A Quick Libertarian Argument for Universalism

A.P.

Many libertarians reject (1). Libertarianism is simply the thesis that we sometimes act freely and that this fact is incompatible with the truth of determinism. The libertarian need not--and probably should not--be committed to the claim that all our decisions are causally undetermined. Moreover, many libertarians would want to insist that, as our character is formed, partly through our own free decisions, we become less and less free to choose otherwise. So it seems your argument relies on an implausibly strong characterization of the libertarian thesis.

Also (2) is questionable, as it is stated. Suppose that R is determined not to accept God's offer of salvation. This is consistent with all his decisions being free. Suppose that every time R is presented with an opportunity to accept God's offer of salvation, he has the following two options: (i) reject or (ii) postpone decision until a later date. Suppose that it is not determined whether he choose (i) or (ii). Then his choice will be undetermined, indeed all his choices can be undetermined, even if it is not possible for him to accept God's offer of salvation.

Hey AP,

How do you get from each recalcitrant agent might be reconciled in premise (2) to each recalcitrant agent will be reconciled in premise (3)? Let A be an agent in w that is a recalcitrant rejector of God. It is true that A might actualize world w' at t in which he repents. So (2) seems safe. But it doesn't follow that A will actualize w' a t. You make it seem just a matter of time before each recalcitrant agent repents. But certainly there are recalcitrant agents who could but won't ever repent. It is therefore pointless for God to wait on such agents.

But if Libertarianism is true, it seems to me that it will never be true, if Libertarianism is true, to say of anyone that they could but won't ever repent. This is too deterministic.

So, your argument depends on the claim that libertarianism entails that, if S is free to do A (in this case to repent) then it is impossible that A does not freely repent at some time t. Let the modality take wide scope, so that you are saying something like this:

P. Necessarily, if S is libertarian free to repent, then, for some time t, S does repent.

But how could (P) be true? (P) entails that it is impossible that S is libertarian free and that S freely chooses at each temporal point t not to repent. Now, if you assume that time is infinite, then it is true that the chances of S's choosing not to repent over an infinite amount of time go to zero. But in that case S might still never go to heaven. Since for every finite amount of time he might fail to repent. On the other hand, if time is not infinite, then the chances are small, but neither infinitessimal nor zero, that S never repents. In either case, it is possible that S never repents and never goes to heaven.

On yet a third hand, you might just be saying that the chances are low that someone consistently fail to repent, given an extremely long period of time to freely modify his views, unencumbered by bad character, bad influences, etc. I guess that's right.

I'd want to go the route Justin suggests. Not all forms of psychological determinism are equal.

A.P.

Your modification still doesn't save (1). The falsity of determinism is consistent with the moderate form of libertarianism I previously described. Indeed, if any form of libertarianism is true, there must be some indeterminism in the world. Indeterminism is also consistent with our forming our characters (via genuinely free decision) such they we are no longer able to do otherwise.

I take libertarianism to entail (at least) that for any time t, there are two possible futures available at t compatible with the history and laws holding until t. That might be a little strong, given some inevitable backtracking, but I think AP's argument is assuming nothing more than this. If that's compatible with some forms of determinism, the argument seems to run the same way. But maybe I'm misreading the argument.

I would question issues packed into 5 and 6. God may frequently desire humans to be reconciled to him not in general, but under certain conditions. For example, he may desire that some divine action is sufficient for a certain person to be reconciled to God within a certain time frame.

There are all sorts of conditions which might make a person repent. For example, perhaps two incarnations would do the trick for some people. This does not mean we are logically committed to two incarnations.

In other words, there simply being a condition under which a person will repent (whether it's a time interval or a piece of knowledge or an event or whatever) does not mean God will actualize that condition, or wait for that condition to materialize. God's will with respect to reconciliation is probably not as general as your argument requires. It involved reconciliation in certain time intervals, under certain conditions, and so on.

A.P. - Does your position commit you to saying that either (1) there are no (human) agents in heaven with libertarian freedom, or (2) there are (human) agents in heaven with libertarian freedom, and each will eventually freely choose to do something sinful?

It seems that your argument could easily be converted into one that commits you to either (1) or (2) (that is, that libertarianism is incompatible with the "impeccability" of those in heaven). Or can you point to some relevant disanalogy?

To echo Justin, libertarianism is not generally taken to entail a specific view about the number, scope, or proportion of actions/decisions that are free (except that the number of free actions is greater than zero).

You seem to suggest that libertarians must deny (or perhaps just ought to find unattractive) the view that one can (freely) form one's character in such a way as to make some class of actions impossible (from then on) for one to perform. But isn't this in tension with your own view of libertarianism, since it would deny that free agents are free to perform a certain class of actions, namely, those actions that would culminate in one's character being formed and "fixed" in the relevant way? Or perhaps not: perhaps you would say that it is logically or conceptually impossible for there to be actions that could "fix" one's character in this way, because it is logically or conceptually impossible for a free agent's character to be "fixed" at all? As my question about the "impeccability of the saints" suggests, I don't know that libertarians (of a certain bent) will find this latter view all that attractive.

Justin, you write,

Indeterminism is also consistent with our forming our characters (via genuinely free decision) such they we are no longer able to do otherwise.

Can that be right? Let t be a time at which, given my character, I am no longer able to do other than reject God. I don't think I'm libertarian free at t. Apart from that, I'm not sure there are character traits so strong that we cannot act against them.

It does seem libertarian freedom requires the possibility that an agent will forever freely reject God's offer of reconciliation. But if universalism is understood as the claim that all will be saved, as opposed to all must as a matter of logical/metaphysical necessity be saved, then this abstract possibility does little to counter universalism.

I think AP's point is a good counter to those open theists who insist on the likelihood of damnation, but it does not counter those who simply deny the logical necessity of universalism.

An interesting question is here is the relationship between inevitability and necessity. Some events are inevitable, but nto necessary. if you flip a coin forever, it will turn up heads, but this is not logically necessary. Universal Salvation may then be inevitable, even though it is not necessary.

An interesting question is here is the relationship between inevitability and necessity. Some events are inevitable, but nto necessary. if you flip a coin forever, it will turn up heads, but this is not logically necessary

If your coin C is fair and "forever" is some extremely long, finite amount of time, then there is some world w in which C never comes up heads. In that world w, not everyone is saved. So universalism, if true, is contingently so. But everyone concedes that, I think. It's trivally true that (Vx)(x is a rational creature only if x is saved) in worlds where God exists alone or with no rational creatures.

Mike,

I don't take libertarianism to entail that for any time t there is more than one possible future. I take libertarianism to be the conjunction of the free will thesis (the thesis that some human persons sometimes act freely) with incompatibilism. Libertarianism is thus consistent with some of our actions not being free.

As for the issue of character formation, one might say that an agent who can no longer do otherwise is no longer free, or one could say that the agent's freedom derives from his ability to do otherwise earlier in his life. I prefer the latter, since the agent (assuming his character was formed in part by his free choices) is still the source of his actions, even though he can no longer do otherwise.

Mike wrote (in response to Justin):

Can that be right? Let t be a time at which, given my character, I am no longer able to do other than reject God. I don't think I'm libertarian free at t. Apart from that, I'm not sure there are character traits so strong that we cannot act against them.

Insofar as libertarianism is simply the conjunction of (i) incompatibilism is true and (ii) the free will thesis is true, then what Justin says is correct. Let's simply stipulate that free will is the kind of control over one's action needed to be morally responsible for that action. (Yes, this is a stipulation, and PvI for one would reject it, but this stipulation is about as widespread in the free will debates as anything is.) In fact, all of the leading libertarian views either explicitly endorse that there could be free actions at t that an agent could not have done otherwise than at t, or are compatible with such a claim: Kane, O'Connor, Stump, Zagzebski, Clarke, Pereboom (just off the top of my head).

Not to shamelessly self-promote (it's self-promoting, but of the 'shamed' variety!), Tim Pawl and I have a forthcoming paper on roughly this issue, though our paper deals explictly with heaven rather than hell. However, I think it's pretty obvious that if what we say is true, then something structurally parallel could be said about hell.

You asked Justin, but I'll answer. Yes, Twearth is (by definition) a libertarian world (or, better, a world which contains libertarian freedom). Libertarianism per se takes no stand on the quantity or distribution of free actions.

Kevin, you write,

. . . all of the leading libertarian views either explicitly endorse that there could be free actions at t that an agent could not have done otherwise than at t, or are compatible with such a claim

But you also say,

Insofar as libertarianism is simply the conjunction of (i) incompatibilism is true and (ii) the free will thesis is true

I believe that (i) and (ii) entail that determinism is false (or that indeterminism is true). If indeterminism is true, then I don't see how there could not be at least two possible futures at any particular temporal point t. If there is only one possible future at t, then the chances of a particular future is certainty or 1. But then it is not indeterminate what occurs at t. So, I'm guessing you have in mind by the claim that, for some t's, "an agent could not have done otherwise than at t", is not that there is not more than one possible future at t, but that one possible future is overwhelmingly likely. Is that about right?

Mike,

You're right that if there is any indeterminism, then the thesis of causal determinism is false. But even if not everything in the universe is determined, there could well be parts of the universe that are determined (either by the conjunction of the distant past and the laws of nature, or by one's freely formed character).

Another way to perhaps put the point is that there are other ways to rule out alternative possibilities than the truth of determinism.

You also say:
So, I'm guessing you have in mind by the claim that, for some t's, "an agent could not have done otherwise than at t", is not that there is not more than one possible future at t, but that one possible future is overwhelmingly likely. Is that about right?

No, I actually have in mind the stronger thesis. Take the case of Odysseus. Suppose that at time t1 he know that he will be in the presence of the sirens, and also knows that unless he takes certain steps, he will be lured by their song and crash his ship on the rocks, resulting in the death of his entire crew. So at time t1 he freely binds himself to the mast. Shortly thereafter at t2, he sails past the sirens, but given that he is bound to the mast, he cannot stear his ship into the rocks because of their song. Odysseus is praiseworthy for saving his men at t2, even though (given what he did at t1) he couldn't have done otherwise. Any libertarian view which has a tracing clause can account for this.

Kane explicitly talks about such a case in discussing Dennett's "Luther example". O'Connor discusses them in "Freedom with a Human Face." Stump explicitly mentions the redeemed in this context. Even van Inwagen, who reserves 'free will' for having the ability to do otherwise allows for responsibility to trace backwards in this way in his restrictivism.

A.P.: If what you are asking is, would a world in which there is only one free action, all others being determined (and thus not free), be a world in which libertarianism is true, then it seems the answer is 'yes'. This would just be a consequence of the definition of libertarianism I gave earlier, which is, I take it, the *standard* definition.

Mike, you write in response to Kevin:

"If indeterminism is true, then I don't see how there could not be at least two possible futures at any particular temporal point t."

There could be pockets of "local" determinism even in a world in which indeterminism is true.

Kevin, you write,

But even if not everything in the universe is determined, there could well be parts of the universe that are determined (either by the conjunction of the distant past and the laws of nature, or by one's freely formed character).

What do you mean 'by laws of nature'? If the world is indeterministic, the laws are at best statistical. Are you saying that in indetermnsitic worlds there are non-statistical laws: certainly, this isn't so. Otherwise there would be indeterminstic worlds in which everything is causally determined. Statistical or indeterministic causation allows for exceptions. So this won't get you what you want. It won;t ensure that all are bound for heaven. On the other hand, one's freely formed character C won't get you to the conclusion that, necessarily, if C then e, for the relevant e. Character does not entail the relevant action. But if entailment fails, then necessarily, there is a world w in which both C and ~e. Again, we don't reach the conclusion that all are bound for heaven.

Justin you write,

There could be pockets of "local" determinism even in a world in which indeterminism is true.

Right, Lewis (in)famously talks this way. I wonder what he could mean in indeterminstic worlds. Here is one thing he does not mean: that there are times at which events are determined by history + laws, where the laws are exceptionless regularities. If he did mean that, then there could be indeterminsitic worlds in which every event is causally determined. I'm pretty sure that can't happen in an indeterminsitic world. But, suppose that is possible. It still wouldn't help. In those causally deterministic enclaves, no action would be free.

"If indeterminism is true, then I don't see how there could not be at least two possible futures at any particular temporal point t."

Suppose there are exactly n free agents in an indeterministic world W. At time t, all n free agents are asleep (or unconscious, etc.). Suppose they won't wake up until t2. Then at t1 (t1 is between t and t2) there is only one possible way W can be arranged (given, that is, how things were at t). Of course, there is some sense in which W has more than one possible future at t: after all, at times later than t2, there is more than one way things might go. But for the interval between t and t2, there are no other possible "world-segments."

I've assumed in this example that the only way indeterminism can ever "manifest itself" in W is by the (consequences of) free actions of those n free agents. But isn't that all coherent? Or do you (1) deny that this is an indeterministic world, or (2) deny that this is an indeterministic world without at least two possible futures?

I think the possibility of "only one possible world-segment from t to t2" is what some here have in mind...

Mike,

Suppose the laws governing everything except free choices are deterministic. But if it is a possible world in which there is libertarian freedom, then the laws governing choices cannot be entirely deterministic. Suppose that what one chooses is probablistic depending on one's character. The Luther and Odysseus examples were suppose to show how even if the laws governing choices are probabilistic, one can do things to make these probabilities for certain choices 0 or 1.

You also write:
Statistical or indeterministic causation allows for exceptions. So this won't get you what you want. It won;t ensure that all are bound for heaven.

I concur. I reject guaranteed universalism (though I do think universalism could contingently be true, and hope that it is, even though I'm skeptical that it is). I rather claimed one could freely form one's character in such a way that certain options are no longer open to one. Does this make sense? If not, I'll try again.

AP,

I think that it is the following that some of having a hard time reconciling with libertarianism:

[given] (ii) that God's essence itself is love, it is "morally necessary" and therefore inevitable (given God's omnipotence and omniscience) that all will be reconciled

If libertarianism is true, can't one always freely reject God's offer of reconciliation?

AP:

That might be exactly my view--though I don't know what "moral necessity" means really. Does it mean this we are free, God has resources well beyond what we can immagine, and in our hearts we really want to be saved anyway, so God never gives up, as loving father. We all really want salvation anyway. It sucks to be unredeemed, so even Satan and his minions will, (as Origen reportedly claimed), be reconicled in the end.

The chance that one will remain unredeemed is small but significant. If we are serious about the existential import of our choices, we have to allow that we, not God, has power over our fate (I mean exercised power). God nudges but does not compell.

Kevin:

"If libertarianism is true, can't one always freely reject God's offer of reconciliation?"

I take it the question is not one of logic but of fact. Sure its possible that one may rebell forever. It is possible I will eat only beef liver for the rest of my days. You correctly say "It is possible GK eats only beef liver for the rest of his life" and I can also correctly say "it is false that I will eat beef liver for the rest of my days."

I guess I disagree with AP if he holds that there is logical necessity in salvation. I too think that is incompatible with libertarian freedom.

Suppose there are exactly n free agents in an indeterministic world W. At time t, all n free agents are asleep (or unconscious, etc.). Suppose they won't wake up until t2. Then at t1 (t1 is between t and t2) there is only one possible way W can be arranged (given, that is, how things were at t).

I guess I see no reason to draw this conclusion. Between t1 and t2 there are uncountably many segments--assuming time is dense--and at any one of them there is some non-zero probability that any subset of the n individuals could awaken. There is, in fact, a non-zero probabilitity that any one of them spontaneously disintegrates. There is a non-zero probability that the atoms composing the left arms of exactly n/2 of them fly off in different directions. So, I don't reach the conclusion that there is only one way things could go in the time between t1 and t2. I deny that in indeterminsitic worlds only free choices are undetermined. I'm not even sure that there are such worlds.

Why not simply redefine Universalism to claim that all may be saved instead of all will be saved? God will not close off the possibility of salvation to anyone although there may be at any time be someone who is not saved. Is this not consistent with a Libertarian view of free-will?

Kevin, you say,

Suppose the laws governing everything except free choices are deterministic.

I'm not sure there are such worlds. I guess I don't see any principled basis for the exception, unless we are talking about worlds of disembodied spirits (or something) ungoverned by physical laws.

Suppose that what one chooses is probablistic depending on one's character. The Luther and Odysseus examples were suppose to show how even if the laws governing choices are probabilistic, one can do things to make these probabilities for certain choices 0 or 1.

But these cases do not make probabilities for certain actions go to either 0 or 1. They rather approximate 0 or 1. The psychological equivalent of tying yourself to the mast will not make the chances of surrendering to the sirens 0. Speaking metaphorically, ropes come untied; boats drift in unexpected ways. Strong dispositions against surrendering are no guarantee that you won't. It is possible to act against one's strongest dispositions, and that possibility ensures that the chances that you do not are less than 1.

John,

I think what you say is right: contingent universalism does not contradict libertarianism at all.

But "may" sounds a little weak for universalism. Suppose I claim that while all could be saved, Satan, Dick Cheney, Hitler, and they guy down the street who kicks his cat for fun will likely not be saved, no matter how long God gives them.

All we need to correct this picture is to claim that God's loving nature and power to non-coercively influence us is so strong, that even Dick Cheney will, in the end, be saved.

AP,

There is a way to run this argument where you get the conclusion you want; but it requires middle knowledge. You need only let God actualize all and only those creatures in each world who would freely (libertarian freely!) accept God as savior. In this case, it is necessarily true that every creature that can be saved is saved and no creature that cannot be saved is created.

But there is at least this argument that God cannot do this. It might be true that, among the worlds in which God creates all and only those agents who will freely accept God as savior, there are some (morally) terrible worlds. Maybe some of these world are so bad that God cannot actualize them.

Gordon - doesn't your form of universalism fall foul of Jon Kvanvig's argument that contingent universalism doesn't really solve the moral problem of hell, because there are still possible worlds where somebody remains in hell forever, yet God is still perfectly loving?
I remember you discuss this objection in your seminal 2006 paper 'Univeraslism for open theists'. But I'm not sure I understand your thinking on this.
Do you hold that God's love is consistent with His allowing an individual to choose an existence that is so filled with unending (but self-inflicted) pain, that the life is on balance, a bad thing?
Thanks for some great papers!

AP
As I teeter into territory I am not that familiar with it seems that we have reason to believe that (H) is false in so far as at least, if the story is to be believed, one person went to Heaven on the day Christ was crucified, namely the thief who recognized the innocence of Christ and asked to be with Him in Heaven.

I also do not see how God's will is thwarted. If He wills x then x. I am presuming that there is a difference between God willing x and God desiring x. If x = we will all be saved, then we will all be saved if that is what He willed. But if He wills x' = we act freely then we can act freely if that is what He willed and reject God as our Savior. So I do not see any issue with U' unless there is an 'end of time.' As long as time continues then there can be any number of people acting freely and rejecting God as their Savior.

There seems to be one way for (U') to be false and that is if no rational person would reject God as his or her Savior. If someone does then reject God as his or her Savior then that person is not rational and should not be punished. At some Time t (the end of time) God, because of His loving nature, will cure all irrational people and then all people will knowingly and freely accept God as their Savior and be saved.

Why not simply redefine Universalism to claim that all may be saved instead of all will be saved?

But that won't get you even contingent universalism. It is likely true that (1),

1. (Vx)M(x is saved)

All might be saved.

But (1) doesn't ensure that (2),

2. M(Vx)(x is saved)

Might be that all are saved.

I'm inclined to believe (2) is true, in any case. But I don't see how contingent universalism creates Hell-problems for God unless you assume that at least some worlds in which not all are saved are hell-worlds (worlds in which there is a place like hell). But hell might be contingent, too. In that case, possibly, every hell-world is one in which universalism is true, and universalism fails in some non-hell worlds. Perhaps these are, for some unfortunates, annihilation-worlds or permanent limbo-worlds.

Yazan:

Wow I did not know I wrote anything seminal! Thanks! First, this possible world stuff: I only understand this terminology if it is translated into potentia. So it is possible for me to rob a bank just means that I have that potentiality in this life. When it comes to nonexistent but possible beings, I go Meinongian. I am not sure how much this matters but talk of "there being a possible world" makes it sound like its some place, and even those who are not modal realists can, perhaps, be besotted by the picture the terminology lends itself to.

But anyway, I think God's loving character is compatible with someone not being saved in this world. I don't think its true, but its compatible. What is not compatible with God's loving character is that God gives up on anyone. As long as reconciliation is possible God will be working on it. So its like a human parent, who never really gives up on her child. She will do what it takes to turn him to the good, it may be "tough love" (weeping and gnashing of teeth) or it may be something more pleasant. carrots and sticks, whatever works.

I have to teach and this is probably not coherent... !

Gordon:

Interesting! Thanks for clarifying that for me and, once again, for your journal paper and blog discussions in defense of contingent universalism - I really appreciate it!

I must disagree.

This is true: if you flip a coin forever, there is probability 1 that it will turn up heads.

But this is false: if you flip a coin forever, it *will* turn up heads.

To see why, consider the following.

This is true: if you pick a real number between 0 and 1 at random, there is probability 1 that it will not be .8675309.

But this is false: if you pick a real number between 0 and 1 at random, it *will* not be .8675309.

DT,

You say that (3) is true and (4) is false.

3. If you pick a real number between 0 and 1 at random, there is probability 1 that it will not be .8675309.

4. If you pick a real number between 0 and 1 at random, it *will* not be .8675309.

I take it you mean that there is a world in which (3) is true and (4) is false. This would be a world in which someone picked a real number. Let w be such a world. In w all of the following are true:

a. S picked a real number between o and 1,

b. S picked .8675309.

c. The probability is 1 that S did not pick .8675309.

Given (a) and (b), (3) is false in w. Given (a) and (c), (2) is true in w. But (b) and (c) are inconsistent. So, it looks like there are no worlds in which (2) is true and (3) is false. How do you avoid this problem?

Hi Mike,

You say that (b) and (c) are inconsistent. This is exactly what I and many others reject. We admit that it's a cost, but it is one that we have learned to live with.

The trouble is this: what's the alternative? Given that it is possible to choose .8675309, and given that each number is equally probable (by assumption), what is the probability of choosing .8675309? I would say 0. But if (b) and (c) are inconsistent, then we can't say 0. So what are we to say? David Lewis, for one, would say that the probability is:

1 - infinitesimal

where this is not equal to any real number. But the motivation for thinking that there are such bizzare numbers as non-real infinitesimals seems to be nothing more than the intuition that, gosh darnit, zero probability events can't happen! I for one would rather chuck the intuition than go in for those kinds of numbers. (But don't get me wrong: I'm not opposed to weird numbers in general---I'd just like to see them carry their weight before I start using them in arguments for the existence of God.)

Sorry, that should say 'an argument for universalism' not 'an argument for the existence of God'.

dt,

Here's what I think is happening. You don't allow probability functions to assign infinitesimal values, but you are nonetheless countenancing more worlds than there are real numbers. When you assign 0 probability to some proposition p in your example, you mean that there are no real numbers that are properly assigned to p. But you affirm in your example that there are (nonetheless) worlds in which p is true. Which worlds are those? Those are the worlds whose chance of actualization is infinitesimal. So you are really accepting infinitesimals anyway, since, if you don't, there won't be worlds enough the get p true in one.
Your approach also generates some worries that Lewis mentions concerning regularity. I take these seriously: for instance, that we should not assign 0 to any proposition that is true in some world or other. But this is another question, I guess.

"When you assign 0 probability to some proposition p in your example, you mean that there are no real numbers that are properly assigned to p."

I'm afraid I don't understand. I assign it probability 0, and 0 is a real number.

Sorry dt,

I meant no 'positive' real.

You can say both (i) the probablity is 0 that p will occur and (ii) it will happen that p only if you assign 0 to propositions that are not impossible. If (ii) is true, then p has to happen in some world, even though it is assigned probability 0. In your case, this occurs because you assign 0 to any proposition p whose probability is not some positive real. But if p occurs in some world anyway, it must have some infinitesimal chance of being true; some chance infinitely approaching 0. There is no other way to get (i) and (ii) true together. So, while you eschew infinitesimal probabilities, it looks to me like you assume some propositions have them.

"But if p occurs in some world anyway, it must have some infinitesimal chance of being true; some chance infinitely approaching 0."

I do not agree that there must be some infinitesimal chance that p is true. (To insist that just begs the question, doesn't it?)

However, I do agree that the probability of p is

Lim(1/x) as x goes to infinity

But the thing is

Lim(1/x) as x goes to infinity = 0

I do not agree that there must be some infinitesimal chance that p is true. (To insist that just begs the question, doesn't it?)

Not that I can see. You've assumed that there are more worlds than there are positive real numbers. That is, the possible ways in which p can come out true are not exhausted by the positive real numbers. So, two possibilitites:

1. If you insist that there really is no genuine possibility represented by the assigment P(p)= 1/oo, then you should not countenance more worlds than there are positive reals. In that case P(p) = 1/oo = 0.

2. If you insist that there is a genuine possibility represented by P(p) = 1/oo and also that P(p) = 1/oo = 0, then just say that, as you use 'zero probability', it does not entail that if P(p) = 0, then p has no chance of being true. It has a chance that is not representable by a positive real number. In this case, we have a recommendation for interpreting 0 probability assignments.

But what I htink you can't say is that (i) P(p) = 0, (ii) p is true in some worlds and (iii) p has no chance of being true.

If P may occcr and has a probability of 0, then what do we say of the probability of something that cannot occur?

If P may occcr and has a probability of 0, then what do we say of the probability of something that cannot occur?

That's also zero. The assignment of probability 0 is a tricky matter. I say, if you're an average gambler, (and you've got some sense) you assign 0 to anything that will not occur (rather than to just those that cannot occur). Bet the farm on my not pole vaulting 22'. It won't happen, though it could. And you'll be all the richer for it.
But if you're using probabilities as credences, then (many say) you should put some small credence in any proposition p that is not impossible. No doing so amounts to asserting (unreasonably) that you could never learn anything that would raise the probability that p is true.

What of an impossible proposition? I suppose I may be mistaken in believing the proposition is impossible. But if it is, then the probability of it happening is, in fact, impossible 0 minus something or other (on the view we are considering).

The probablity of my building a round square in my basement is less than that I will fly flapping my arms on the way to class today.

bad example! of course the probability of me flying is greater than zero.

but take the numerical case. The probability of randomly picking any given number between 1 and 2 is, DT says, zero. Yet such a number could be picked. But I cannot pick a number between 1 and 2 that is greater than 3.

"If P may occcr and has a probability of 0, then what do we say of the probability of something that cannot occur?"

It depends on what sense of 'probability' and 'cannot' we have in mind. Here's an example. Consider the following proposition:

1. Light travels faster than 186,000 mps.

If modern physics is correct, then (1) cannot be true, in the 'physically possible' sense of 'cannot'. However, I don't assign probability 0 to (1), in the 'credence' sense of 'probability'. (I'm arrogant, but not that arrogant!)

Is there a sense of 'cannot' and a sense of '0 probability' according to which the two are equivalent? I'm not sure there is. It seems to me that under any interpretation these are simply different concepts.

By the way, Mike, here is where Lewis agrees with you:

http://books.google.com/books?id=NprZWnApecIC&pg=PA176&lpg=PA176&dq=david+lewis+probability+zero&source=bl&ots=sCDlOmKaH1&sig=l3jTItal1dgxl3LH_yC_p9t5DCU&hl=en&ei=Nsf5Scj-FIq0NfvXpbwE&sa=X&oi=book_result&ct=result&resnum=1#PPA176,M1

As far as I can see we have two choices: either say that events with 0 probability can occur, or else invent a new number, the 'infinitesimal', and do your math with that. I guess it's a pick your poison kinda situation.

1. If you take the position that dt seems to have taken: any proposition whose probability is not equal to some positive real number has probability 0. In that case, propositions that are possible, but whose probability is (some might say) infinitesimal, get assigned 0. And propositions that are metaphysically impossible also get assigned probability 0.
2. If you take the position that there are infinitesimal probabilities, then the former propositions get assigned positive 1/oo and only the latter get assigned 0.
3. If you take the position that it is impossible that any proposition occurs whose probability is not equal to some positive real number, then all such propositions describe metaphysical impossibilities and get assigned 0.

I'm a little late to the discussion, but I have some questions that the comments do not seem to address. Just to let you all know, I'm fairly new to the world of philosophy. Right now I'm attending a community college for general studies, but I plan to major in philosophy once I transfer to a university. So, please excuse me if I make a fool of myself.

I'm wondering how justness fits into the discussion. Is not God, for example, all-just? If he is, then (4) should be revised to say: (4') It is no cost to an omnipotent, all-just God to sustain R in existence until tF. (Now maybe an all-just God annihilates those who fail to accept the gospel, in which case God does does not sustain R in existence until tF).

If God is all-just, then could it not be that if God sustains R in existence until tF, then at the time tF, God as all-loving and all-just is committed to both continuing to punish R and reconcile R (assuming for the moment that R's choices warrant R's being eternally punished, in which case R's punishment has not terminated at an earlier time t0, and a deliverance of God's all-goodness is his desiring to reconcile all people)? This would be a sort of conflict of properties: God's all-goodness would contradict his all-justness.

Consider the following analogy. Jim lives in a good but just society. One day Jim ends up killing his friend, Joe, in cold blood (for whatever reason). Jim never repents for his crime and would probably do it again if given the chance. According to the law, the government legitimately sentences Jim to life in prison. However, at a later time t (maybe, say, 20 years into his sentence), Jim finally comes to realize the significance of his actions. He understands what he did was wrong. He earnestly wants to repent and change for the better. Maybe he even becomes a Christian. Now, it would seem that if the society is truly good, then Joe's family should forgive Jim for his mistakes (especially if Jesus' commandments to forgive and turn the other cheek are objectively good commandments), and the government should offer him a second chance and incorporate him back into society. On the other hand, if the government is just, then Jim should not be absolved of his punishment (life in prison). So which should the society do? Keep Jim in prison (since justice demands it)? Or offer Jim a second chance (since prima facie goodness demands it)? I have no idea!

I wonder if predicating the property of "justness" to God presupposes other philosophical theories or views, like retributive justice. How might those presuppositions effect the eternal torment-universalism debate?

Three more things...

First, it seems like the view floating around with most evangelical Christians about hell is really inadequate. Hell went from a place of torment or punishment (created and sustained by God) to a mere separation from God dependent on choice. But how exactly is a separation from God dependent on one's ongoing choice really a punishment? Say Jim asks Joe to go to a party. The party is the talk of the town and by everyone's standard will offer a great time. Joe, however, just isn't interested. He chooses not to go. Instead, he stays home and watches TV. Now maybe nothing is on TV, and Joe ends up having a really lousy, and boring time, and he misses out on the great time he could have had at the party. Or maybe Joe genuinely enjoys his time watching TV: there's a documentary on about Christian philosophers debating about hell. In this case, Joe is happy he chose not to go to the party. Now, presumably if we're comparing Joe to someone that finds himself in hell, then Joe in fact had a bad and boring time staying home watching TV. Now the question is, how exactly is Jim choosing not to go to the party even remotely comparable to a "punishment?" It seems like that if people actually choose to stay in hell, then hell is no punishment at all. I also don't see how separation from God is a punishment, since according to many Christians, atheists are "separate" from God (God does not live in their hearts and so on), but many of them seem quite happy. Look at Joe: he might enjoy missing the party to watch TV.

Second, if Molinism is true (and I understand it correctly), could not God have actualized a world such that all men find themselves in situations where they choose to accept the gospel? If God desires all men to saved (which it seems like an all-good being would surely desire), then wouldn't he actualize such a world? Maybe it's possible that there are some people whose essences are such that no matter what possible world they are instantiated in, they still reject the gospel (presumably even after death). Still, God could have simply chosen not to instantiate those individuals (if he knew they would not be able to choose the gospel and God did not desire that). It might be possible, moreover, that a necessary condition of many people choosing to accept the gospel is God instantiating those individuals (who will never choose to accept the gospel), in which case instantiating those individuals would serve a greater good. In that case, would God desire those individuals to end up in hell, or would he not desire it, but he had no other choice? Nevertheless, wouldn't a God who actualizes a world such that all people eventually choose to accept the gospel or believe in God be "better" than a God who couldn't? So if we're merely dealing with possibility, then shouldn't we, given Molinism, prefer the view that it is the case God actualized such a world (and universalism is true)?

I might completely misunderstand Molinism, of course.

Finally, why is the Christian who accepts libertarian free will committed to holding that God desires/wills all people to believe and be saved? Is this desire a necessary deliverance of God's all-goodness? That is, if God did not desire/will all people to believe and be saved, then would that contradict his all-goodness?

My post ended up a little long. Hopefully my points are coherent.

Just stumbled across this short argument and the discussion thread. At the risk of commenting on a post that has been long abandoned by others, I thought I'd mention that the argument presented in this post is a variant of one I developed in my article, “A Guarantee of Universal Salvation?” in Faith and Philosophy vol. 24 (2007), pp. 413-432. I'd be curious to what extent readers think this fuller development of the argument anticipates and avoids the concerns raised here, and to what extent readers of this blog find it vulnerable to critique. I ask in part because the argument plays an important role in the book project I'm working on now (a multi-pronged critique of the doctrine of hell, co-authored with John Kronen over at U of St. Thomas).