Mijin

Anyway, if you believe that you can derive a finite, greater than zero probability from an uncountably infinite set divided by an uncountably infinite set, then colour me impressed.

It’s possible to make sense of probabilities in infinite contexts using measure theory. And it’s actually part of my daily work—generically, quantum observables have continuous spectra, and the Born rule assigns probabilities to outcomes lying in subsets of those spectra. Here’s an introduction to probability measures on uncountably infinite sets.

@Mijin

I didn’t understand this, but it smells a bit like assuming the conclusion.

If there’s a world W such that God would instantiate it (i.e. it’s compatible with tri-omnihood and whatever other constraints, like free agency, weigh on that fact), and that world contains evil E, there can’t be a world W’ such that God would instantiate it containing evil E’ < E, since then, God could not instantiate W, since W can’t be compatible with omnibenevolence, thus yielding a contradiction.

@Mijin

But are you conceding the point? Even though godzilla is compatible with our universe* we don’t have good grounds to expect to see 1 or more godzillas, and furthermore that “compatible” does not mean anything like “plausible” or “likely”?

If Godzilla is compatible with our universe, then we have grounds of expecting way, way less than one Godzilla—although obviously I can’t put a number on it, whether it’s 10^-50 or 10^-500 or 10^-5000. Expectations don’t have to lie in the range of values for possible outcomes (indeed, they generally won’t—the expected value for a throw of a die is 3.5, for instance). So it’s not that you have to expect either one Godzilla or none.

[Voyager]

@Half_Man_Half_Wit

Without any further definition of ‘minimal’, however, that just amounts to the same thing: there’s some amount greater than zero.

The usual way to amend this is to argue that a certain amount/kind of evil is gratuitous, i.e. such that a tri-omni God both could and would avert it. This leads to the evidential argument from evil.

I think it might be possible to make an argument that zero suffering (I don’t like to use evil since it is ill-defined) is logically impossible, given a wide enough definition of suffering and perhaps free will. So more suffering than this logically necessary amount would be a defeater of omnibenevolence.
If maximal evil is compatible with omnibenevolence, then it must be equivalent to minimal evil. That’s an interesting concept. It seems to constrain the world quite a lot.
And we are all ignoring the problem of heaven, which is an existence proof of the possibility of a world with zero suffering. I’m fine with ignoring it for this discussion.

[Der_Trihs]

@Voyager

I think it might be possible to make an argument that zero suffering (I don’t like to use evil since it is ill-defined) is logically impossible, given a wide enough definition of suffering and perhaps free will.

Don’t make a universe in the first place; then there’s no suffering whatsoever. If the goal is zero suffering, then that’s always an option.

[Max_S]

@Mijin

We see suffering. A lot of it, particularly natural suffering, does not fulfill any wider objective that we can see … each instance leaves an explanatory gap and the more instances, the more implausible that all the gaps will be filled.

Why would it matter if we can see whether observed suffering fulfills a wider objective? That can be deduced. You haven’t asserted the premises necessary to tie observed suffering to the concept of an “explanatory gap,” nor tied explanatory gaps to your main inductive argument.

~Max

[Mijin]

@Max_S

Why would it matter if we can see whether observed suffering fulfills a wider objective? That can be deduced.

It hasn’t been deduced, it’s been hypothesized.
Let me be clear: I have conceded that it’s hypothetically possible for an omnimax god and suffering to co-exist in the same sense that it’s hypothetically possible that humans have 3 hands and we’ve just always counted incorrectly.
It’s a long way off plausible, let alone a logical deduction, that this actually is the case.

@Max_S

You haven’t asserted the premises necessary to tie observed suffering to the concept of an “explanatory gap,” nor tied explanatory gaps to your main inductive argument.

You keep saying that, but what else is needed?
To go back to my first analogy. The claim is that Bob is a vegetarian. You happen to pass by his home, and catch a glimpse of him sat alone at a dining table, with a plate of various roast meats and he’s holding a chicken leg in his hand and moving the chicken leg towards his mouth and opens his mouth as though he is about to take a bite…before your view is blocked.

Do we have reason to doubt the claim that he’s a vegetarian?
The answer is yes, even though, in the technical sense, our observation is not incompatible with his being vegetarian. And even though we can hypothesize reasons that Bob might have for getting so close to looking like he was going to eat meat.

Or do we not have reason to doubt because something is “missing”? What’s missing?

ETA: I suspect as well that you are taking “doubt” to be interchangeable with “disbelief”. This isn’t the intention. You might still believe that Bob is a vegetarian. But even so, your observation should have affected your confidence in that claim.

[Half_Man_Half_Wit]

@Mijin

Do we have reason to doubt the claim that he’s a vegetarian?

Suppose you know Bob has a dog that requires up to 3 kg meat per week. Then clearly, for any amount of meat up to that, you’d be wrong in lowering your confidence that Bob’s a vegetarian. Doing so would objectively be a mistake. However, if Bob significantly exceeds this amount, it would be rational to doubt his vegetarianism.

Now suppose you know Bob has some dogs, but don’t know how many. When you see him buy some quantity of meat, what can you conclude? Well, the logic hasn’t changed: there is some amount of meat x up to which you shouldn’t doubt his vegetarianism, and beyond which this doubt becomes rational. However, you don’t know this amount. So, you don’t know whether the observed evidence agrees with your hypothesis or disfavors it. You need more information.

This is the situation we’re in with a tri-omni God. We know that we should expect some evil, but not how much. So for the evil we do observe, we don’t know whether it’s in agreement with the hypothesis of God’s existence, or disfavors it. Making either judgment then is irrational. We need additional information.

That’s the reason why any formulation of the evidential problem does introduce such additional information, by means of arguing for the probability of some evil being gratuitous. Then, one can use the observed evil to lower belief in the hypothesis of God’s existence, conditioned on the reasonableness of your argument for it’s gratuitous nature.

[Voyager]

@Der_Trihs

Don’t make a universe in the first place; then there’s no suffering whatsoever. If the goal is zero suffering, then that’s always an option.

But then who will kiss God’s butt? It’s obvious from the Bible that this is his most important desire.

[Half_Man_Half_Wit]

@Voyager

I think it might be possible to make an argument that zero suffering (I don’t like to use evil since it is ill-defined) is logically impossible, given a wide enough definition of suffering and perhaps free will.

I think that this is very likely. After all, if there’s several individuals with their desires around, and they won’t all be satisfiable simultaneously at all times—we can’t all eat the same piece of pie. So some desires or intentions will go frustrated.

Also, there’s an argument to be made that a world without suffering—in the sense of frustrated intentions—might literally be a zombie world without conscious experience: theories of consciousness like predictive processing essentially hold that consciousness happens where we have to update our map of the world, because something we intended to do turned out not to be possible. The philosopher Alva Noë gives the example of waking up in a dark room, trying to feel your way around, and stubbing your toe: your intention of movement is frustrated, and as a result of this, you need to update your mental model, which is a conscious signal. So we experience the world where it frustrates our intentions, where we bump into it. If there is no such frustration, basically there’s no need for a model, because everything we intend will succeed, we never stub our toes on the world, so to speak, and can remain in the dark.

[Half_Man_Half_Wit]

@Mijin

Do we have reason to doubt the claim that he’s a vegetarian?
The answer is yes, even though, in the technical sense, our observation is not incompatible with his being vegetarian

Alright, I’ll try a different tack on this for one last time. The above is true, but there’s a condition. The probability that Bob is a vegetarian, given that he’s seen buying meet, is adjusted downwards from the prior probability that he is a vegetarian (i.e. P(\mathrm{isVeg}|\mathrm{buysMeat})<P(\mathrm{isVeg})) exactly if the probability of a vegetarian buying meat is lower than the probability of anyone buying meat, i.e. P(\mathrm{buysMeat}|\mathrm{isVeg})<P(\mathrm{buysMeat}). Then, so to speak, it’s ‘surprising’ that Bob would buy meat if he is a vegetarian. If we didn’t know that to be the case, then making such a probability adjustment wouldn’t be rational.

This is something we can take for granted in the case of vegetarianism, so yes, it’s true that Bob’s buying meat, in the absence of any other information, will lower our judgment of the likelihood of his being vegetarian. But there’s two issues with porting this over to the problem of evil. First, we do have additional information: that the logical problem of evil fails tells us that there is a certain amount of evil we should expect. Second, we don’t know that P(\mathrm{Evil}|\mathrm{God})<P(\mathrm{Evil}), i.e. that the likelihood of evil occurring in a world that God would instantiate is less than the likelihood of evil in any world at all. This is additional information that we don’t get to assume without warrant.

In fact, it seems rather plausible that the evil in worlds which God would instantiate exceeds that in worlds they would not choose to create. After all, most worlds, as far as we can tell, are incapable of containing any complex life at all—this is the famous fine-tuning problem. So in all of these worlds, there would not be any suffering, or evil, at all. But these aren’t worlds God would create, at least plausibly, for the same reason that God-created worlds contain evil: God might want to maximize free agency, for instance. So it’s very well possible that the average world that God would not create contains vastly less evil than should be expected in those they would create, even if there are worlds that God wouldn’t create due to containing horrendous amounts of evil. And with that, it’s also at least possible that the probability of observing evil in worlds which God would create in total is less than the probability of observing evil across all worlds.

Note that I’m not claiming to have established the above: all this shows is that the opposite, namely that we should expect less evil in a God-created world, is likewise not conclusively established. So again, in order to have any occurrence of evil lower the likelihood of God’s existence, we need further argumentation to make it plausible that P(\mathrm{Evil}|\mathrm{God})<P(\mathrm{Evil}).

[Max_S]

@Mijin

The claim is that Bob is a vegetarian. You happen to pass by his home, and catch a glimpse of him sat alone at a dining table, with a plate of various roast meats and he’s holding a chicken leg in his hand and moving the chicken leg towards his mouth and opens his mouth as though he is about to take a bite…before your view is blocked.

Do we have reason to doubt the claim that he’s a vegetarian?
The answer is yes, even though, in the technical sense, our observation is not incompatible with his being vegetarian. And even though we can hypothesize reasons that Bob might have for getting so close to looking like he was going to eat meat.

Or do we not have reason to doubt because something is “missing”? What’s missing?

There is no reason to doubt whatsoever until we add background knowledge that (1) someone holding a chicken leg to his mouth as if about to take a bite is likely eating the chicken leg, and (2) that vegetarians never eat chicken legs.

Would you do me the favor of completing the analogy?

Bob the Vegetarian	Tri-Omni God
Bob	God
is vegetarian	is omnibenevolent
never eats meat	never allows (cosmic) unnecessary suffering
eating meat	allowing (cosmic) unnecessary suffering
Observed holding chicken leg to mouth as if about to take a bite	observed suffering (inference that God allows observed suffering)???
Knowledge that holding chicken leg to mouth is likely eating	hidden assumption that observed suffering is likely unnecessary (in a cosmic sense)???

~Max

[Mijin]

@Max_S

There is no reason to doubt whatsoever until we add background knowledge […]

Firstly, can we start with an acknowledgement that, with the analogy as is, and the obvious common knowledge like the definition of “vegetarian”, we have reason to doubt Bob’s claim?
Because right now, it seems some here are questioning the very notion of gaining confidence or doubt in claims on the basis of (circumstantial) evidence. I think you’re agreeing with me that of course evidence can support or harm a case, without proving or refuting it outright. But I would prefer we spell it out because we aren’t going to get very far without any basis for reasoning about empirical claims.

@Max_S

Would you do me the favor of completing the analogy?

Good job on the table, let’s go into where I disagree and complete (as you put it) the analogy.

Firstly, I wouldn’t include unnecessary suffering in this table. Yes, ultimately we need to arrive at talking about whether it is unnecessary, or the best possible world or whatever, but injecting it in right from the start only serves to confuse matters. Particularly since one of your defences of ePOE was that suffering is not necessarily bad while another is the “best possible world” argument.

For the sake of this argument, I am defining suffering as intrinsically bad, and omnibenevolence as ideally not wanting any suffering to occur. If you disagree with those definitions then we need to stop the table reasoning right there and have a separate discussion about that, because I want to take on one argument at a time.

Now, in terms of the reasoning, I would put it like this:

¬ The expectation was not seeing Bob putting meat into his mouth
¬ The expectation was not seeing God causing suffering

¬ But Bob might have reasons though
¬ But God might have reasons though

¬ Nonetheless the observation in itself is contrary to our expectation, given the definition of vegetarianism
¬ Nonetheless the observation in itself is contrary to our expectation, given the definition of omnibenevolence

¬ Just as each time Bob eschews meat might build our confidence that he truly is vegetarian, so observations of him apparently about to eat meat should reduce our confidence
¬ Just as each time God does kind acts might build our confidence that he truly is omnibenevolent, so observations of him apparently causing suffering should reduce our confidence

[Mijin]

@Mijin

so observations of him apparently causing suffering should reduce our confidence

Missed edit window: scratch the word “apparently” here. We don’t need it, assuming we’re talking about not merely an omnibenevolent god, but an omnimax one.

[Half_Man_Half_Wit]

@Mijin

¬ The expectation was not seeing Bob putting meat into his mouth
¬ The expectation was not seeing God causing suffering

I think it’s a little unfortunate that you’ve chosen this particular framing, because the initial discussion is one about whether God exists, and whether the observation of evil yields grounds for doubting the existence of a (tri-omni) God, while this presupposes the existence of God, and just questions (possibly) their tri-omni nature. These are relevantly different cases: the latter is like seeing, as a child, a parent do something that you deem harmful, which may however be aimed at preventing greater harm coming to you.

In other words, it makes a ‘mysterious ways’-type of defense much more reasonable: if God exists, as this framing concedes, then they’re unquestionably much greater both in knowledge and power than you—so what expectation could we have to correctly frame their actions? Complaints of their harm done towards us may be akin to complaints of the dog upon having to endure a trip to the vet, with needles and being confined to a small box and all: the dog simply doesn’t have the frame of reference to properly contextualize these actions. To it, that its ordinarily kind and beneficial master suddenly subjects it to such needless torture must be a great cruelty.

Hence I think the original framing where the existence of (tri-omni) God is the open question is the better one. And there, the analogy would be that one should be surprised at seeing evil in a world created by a tri-omni God. But again, that is only true if there is less evil in worlds created by a tri-omni God than there is across all logically possible worlds—but this isn’t something you know, so this is exactly the sort of extra information that I’ve been talking about. Again, there may well be less evil in worlds God wouldn’t create, since most of those may be utterly barren.

@Mijin

¬ Just as each time God does kind acts might build our confidence that he truly is omnibenevolent, so observations of him apparently causing suffering should reduce our confidence

The dog judging its owner cruel for taking it to the vet makes an assumption that it has correctly assessed their reasons for doing so. This assumption is clearly an unreasonable one—the owner neglecting their dog’s health concerns, sparing it the trip to the vet, may well be the more cruel one. So this sort of framing assumes that, when we see God causing suffering, that this is surprising given omnibenevolence—so the assumption being made here is that P(\mathrm{suffering}|\mathrm{omnibenevolence})<P(\mathrm{suffering}). But this is something that can be false: a caring owner may subject a dog to more vet trips than an uncaring one. So a justification is needed to make this assumption: it doesn’t just come for free.

In the end, in every form of the argument you’re making, if you’re asserting that evidence E should lower our belief in hypothesis H, you’re always claiming that P(E|H)<P(E). But without any reason for this to be the case, that remains a stipulation that anybody else just can reject as being unsubstantiated.

[Max_S]

@Mijin

Firstly, can we start with an acknowledgement that, with the analogy as is, and the obvious common knowledge like the definition of “vegetarian”, we have reason to doubt Bob’s claim?

Yes, we certainly do.

@Mijin

Firstly, I wouldn’t include unnecessary suffering in this table. Yes, ultimately we need to arrive at talking about whether it is unnecessary, or the best possible world or whatever, but injecting it in right from the start only serves to confuse matters. Particularly since one of your defences of ePOE was that suffering is not necessarily bad while another is the “best possible world” argument.

For the sake of this argument, I am defining suffering as intrinsically bad, and omnibenevolence as ideally not wanting any suffering to occur. If you disagree with those definitions then we need to stop the table reasoning right there and have a separate discussion about that, because I want to take on one argument at a time.

Unnecessary suffering is a critical term. The defenses that I put forward–that suffering is not inherently bad and that this is the best compossible world–are unified, not distinct. If this is the best compossible world, then teleologically speaking, God’s acts of allowing suffering always have instrumental value in facilitating that end. If suffering can have instrumental value (and you seem to admit it can) then it isn’t intrinsically bad and we would not expect an omnibenevolent being to want zero suffering. We expect God to allow some suffering, but we never expect God to allow unnecessary suffering.

Consider an analogy to a parent and child. A parent knows his or her child will necessarily suffer, and had the power to prevent all the child’s suffering by not having children in the first place. By your logic, regardless of the parent’s potential justifications, the act of having children itself appears to be evidence of evil intent. Having the child is an act allowing suffering; if allowing suffering is always evidence of evil intent in the first instance, we have at least some evidence that most parents are evil. Is it reasonable to think of having children as a naturally evil act, and the parent’s potential reasons as mere excuses? I think common sense says otherwise.

This is all very standard, theologically. No major Christian denominations embrace consequentialism (the best compossible world argument is one that coincidentally aligns with theology, which focuses more on glory of God). No major Christian denominations view suffering as categorically bad; all admit it can serve instrumental value. All would reject your argument that allowing suffering is necessarily evidence of evil; most if not all would look at the actor’s intent (which usually does not involve perfect knowledge).

Note that if this is the best compossible world, allowing suffering is not good in itself. Any more suffering than is necessary loses its instrumental value, and an omniscient God would be evil to allow unnecessary suffering. But before we have any evidence of evil, from this line of inquiry, we must have evidence of unnecessary suffering.

~Max

[Mijin]

@Max_S

Yes, we certainly do. [agree about the vegetarian example]

Awesome

@Max_S

Unnecessary suffering is a critical term.

And I maintain that it is confuses the issue.

Why did we not need to set a “unnecessary meat adjacency level” for the vegetarian example? Why don’t we set a threshold of “unnecessary butler did it evidence” or “unnecessary fairy evidence”?

It’s because we need to first start with what an observation implies and then consider the confounding factors or the likelihood of such factors existing. Otherwise we can’t begin to reason about anything.

If omnibenevolence means not wanting to cause suffering, and we’re observing God causing suffering, of course it gives us reason to doubt that God has that particular omni. Not a refutation, but certainly doubt, if we’re being consistent with how we reason about empirical claims.
Or indeed being consistent with observations of God performing healing miracles or whatever as supporting the claim of a loving God – when we could make exactly the opposite argument about a perfectly evil God being forced to sometimes heal people to overall make the worst possible world.

Up to / until evidence of mitigating factors is presented.

[Max_S]

@Mijin

Why did we not need to set a “unnecessary meat adjacency level” for the vegetarian example?

The example is disanalogous in that respect: eating meat is never necessary for a person to be vegetarian, but allowing suffering is sometimes necessary for an entity to be benevolent.

I only admit that an omnibenevolent being wants no unnecessary suffering.

~Max

[Der_Trihs]

@Max_S

I only admit that an omnibenevolent being wants no unnecessary suffering.

No suffering is necessary for a tri-omni creator, therefore there is no such being.

[Half_Man_Half_Wit]

@Mijin

If omnibenevolence means not wanting to cause suffering, and we’re observing God causing suffering, of course it gives us reason to doubt that God has that particular omni.

This assumes that P(\mathrm{suff}|\mathrm{omni}) < P(\mathrm{suff}). Why do you get to assume that? It’s readily clear in the vegetarian case: we know that typically, people have control over what they buy, that they usually mostly buy what they want, that there are few good reasons for a vegetarian to buy meat, and so on; i.e. that typically, being vegetarian means that you’ll buy less meat than the average person.

But none of this is given in the case of God. We have no relevant background knowledge beyond God’s nature. The entailment from this to the above assumption is not trivial, so at least the argument has to be made. First of all, God’s omnibenevolence alone is clearly not enough: a perfectly omnibenevolent, but completely impotent and maximally ignorant deity may be neither able to avert suffering, nor sufficiently knowledgeable to see that what they’re doing is causing suffering. So God needs both a certain power and knowledge to act on their intentions. That’s fine, of course: the God in question is generally thought to be both omnipotent and omniscient (and omnipresent), so there ought to be no trouble on that front (although it should at least be acknowledged that we’re at best doubting the tri-omni nature of God, not omnibenevolence on its own).

But does this nature suffice to entail that P(\mathrm{suff}|\mathrm{omni}) < P(\mathrm{suff})? (‘Omni’ now referring to the conjunction of the three omni-characteristics.) Again, I don’t see that it does. First of all, as already noted, starting out the argument with the admission that some form of ‘God’ exists significantly weakens it: while it would be fallacious ad hoc reasoning to appeal to ‘mysterious ways’-type arguments when the question is God’s existence, if that question is settled, then we should indeed expect the actions of a being of vastly greater power, knowledge, and understanding than ourselves to be largely incomprehensible, and indeed, as impossible to make sense of for us as it is for a dog to understand calculus.

But even beyond that, there are further mitigating factors. One is that even a tri-omni being is not free from constraints. This is the lesson of the free will defense: just because the absence of suffering is God’s intention, it doesn’t automatically follow that it is in God’s power to bring this state of affairs about. Furthermore, as @Max_S also notes, benevolence doesn’t necessarily entail the prevention of all suffering: the dog owner that brings their dog to regular vet visits may be the more benevolent one.

So is it the case that P(\mathrm{suff}|\mathrm{omni}) < P(\mathrm{suff}), even given constraints on God’s power, the possibility of necessary suffering, and our imperfect understanding of God’s actions? It certainly might not be. At the very least, further argumentation is needed to argue that it is—which you have consistently declined to provide.

This also fits very well with the available evidence: evidential arguments from evil, as they are found ‘in the wild’, start with arguments to the effect that some particular kind of suffering is very unlikely, given that there is a tri-omni deity—which then grounds the inference from observing that sort of suffering to a reduction of credence in the existence of that deity. So the observed evidence—i.e. the arguments that are actually being made—is much more in line with the idea that additional argumentation is needed than with the idea that any observation of evil suffices as countervailing evidence. But of course, this particular kind of evidence has so far failed to nudge any assigned credences, and will continue to do so.

[Mijin]

@Max_S

The example is disanalogous in that respect: eating meat is never necessary for a person to be vegetarian, but allowing suffering is sometimes necessary for an entity to be benevolent.

But that’s your position – it cannot be injected as a premise.
I have not agreed that there is such a thing as necessary suffering in this universe*; that it is “sometimes necessary”. That would need to be demonstrated.

I have only agreed (in the context of the logical problem of evil), that it is not impossible.

It’s like if we were talking about fairies.
Are fairies necessarily impossible? No.
Does that mean we can make an argument which includes the existence of fairies is a premise? Also no.

* ETA: I mean in an ultimate sense. Yes, from a human’s point of view, there can be necessary suffering (e.g. side effects of a life-saving drug) because we have limited knowledge and abilities. It’s not evidence of necessary suffering for a God.