I will argue that no amount, severity or distribution of evil constitutes evidence that a world is not morally perfect. Let W be a compensating world and define ‘compensating world’ as follows.
C. W is a compensating world iff. the value of W = N, and for any agent S, action A and event E, if S performs A or event E occurs and the value of W is thereafter N - M, then a compensation P occurs such that P + (N - M) = V = the value of W’.
Intuitively, for any action or event that produces an overall negative outcome, the additional value P makes the value of W equal the value of W’. Now we can let W’ be any world we like. Suppose W’ is a morally perfect world. Now consider two hypotheses.
Hypothesis H1: The overall value of W determines it’s level of perfection.
Hypothesis H2: The overall value of W does not determine its level of perfection.
Notice that either H1 and H2 is true. But if either is true, then so is T1:
T1. There is no amount, severity or distribution of evil E such that W contains E only if W is not morally perfect.
Worlds with extreme amounts of severe evil might be morally perfect worlds. Such worlds might be morally perfect worlds either because (i) the overall value of W determines its moral perfection and some morally perfect compensating worlds have large amounts of severe and unjustly distributed evil, or because (ii) the overall value of W does not determine its level of perfection. Now two points about the epistemology of compensating worlds.
E1. It cannot be determined apriori that a world W is not a compensating world.
E2.It cannot be determined aposteriori that a world W is not a compensating world.
Conclusions:
C1. If (ii) is true, then no amount, severity or distribution of evil E constitutes evidence that W is not morally perfect.
C2. If (i) is true, and (E1) and (E2) are true, then no amount, severity or distribution of evil E constitutes evidence that W is not morally perfect.
Why believe (E1) and (E2) are true? I take (E1) to be obvious. There is no apriori reason to believe that a world is not a compensating world. But what about (E2)? It seems almost certainly true. There are so many epistemically possible compensating worlds consistent with eveything we might observe. For some compensating worlds, the value increases imperceptibly over time; for other compensating worlds, all of the greatest value occurs in the distant future; for still others, much of the compensating value occurs long before sentient beings exist. The problem is that, despite the epistemic possibilities, we have no idea what the distribution is of metaphysically possible compensating worlds. Since we don’t know that, we don’t know whether an observation is inconsistent with a possible compensating world or not. But then we don’t know when an observation of evil constitutes evidence that a world is not morally perfect
I think that the argument for (C2) may rest on a view of evidence which is open to counterexample. The view of evidence is that if there is prima facie evidence E for a hypothesis H, but if P were true, then H would be false; and we cannot determine, either a priori or a posteriori, whether P is true, then E isn't evidence for H.
Suppose: (1): There is substantial prima facie evidence that Caesar crossed the Rubicon in 49 B.C.
(2): If the world was created five minutes ago, with a population that "remembered" an unreal past, then it is false that Caesar crossed the Rubicon in 49 B.C.
(3): We can't tell, by a priori or posteriori arguments, whether the world was created five minutes ago, with a population that "remembered" an unreal past.
Then,(4): (1) is defeated.
(4) is implausible, so the view of evidence on which the argument for it rests should be rejected.
Thanks David, but I don't think I follow the argument. This is Russell's skeptical argument, right, and the skeptical hypothesis is in (3). Your rejection of (4) seems to be dogmatic (a popular but strange response to skeptical hypotheses) as far as I can tell, since you've given me no reason to believe that the skeptical hypothesis is false. As with all decent skeptical hypotheses, it fits the phenomena as well as any non-skeptical hypothesis.
In any case, my argument is not a skeptical argument. I don't think there is prima facie evidence that ours is not a compensating world. I do not advance any wild skeptical hypotheses. One could take the atonement as evidence that our world is a compensating world, for instance, but I don't expect everyone to believe that the atonement occurred.
Mike,
Would you mind being a little more explicit about how you got C1? I don't immediately see how it follows. In fact, I don't see how it follows that (ii) implies the mere possibility of a world with high quantities of evil being morally perfect.
(ii) says this,
(ii) the overall value of W does not determine its level of perfection.
If that's true, then no amount, severity or distribution of evil E constitutes evidence that W is not morally perfect. That is, you cannot determine moral perfection by appeal to the existing value or disvalue in a world. This is basically (oversimplified a bit) the possibility that the moral value of a world might be independent of welfare of sentient beings. For instance, think of the moral ranking of worlds as being a function of only the justice/injustice in those worlds.
Thank you. In that case, I will take issue with your reasoning. (ii) says that the OVERALL value of a world does not determine its level of perfection. I take this to mean that the sum total of the values in the world do not determine its level of perfection. If this is not what you mean, I'm sorry for misunderstanding you (although I would then be confused on what you mean by (i)). However, if this is what you mean, it does not rule out situations where, for instance, the level of perfection is determined by lack of negative values. Say we quantify values in a very small world, such that it has five values, each given the cardinal 1. Say this world would have an overall value of 5. Perhaps it is morally perfect because of its lack of negative values, despite having an overall value that seems relatively small. It seems that in such cases, a world's level of perfection can be determined by values without being determined by overall value.
I am not saying that this is the most intuitive way to determine a world's level of perfection, but I do think it is consistent with both (ii) and the denial of C1.
Perhaps it is morally perfect because of its lack of negative values, despite having an overall value that seems relatively small. It seems that in such cases, a world's level of perfection can be determined by values without being determined by overall value.
Let n be the overall value of W. I take it that a negative value would include one like -2. Suppose W included that negative value, in it's overall value, n. Would the overall value of W change if it lacked that negative value? Yes, of course it would. It's overall value would then be n+2. So, I don't know what you mean when you distinguish overall value and the lack of negative value.
But (ii) reads "the overall value of W does not determine its level of perfection."
For your response to work, it should read: "nothing that affects the overall value of W determines its level of perfection." This, of course, would leave middle ground between (i) and (ii).
I think the distinction can be important in many cases, including your own. Here are a couple examples.
Suppose that our original 5-valued world had another value added in. Let that value be -a. Now if the world were compensating, another value (say for simplicity, +2) would be added into it. Thus, the addition of a negative value (along with a compensating value) did not change the overall value of the world. But it certainly can, I think, change how we judge the level of perfection of the world.
Here's another one. Suppose world A has five values, all given the number 1, for a total world value of 5. Suppose world B has two values, 15 and -10, for a total world value of 5. If the overall value of the world determines its level of perfection, the perfection of these two worlds are indistinguishable. We are in agreement here. You have argued that if the overall value of the world does NOT determine the moral perfection of a world, we cannot make a judgment between them based on the information given. But I disagree. Why can't we make a judgment that A is better because it has fewer bad things? Again, I'm not claiming that this is the judgment I would make. I am only claiming that this judgment is consistent with what we have assumed (namely, that the overall value does not determine a world's level of moral perfection).
But you are still committed to saying that the overall value of worlds does matter to its moral perfection.
Proof:
1. Suppose for reductio that the overall value of W does not matter to its moral perfection.
2. Let -N be the amount of evil such that, you would say, any world with -N or greater evil is imperfect (as in your examples above).
3. It follows from (2) that the overall value of any morally perfect world must exceed -N.
4. Therefore (from (3)) any world whose overall value does not exceed -N is morally imperfect, contrary to our suppostion.
5. Therefore, the overall value of a world matters to its moral perfection. (Contradiction!, 5 and 1)
I agree with your point. It does matter. If I was unclear, I apologize. I am just trying to draw a distinction between it mattering and it determining. I find a very large difference between the two, and I think that you've conflated them in some of your argumentation.
Let's see whether I've conflated determining and mattering. It is a sufficient condition on a world being imperfect that it fails to exceed -N in overall value. That obviously determines a world's moral status. So, I don't see the worry. If a world has an overall value of less than -N, it is NECESSARILY NOT a perfect world. And by virtue of the converse of that conditional, if a world is perfect, then it is NECESSARILY a world with greater overall value than -N. So, the overall value of a world does determine it's level of perfection.
But in case there is still any doubt, just to get on to something more interesting, let me hereby stipulate that if it's a sufficient condition on a world being imperfect that it fails to exceed some -N in overall value, then the overall value of a world determines its moral perfection. The stipulation does not affect the argument in any interesing way at all.
This time I don't believe I can grant your argument. Maybe to avoid further confusion, defining terms would be a good plan. I've been working with a fairly commonsense concept of "determining" a level of perfection. But in the interests of clarity, let's see if we can formalize it.
Let's say that the overall value V of W determines W's level of perfection under these conditions:
(1) For any world W, if W attains a certain level of perfection, then any world W' with value V' will attain that level of perfection if V' > V.
(2) For any world W, if W fails to attain a certain level of perfection, then any world W' with value V' will fail to attain that level of perfection if V'
You've argued that we may define perfection in such a way that there is a lower bound on perfection. In your examples, (2) is obviously true. However, I fail to see that your examples cover (1). I argue that we may conceive of moral perfection in such a way that there is no upper bound on imperfection. That is, there is no such integer N such that we know that any world W with value V such that V > N is a morally perfect world. Under my definition, examples like the above would still fall under (ii).
If this has not been your sense of the word, I suppose we've been talking past each other. At any rate, I will try interacting with the stipulation that you just gave. I will treat it as a definition of determining moral perfection. If that's not how you intended it, I apologize, and we should probably figure out what we're talking about (by either using my proposed definition or another) before continuing this debate.
Under your definition, if there exists a value N such that any world with a value less than N is not morally perfect, the overall value of the world determines its moral perfection. If this is the case, I no longer believe that (i), (E1), and (E2) jointly imply (C2). My objection is the same as it was earlier, just in a different place now.
Suppose one wants to take the position that any evil (something with negative value) makes a world morally imperfect. This is the denial of your conclusion in (C2). This sets a lower bound on the value of a perfect world at 0, and we thus meet the definition for value determining perfection. Now suppose there is a great evil, but there may be great compensating goods of which we are unaware (i.e. suppose (i), (E1), and (E2)). In fact, suppose we KNOW of great compensating goods such that the value of the goods outweigh the value of the evil (I ask you to grant this because it makes things simpler and doesn't hurt your argument). That is, the original value of the world W is less than the value plus evil M (which takes a negative value) and compensation P. W
I'm sorry to sound like a broken record, but I think this is more than a semantic difference. If you define things as I proposed, (C1) doesn't follow. If you define things as you proposed, (C2) doesn't follow. Either way, I don't immediately see a way around it.
I think what you want to argue is not that (C2) does not follow, but that T1 does not follow from H1 and H2. Isn't that right?
T1. There is no amount, severity or distribution of evil E such that W contains E only if W is not morally perfect.
You want to argue that it is a sufficient condition on being imperfect that a world contains a certain amount of evil N. If that's so, then (contrary to T1) there is an amount of evil E such that W contains E only if W is not morally perfect. Is that about right?
Yes, you are right. Oversight on my part. If T1 does not follow, then obviously that will lead to my criticisms on C1/C2 (depending on definitions). But it would've made more sense for me to consolidate things and just question T1 in the first place. Don't ask me why I didn't do that.
It isn't that I necessarily disagree with T1, I just don't think it follows from (H1 OR H2). And I do think it's pretty central to your argument.
It is central to my argument. I argue here Good enough that there is no credible position on value according to which, for some -N, if a world includes -N or worse, then it is imperfect. Here's the problem in brief. Let W be a world that contains an amount of evil equal to -M, where -M is slightly less evil than -N. And let W contain no positive value. So we do not know that W is imperfect, givne the amount of evil it contains. Let W' be a world whose evil is equal to -N and whose positive value is infinitely positive, oo+. We know that W' is imperfect. It follows that W' is a worse world than W. That's not credible. W has no positive value and only slightly less negative value than W'. W' contains infinitely positive value and only slightly more (make it infinitessimally more) negative value than W.
Okay. I'm sorry that it took so much confusion to get to the central point here. I don't believe I'd read your other argument, but here is my comment on your summary as it relates to my argument.
You have shown in a way that convinces me that there is no such negative value N such that any world that includes a value less than or equal to N is imperfect. However, I do think that special cases arise when we get near zero. This seems like a strong initial intuition, so we should make sure to consider it. I will grant that we cannot necessarily say that a world in which somebody gets a cut on their finger and then a brain freeze while eating ice cream is necessarily imperfect while a world where someone only gets a cut on their finger is not. But can we say that a world in which someone gets a cut on their finger is necessarily imperfect while a world in which nothing bad happens is not? This seems much more intuitive. Two stains don't mar perfection much differently than one does, but one stain mars perfection quite differently than zero.
Thus, I would say that it has not been ruled out that for the value N = 0, N has a value such that any world containing a value strictly less than N is imperfect.
Thus, I would say that it has not been ruled out that for the value N = 0, N has a value such that any world containing a value strictly less than N is imperfect.
I might be losing the point. We were focused on worlds containing a certain amount of disvalue N and trying to determine whether there is any N such that a world that contains N is imperfect. You seem to be proposing that any world with any amount of disvalue (i.e. any N s.t. N less than 0) is imperfect. But this is susceptible to the same sort of counterexample. Let W include the infinitessimal disvalue -1/oo and the positive value oo+. Let W' include no disvalue and no positive value. Since W contains some disvalue it is imperfect, and since W' does not contain any disvalue, we cannot conclude that W' is imperfect. But it's obvious that W is better than W'.
I agree with the first statement.
I first fell in love with scientology at 12 and ever since have been promoting the campaign. As the dalai lama stated 'The man who smiles when things go wrong has thought of someone to blame it on'
We know that W' is imperfect. It follows that W' is a worse world than W. That's not credible.
I don't see how it follows that W' is worse than W. W meets the condition for being perfect that it contain less evil than N; but unless this is a sufficient condition for perfection, W might fail to meet other requirements and also be known to be imperfect. Suppose, e.g, that one holds a perfect world has to contain both evil less than N and infinite positive value.
Aside from this, why should one accept the principle, "If we know that W' is imperfect but don't know that W is imperfect, then W' is worse than W."? To know only that W', but not W, has been ruled out for perfection still leaves us in the dark about the relative rank of W and W'.
The point is that I think there is special behavior around 0. Although I think that your infinitesimal account is fine for negative values, your most recent example can be debated. One can say that a world with an infinity of positive values but one infinitesimally small negative value IS in fact more imperfect than a world with no values either way. This is not an argument that I personally sympathize with, but I do think that it is a coherent and respectable position to hold.
I'm with you on disagreeing with the idea. If you don't think it's a coherent or respectable position, then we'll agree to disagree on the point. If you do think it's coherent and respectable, then I think your argument ought to deal with it.
David, the argument went this way.
I could have concluded this better. What's not credible is that, given the distribution of value in W and W', W' is known to be imperfect and W is not. It's obvious that W' is better than W.
One can say that a world with an infinity of positive values but one infinitesimally small negative value IS in fact more imperfect than a world with no values either way.
I doubt you believe that's a respectable position. Imagine being given the choice between living eternally in a neutral mental state--neither positive nor negative--and, on the other hand, suffering the smallest possible pain and thereafter living eternally in perfect bliss. Suppose all else is equal, nothing more is on the line. Do you really think it is a respectable position to advise someone to take the neutral state eternally? Suppose it is someone who trusts you and is likely to do what you advise. I think it would be a serious wrong to advise someone (someone naive enough to take the advice) to take the neutral state.
But now are we mixing up the value of a world with its moral perfection? Of course the one with the small pain and vast bliss is more valuable than the other. But the question is whether it's necessarily less imperfect. That is a question that doesn't have quite as easy of any answer.
An instance that comes to mind is the [possibly apocryphal] story of a musician who went to a solo piano concert and, when asked about it, claimed it was terrible, for there was not one wrong note. Missing a note can be symptomatic of great positives, and when it is accompanied by these positives, it can lead an overall very valuable concert. On the other hand, a piece can be technically perfect and still be lacking in some quality that draws you in. But still, the former is called imperfect and the latter is not, value claims aside.
What's not credible is that, given the distribution of value in W and W', W' is known to be imperfect and W is not. It's obvious that W' is better than W.
Mike, thanks for the clarification. You are---forgive me---perfectly right that W' is obviously better than W. But why is this inconsistent with W', but not W, being known to be imperfect? Suppose someone claimed a perfect golfer would win every golf tournament he entered. I've never entered a golf tournament, so I'm not known to be an imperfect golfer. Tiger Woods is known to be imperfect, since he has failed to win many tournaments. I don't think it would be right to say that the claim about the perfect golfer is obviously mistaken, because Tiger Woods is an enormously better golfer than I am.
James, you say,
One can say that a world with an infinity of positive values but one infinitesimally small negative value IS in fact more imperfect than a world with no values either way.
Then you say,
Of course the one with the small pain and vast bliss is more valuable than the other. But the question is whether it's necessarily less imperfect.
I'm pretty confused. In the initial quote above you adduce facts about the relative value of worlds to make a claim about their relative perfection. Now, in the second quote, you want to distance claims about relative perfection from claims about relative value of worlds.
The latest view introduces a notion of 'perfection' that is too technical to matter. I took us to be talking about the familiar notion of perfection according to which the perfection of a world is some function of the goodness of a world (I don't use 'goodness' here in a way that precludes deontological assessments of worlds. There are representation theorems (see G. Oddie, among others) that allow to rank to 'deontological goodness' of worlds in the same way that we rank them axiologically). For reasons of simplification, we stipulated that the only morally important features of the worlds under consideration were the values described: W contained an infinitessimal amount of disvalue and an infinite amount of positive value. W' contained no positve or negative value. Those are the relevant features. Under those assumptions you say,
... a world with an infinity of positive values but one infinitesimally small negative value IS in fact more imperfect than a world with no values either way
Given that those are the morally relevant features of those worlds, it's not coherent to claim that the values they contain are not directly related to their perfection. You were defending the reasonableness of a position which claimed that W' is more perfect than W, given just those distributions of values. That view is not reasonable, I claim, since it is intuitively obvious that W' is a morally better world.
I'm not sure where your confusion lies, as my two lines that you quoted say pretty much the same thing.
At any rate, I don't see this notion of perfection as too technical or lacking intuitive appeal. I believe that David, in fact, is using a very similar (although perhaps not identical) notion.
David, you write,
Suppose someone claimed a perfect golfer would win every golf tournament he entered. I've never entered a golf tournament, so I'm not known to be an imperfect golfer. Tiger Woods is known to be imperfect, since he has failed to win many tournaments. I don't think it would be right to say that the claim about the perfect golfer is obviously mistaken, because Tiger Woods is an enormously better golfer than I am.
In your case we'd need to add that we know that Tiger is a better golfer than you are. Ok, so let's be sure this is analogous to the case I'm describing. So, you and Tiger enter a tournament, and no one else enters. I know that Tiger is an imperfect golfer and I know that you are a worse golfer than Tiger. If so, then I know it's false that a perfect golfer will win. If Tiger is an imperfect golfer and you're an even worse golfer, then I don't think I can coherently conclude that you might be a perfect golfer. Otherwise, I'd have to affirm that, possibly, perfect golfers are worse golfers than imperfect golfers. Is there a world in which that statement is true?
Mike,
Thanks--that's extremely helpful. If A is known to be worse than B, it can't be the case that A is not known to be imperfect but B is known to be imperfect. But this leaves open that B can be known to fail a necessary condition for perfection, while A is not known to fail this condition.
In my Tiger case, I am not known to fail the requirement of winning every tournament entered, while he is known to fail this. But I am known to be worse, and hence imperfect, by other plausible requirements, e.g., my inability to break 100--for a single hole.
Likewise, in your original example, it is only relative to the requirement that a world must contain less evil than N that W' is known to be imperfect and W isn't.
By the way, I'm eagerly anticipating your response to the comment at 2:20 P.M. on March 2.
I am not known to fail the requirement of winning every tournament entered, while he is known to fail this.
No, right, but that wasn't part of the case, I don't think. The question concerned whether a perfect golfer would do X. We can know that a perfect golfer wouldn't do X, for any X, since Tiger and you are imperfect golfers. Sorry, I missed the other comment. I'll try to reply.
good...thanks