Set-Theoretic Misunderstandings
Several considerations are relevant for this section. In my original piece, I pointed out that the claim that there are "a number of people too large to be a set" is strictly speaking, incoherent. This point is neither here nor there for the purposes of the Anthropic Argument. So, you can skip this section if you're not particularly interested in my disagreement with BB on set theory. BB replies as follows:BB replies by bringing up the notion of "proper classes" that are "too large" to be a set, citing Wikipedia. Unfortunately, as a response to what I said, this simply misunderstands the dialectical context. It's true that in set theory, there is something known as 'proper classes', such as the 'class of all sets' or the 'class of all ordinals'. BB's claim, however, was that there is "a number of people too large to be a set". Proper classes, since they are not sets, also aren't actually a 'number of things'[1]. Now, true enough, it is sometimes said that a proper class, is 'too large' to be a set, but it's important to be clear about what that means. What it does not mean, is that there is a size S that denotes the size of the class of all sets, such that S is measurably greater in amount than C, where C is any arbitrary cardinal number. Proper classes don't have cardinalities, in the sense that they do not have measurable sizes with amounts. Rather, what it means is that the axioms of set theory prevent us from treating them as sets, and they cannot participate in the same operations that sets do. This is so as to avoid paradoxes, such as, famously, Russell's paradox. Such paradoxes arise when attempting to treat collections that include all conceivable instances of a type (like all sets or all ordinals) as unified wholes, which would require them to be self-containing. Proper classes aren't 'actual objects' in the universe of set theory[2], such that we can actually collect all their members into a single whole, they are a theoretical construct invented so that we simply can talk about "all cardinals" or "all ordinals" without contradicting ourselves, even though it's not actually possible to form a real measurable collection of "all ordinals" or "all cardinals".The claim that there is a number of things too large to be a set is not at all true. A set has a well-defined formal definition. This isn’t controversial—the Wikipedia page on limitation of size, describes inconsistent multiplicities “that cannot be sets because they are "too large".” That there are classes too big to be sets was axiomatized by Von Neumann—famously no slouch. That’s why classes that are sets are sometimes called small classes—there’s an important sense in which they’re smaller than classes that aren’t sets. In fact, I gave a quite clear explanation of the sense in which it’s too big to be a set—for any set of any size, one can construct a larger set from subcomponents of the collection of all truths.Take, for instance, the class of all sets. There’s an important sense in which it’s bigger than any set (for any set of any cardinality, it contains a set of higher cardinality—namely, the power set of the original set). Similarly, the collection of truths is not a set for the reason described by Grim."
Revisiting our topic of "all possible people," let’s consider BB’s use of proper classes for his claim. When BB states that "all possible people is an amount or number too large to be a set," are we meant to interpret this as "the number of all possible people is a proper class"? If so, the use of "number" or “amount” in this context is problematic. If "number" or "amount" is meant to imply cardinalities as defined within standard set theory, then the claim becomes incoherent—proper classes, by definition, do not have cardinalities[1]. If, alternatively, "number" or “amount” refers to something beyond any set-definable cardinality, then he is using "number" or "amount" in a way that is unfamiliar and, frankly, completely unclear.
Worse yet, describing "all possible people'' as a proper class introduces significant complications. First, this would mean that the total number of individuals encompassed by "all possible people'' is undefined (since proper classes aren't assigned any cardinal number), making any probability calculations concerning this group also undefined. Thus, if "all possible people'' is a proper class, calculations like, say, P(Y|~T) become completely undefined since there isn't any value we could assign to it, which would entail that P(Y|T) is also undefined. Second, it probably implies that "all possible people" exists only as a theoretical construct[2], which suggests that it's incorrect to update in favor of a view that predicts "all possible people'' exist when, in reality, such a group probably could not really exist in any tangible form. For these reasons, I thought it appropriate to 'rule out' that BB was taking "all possible people '' to be a "proper class', but that was a mistake in retrospect and I should have had a paragraph or two on my original blog post addressing that possibility.
BB next says:
In response to this worry about there being no collection of all truths, Truth Teller claims “that there is no such thing as "all truths" on pain of contradiction.” This is not what the Grim proof says—it says there’s no set of all truths but not that there is not such thing as all truths. Obviously, there are all truths—for if each truth is true, then all of them are true, by definition.
"I think the argument will also apply, however, against any class or collection of all truths as well. In the Nous piece the issue of classes was addressed by pointing out intuitive problems and chronic technical limitations that seem to plague formal class theories. But I also think the issue can be broached more directly - I think something like the Cantorian argument can be constructed against any class, collection, or totality of all truths, and that such an argument can be constructed without any explicit use of the notion of membership ..."
Why that is is rather complicated and would take us too far afield, but for those interested Grim gives a rather technical argument for that in this paper. The idea is that the problem doesn't arise simply from the nature of sets, but from the fact that the properties needed to describe all truths, among which would be self-referential properties, cannot be fully captured in a contradiction-free manner. Importantly, this doesn't stop us from saying of each and every true atomic proposition, that they are true, using plural quantification. The problem arises for a collection which is supposed to encompass "all truths". If Grim is right, there is no such collection, and indeed there couldn't be.
BB also addresses my argument that if it were true that "all possible people" is "too large to be a set" then, by parity of reasoning, we can argue that this implies that God can't create all possible people.
I reject 2 because Cantor’s theorem only applies to a set and there are unsetly many world segments. I also reject 3, because once every possible agent is created, you can’t just make new ones to think about things.
His rejection of 3, for instance, would undermine his own argument that there is no set of all minds. In laying out my argument I assume for reductio that there is a state of affairs with all possible minds, and show that this assumption entails a contradiction, in exactly the same way that he allegedly proves that a ‘set of all minds’ entails a contradiction. Here's a parody of his response that undermines his own argument against their being a set of all minds: "once you have a set of all minds then you can't have new ones to think about things, so you can't have new minds thinking about each element of the powerset of the set of all minds". If BB's response to 3 is sound, then he should also think his own argument that there isn't a set of all minds is a failure.
Infinites and the Self Indication Assumption
None of the above is actually all that important to the anthropic argument for theism itself, as BB himself correctly notes. So, let's get to the meat now. The first objection I make to the anthropic argument, so construed by BB, is that updating in favor of a scenario, or hypothesis that entails more observers in light of your existence, is not at all straightforward in the case where the total class of observers is infinite. In fact, it appears that such an inference does not go through. BB starts by claiming the following:Truth Teller suggests that SIA applies in cases where the number of possible people is finite but not infinite. If there are, say, 100^100 possible people then it’s easy to do the math around them—a theory that predicts 100^100 possible people is 100x as likely to contain you as one that predicts 100^99 possible people, for the other theory predicts only 1% of possible people exist. However, if there are infinite possible people then no matter how many people are created, 0% of possible people are created.
But let’s just be clear on what SIA says. It says—and this is what’s supported by all of the arguments I’ve given for SIA (read through any of the arguments here, for instance)—that more people existing is more likely. It says nothing about the percent of possible people existing (in fact, part of the motivation for adopting this formulation of SIA is that it avoids trickiness about counting up percentages of infinity). That’s one way to formulate it, but it’s certainly not required, and does generate trickiness for the reasons TT describes.
This is going to be one of the reoccurring themes in BBs response. BB claims that my objection (that in the case where the universal class of observers under consideration is an infinite cardinality there is no reason to update in favor of hypotheses which posit more observers) is really just an objection to the SIA itself, and that any argument for the SIA would be an argument against that claim. This would be big if true, but unfortunately, it's just false. There are two principles here which BB is implicitly conflating, one is a more modest one, which is the SIA as standardly understood (Nick Bostrom's formulation), another is a much stronger one. Which I will call the "N-SIA"[3]. The standard SIA is as follows:
SIA: All other things equal, an observer should reason as if randomly selected from the set of all possible observers
Basically, you should reason as if you are a member of a universal class of all possible observers each of which are equally likely to exist. This entails that, all else equal, hypotheses or scenario's on which there are more observers are more likely conditional on your existence, because, of the observers that could exist, there is a larger share of those observers that do exist and a higher chance for you to be among the observers that are instantiated. Here is the much stronger claim that BB appears to be endorsing:
N-SIA: Necessarily, if a hypothesis H1 posits n-times as many observers as H2, then, upon updating on your existence, you ought to think H1 is n-times more likely than H2.
Notice that this isn't entailed from the SIA. The SIA has an "all else equal" clause, the N-SIA does not. The SIA just tells you how to reason about yourself as an observer among a set of possible observers (that is by taking yourself to be a random sample), the N-SIA proposes a necessary condition on which hypotheses one ought to think are more likely conditional on your existence. While the SIA does imply that scenarios with more observers are more likely when all else is held equal, this does not entail, contra BB, that it is necessarily the case that, conditional on your existence, you should think hypotheses which entail more observers are more likely.
This is a crucial distinction that I must stress, because only the N-SIA, and not the bare SIA would, if true, undermine my objection that there is no reason to update in favor of hypotheses with more observers where the total class of observers is, say, ℶ2 (an uncountable infinity that represents the power set of the real numbers). The standard SIA is perfectly consistent with my objection, in fact the objection stands on firmer ground given the SIA. The idea is, if we reason as if we are a random selection from the set of all possible observers, then no matter how many observers of the class of possible observers are instantiated, the probability that you are randomly selected to be instantiated as an observer in that set would always be 0, if it is even defined (which isn't clear to me).
I'll stress that the SIA, unlike the much more radical N-SIA, does not imply that having more observers increases the probability of your specific existence if the pool of possible observers under consideration is proportionately larger, it only implies that when all else is equal (and this would include holding equal the size of the pool of possible observers) more observers are more likely given that you exist. In fact, the problem is much worse, the standard SIA actually implies that the N-SIA is false. This is because reasoning as if one is randomly selected from a set of all possible observers would only lead one to update in favor of scenarios with more observers if it means a greater proportion of observers in the total class of possible observers which includes you exist, which isn't the case when the total class is infinite. Let me illustrate this with a simple analogy:
Suppose you have a deck of cards, where you know that within the deck is the ace of spades. There are 2 scenarios, and in each scenario the deck of cards is randomly shuffled.
1. You have a deck of 10 cards, and you know the ace of spades is in that deck. You select 1 card at random.
2. You have a full 52 deck of cards, and you know that the ace of spades is in that deck. You randomly select 4 cards from that deck.
Now, if we treat the cards as observers and reason as the SIA tells us to, you would reason as though the ace of spades is a random sample, equally likely to be any one of the individual cards. So, in scenario (1.) there's a 1 in 10 chance that the card you pull is the ace of spades. In scenario (2.) there's a 1 in 13 chance, that, of the 4 cards you pulled, one of them is the ace of spades. But, wait: That means that the 1 card you pull is more likely to be the ace of spades in scenario (1.), than any of the 4 cards you pull being the ace of spades in scenario (2.), despite the fact that you pull out 4 times as many cards in scenario (2.). But this means, if you reason as if each card is a random sample from the set of all cards in the relevant domain of quantification, it's not the case that, necessarily, having more pulled cards makes it more likely that one of those pulled cards is the ace of spades. What matters is the relative proportion of pulled cards to the total set of cards being pulled from which contains the ace of spades.
Let's now apply this analogy directly to an example scenario involving observers: suppose there are 2 random-observer-making machines, M1 and M2. M1 is programmed to select among a class of 500 possible observers, among which you are one of them, and to select 100 of those observers to create at random. M2 has a much larger class of possible observers it can create, 1 trillion let's say, among which you are one of them. M2 selects 200 million of those observers to create at random. Only one of the machines performs their function. You learn that you exist and were created by one of the machines. Which machine should you think performed its function?
The probability that you exist given that M1 performs its function and creates 100 observers, is 1 in 5 or 20%. The probability that you exist given that M2 performs its function and creates 200 million observers, is 1 in 5000, or 0.02%. So, it's 1000 times more likely, given that you exist, that M1 performed its function than that M2 did. You should therefore think you were created by M1, even though the fact that you were created by M1 implies that much less observers are created than if you were created by M2. This is completely consistent with the SIA, in fact if anything it is probably implied by how the SIA tells you to reason about your existence. The SIA says to treat yourself as a random sample in the set of all possible observers, not to treat the probability of your existence as being necessarily higher in relation to any arbitrary scenario where you happen to exist alongside more observers. Treating yourself as a random selection from the pool of possible observers, and indeed, simply appealing to standard ways of assigning epistemic probabilities in situations of ignorance such as the principle of indifference implies that there are possible albeit abstract hypotheticals, like this one, where you have more epistemic reason to update in favor of scenarios which entail less observers in light of your existence.
BB also claims that the motivations he has for the SIA motivate the claim "that more people existing is more likely" regardless of the percentage of possible people instantiated. This is false, the motivations he provides in the linked article are motivations for the regular SIA, many from Joe Carlsmith. But the SIA, as I've argued, actually implies that the percentage of people instantiated does matter. I don't want to waste too much space so I'll just address one scenario BB provides in his article where he cites Carlsmith:
I could go on addressing other motivations BB offers for the SIA. But the point is, they are motivations for, well, the SIA over rival views like the SSA (which I'm by no means committed to). If BB thinks he has motivations specifically for the much stronger N-SIA that does undermine my argument, it's his burden to provide them, a burden he has completely fallen short of meeting. The motivations present in the article linked, so far as I can tell, if accepted, don't give us any reason whatsoever to accept the N-SIA in favor of the more modest SIA. Even if they did, we should still reject the N-SIA, since as we've seen, it entails incorrect reasoning about epistemic probabilities (and we'll see later in the last section that the N-SIA can be proven false another way).
In sum, BB is wrong that my argument against updating in favor of more observers in infinite contexts depends on rejecting the SIA or the motivations BB has offered for it, they only depend on rejecting the N-SIA. However, not only has BB given no good reason to accept the N-SIA so far as I can tell, the N-SIA is plainly false, and so it's no cost that my argument depends on its falsity. I’ll also note that even if my argument did entail the falsity of the bare SIA, BB would still need to engage with my argument directly, and it’s not at all sufficient to point to his independent motivations for the SIA. Even if the SIA does better than any rival views in anthropics, such as the SSA, it may well be the best of a bad lot.
BB tries to directly address my objection with an analogy:
BB claims that if you're in Hilbert's Infinite Hotel and you learn that your room is painted red, upon updating you should think it is much more likely that googolplex rooms are red than one. But he doesn't show how this update actually bears out probabilistically, he just baldly asserts it as if it were an obvious truism. Contra BB that's not at all obvious, and the only reason to think it's obvious, so far as I can tell, would be if one inappropriately extends one's intuitions about probabilities regarding finite sets to infinite sets. In fact I contend that it is false, both hypotheses have a conditional probability of 0 given that you see yourself in one of finite red rooms and there are infinitely many non-red rooms each of which you had an equal chance of being in, since, both googolplex and one are 0% of even the smallest transfinite cardinal.
Bizarrely BB seems to concede this, but yet he still says one should still think the hypothesis that a googolplex rooms are painted red is more probable. But that's incoherent, if they both have a probability of 0, then, well, they both have the same probability, 0. Charitably, maybe he's just saying you should still think googolplex rooms being painted red is more likely, even though it's false that it's actually more likely. But that looks odd, surely on any plausible account of epistemic rationality you shouldn't believe false things. More charitably, maybe he's saying that we should use non-standard analysis so we can say a googolplex within infinity is a larger infinitesimal than one within infinity. But even if that is granted, how this would bear out probabilistically is at most highly unclear and BB has not even attempted to demonstrate that this would permit a substantial Bayesian update in favor of more red rooms. Overall, at the very least BB's claim demands serious clarification and motivation. Recall that precisely what's in dispute is whether Bayesian updating in line with the SIA works in infinite cases in a way that allows the anthropic argument for theism to succeed. Simply insisting that it does work that way is dialectically inert.
Next, BB addresses my pre-emptive rebuttal to the claim that God would create all possible people not just the same cardinality. He says:
BB says that creating a happy person is a valuable action, and that doesn't change just because you already have infinitely other valuable people. However, I'm not at all committed to denying that. In fact, it can be granted that for any happy person that is created, there is value in creating them, regardless of how many others are already created. The problem comes when we think about God's decision regarding which set of ℶ2 people to actualize. Let's say you have S1 which contains all possible people, and S2 which contains the same amount of possible people as S1, but we take out some arbitrary number of specific elements, in this case possible people, including you and 'Suzie'. We know the act of creating a happy person for the sake of their good has intrinsic value. We can label the intrinsic value of creating a person with a good life, V. We can then have a function which maps every person in S1 or S2 with a distinct Vi, where Vi is the intrinsic value of actualizing that particular person.
Here's the kicker. Since both sets have the same cardinality that means that all elements from S1 can be put in one-to-one correspondence with all elements of S2. In other words, for every created happy person in S1, there's a corresponding created happy person in S2. This also means, for every Vi that corresponds to an element in S1, there's a Vi that corresponds to an element in S2. So, the intrinsic value of creating the people in sets S1 and S2 is equal, since every act of creating each person in S1 has a corresponding act of the same intrinsic value, of creating a person in S2. So, the real issue isn't whether there's a reason to create Suzie when you already have infinite people; it’s whether, prior to creation, God has any reason to choose to create an infinite set that includes Suzie over an infinite set that doesn’t, if both sets have the same cardinality. Each person in the set without Suzie can be mapped with a person in the set with Suzie, and each act of creating those people would confer the same intrinsic value, V. Thus, BB’s claim that God would create all possible people, even if this doesn’t imply creating a greater cardinality of people is unmotivated, the reasons he tries to give are uncompelling.
BB also uses an example of getting stabbed twice versus once for an infinite amount of days. But again, since our ways of aggregating and comparing quantities simply don't work the same way for infinites, it's unclear why we should put much stock in BB's intuition. But assuming we do hold that a person getting stabbed twice a day for infinite days is worse than them getting stabbed once for infinite days, this could not be because the total amount of suffering endured is greater (they're the same). It would have to be because for each individual day, the suffering the person undergoes is more intense and painful. But then how this analogy is supposed to apply to God's decision to actualize S1 over S2 is unclear. Each member of S1 can be mapped to a member of S2, where the intrinsic value, V, of creating each person and giving them a happy life is present equally in both cases.
Next BB says:
I'll briefly note that I never appealed to the existence of infinite people 'sounding weird'. But insofar as BB wants to make a distinction between positions "sounding weird" and "seeming actually false," it's unclear why any of the relevant intuitions BB wants to appeal to about infinites don't themselves just fall into the "sounding weird" category[4]. Especially since it looks like they can be explained by infinites having weird properties that don't neatly conform to BB's intuitions. If BB wants to dismiss intuitions that count the other way as merely "sounding weird" without giving a principled criteria for what intuitions actually "count", then there's nothing to stop me from playing that game too.
He continues:
Second, BB misconstrues my point in that section, though perhaps I could have been more clear. It wasn't that being unable to list the real numbers prevents us from having a bijection on a set as large as ℶ1 or ℶ2. Nor was it simply some argument from ignorance from our lack of knowledge of a way to map concrete people in the world to a higher-order infinite set. It's that there is a positive intuition that it is impossible to do such a mapping. For any amount of actual people, even if there were infinitely many, it seems we can conceptualize them as first, second, third, and so on, mapped to 1, 2, 3, etc., but we'd never at any point map people to irrational numbers like π or complex subsets of real numbers. There could never at any point be a 1.78657089….th person, or so it appears. It seems like a set of actual concrete people would have to be discrete, not continuous like the set of real numbers or bigger sets like ℶ2. People seem to be distinct countable entities that can, at least in principle, be listed and individually identified with no intermediate values between them. But higher-order infinite cardinalities are not listable, between any two real numbers there are infinitely other real numbers. It’s perfectly fine and sensible to think about such sets abstractly, but how could we even begin to map concrete actual people in reality to a set with such properties? I'll stress that this doesn't deductively prove that it is conceptually or logically impossible for there to be ℶ2 people. It simply illustrates that there is something counterintuitive about it. On reflection, myself and likely many others will find the notion of ℶ2 people existing far more counterintuitive than the intuitive appeals in infinite cases that BB likes to rely on.
Next BB replies to my objection that while all-else-equal the SIA will lead us to favor hypotheses with more observers, the 'total evidence' does not favor hypotheses with more observers. While more observers make your existence more likely, they also make you less likely to be you and not someone else, as there are more possible identities to assume. BB starts with the following;
The SIA entails that all else equal taking into account the data of your existence leads to an update in favor of scenarios with more observers, it does not entail that you can’t have other evidence which leads to an update in favor of less observers. So, on its face, it's simply not true that any argument for the SIA would undercut the TE objection. The purported absurd results BB provides are ones which afflict alternatives to the SIA like the SSA, which I'm not committed to adopting. If BB thinks he has a rebutting or undercutting defeater for the TE objection, it is his job to provide it. However, he provides no such defeater, so as a result his claim is simply unmotivated.
BB however does attempt a direct attack on the TE objection:
If, following the analogy, you discover you're in room 1, this new evidence cancels out the initial update in favor of tails, since the scenario of being created in room 2 is ruled out and you’re only left with two possible outcomes. Thus, if BB’s analogy allows for a second update upon gaining new information about your location, it should similarly allow for a second update when taking into account evidence of your specific identity as an observer among other observers, including your unique body, experiences and properties—features that differentiate you from other identities that exist. As I argued in my original piece, this update would counterbalance the update in favor of hypotheses with more observers when taking into account your existence. So, even granting the SIA, the total evidence doesn't favor a scenario with more observers all-things-considered. Nothing BB says here remotely undermines this point.
With respect to his second point, the idea that "even if such and such has a low probability, you should update in light of your existence, on theism plus such and such". This is another reoccurring theme in BB's reply, which I'll say a lot about later in this section. I'll make a couple points for now. First, I'll note that the claim that your existence significantly boosts the probability of theism does not merely hinge on pro-natalism, it hinges on a more specific interpretation of divine motives. As I noted in my original piece, it assumes not just that God would make people, but that God would create indiscriminately—that it is always all-things-considered better than not to create a possible person with a happy life. But even if creating happy people is generally a good thing, it does not follow that it is all-things-considered good to create more and more possible people with happy lives without any restriction. Many possible theistic narratives, for instance, allow for a God who creates selectively, according to divine wisdom or a larger cosmic plan, prioritizing other goods over sheer numbers of happy individuals.
Second, BB assumes, without any justification, that it is methodologically sound to update our beliefs about ethical views like pro-natalism based on their utility in explaining the data when combined with theism. But I am not at all inclined to grant him that much. I take it ethical views are the kinds of things that must be independently justified, by virtue of their accordance with our moral reasoning and practices. They aren't themselves variables to be adjusted at our behest based on what best accommodates the already observed data when combined with whatever metaphysical postulate we're using to explain that data. Relatedly, I suspect there may be a looming circularity worry here. Why do we exist? God creates us. Why think God would create us? Because such-and-such ethical assumption. But why think such-and-such ethical assumption is true? Because it explains why God would create us and accounts for the improbability of our existence. Of course, this wouldn't be an issue if one holds to the ethical assumption on independent grounds, but insofar as one updates in favor of the ethical assumption for the sole purpose of explaining some otherwise improbable data regardless of it's independent plausibility, that looks to me like a vicious explanation. You don't use the same data as evidence for a theory, and as a basis to tailor the theory's assumptions. You can't use the data twice, as that would be double-counting, and as we will now see, BB acknowledges the folly in double-counting:
Nevertheless, if I am being charitable, what BB means to say is the following: when considering the anthropic argument, we must hold certain data, k, in this case the existence of a universe that God created in our background knowledge. We then use k in conjunction with theism to predict the new data. Of course, this would not at all vindicate his "double-counting" charge, since failing to hold certain data in the background when evaluating what to expect from the theistic hypothesis, isn't anything close to double-counting. In any case, I concede that this move can be made in my original post, which is precisely why I introduced the duplication/godman objection. My argument concerning the imperfection of creation was to highlight that if we evaluate theism purely on antecedent grounds, without any presupposed knowledge of creation, we wouldn’t expect God to create at all, much less expect our specific existences. I never suggested that this precludes a theistic update if we assume God's act of creation as a given. Again, that is precisely what my godman/duplication objection aims to undermine.
When it comes to his accusation of double-counting in the context of the duplication objection we run into trouble, and it is hard to see how BB’s accusation could be charitably interpreted. That objection is directly about the probability that God would make a person like you, it is not something that can be held in the background when assessing the anthropic argument. True enough, it is a byproduct of my critique that the world we observe, with very unimpressive beings and blemished creation is very, very unlikely under theism, and thus that theism is probably false (at least if we only consider that specific evidence). But that's just what virtually ANY critique of the probability judgment that 'P(Y|T) is not hopelessly low' or 'P(Y|T) is higher than P(Y|N)' would entail. It certainly does not mean I am double-counting. The dialectic is simply that BB put forth an argument for theism, and I responded by directly challenging one of the premises of that argument—the assertion that 'P(Y|T) >> P(Y|N)'. We would only massively update in favor of theism if that assertion is true. This isn't at all illicit; in fact such antecedent analysis of the relative conditional probabilities is integral to how Bayesian reasoning operates.
BB's resistance to this line of critique, with his dogged insistence that we would just reject my objections conditional on the truth of theism, if taken seriously, would cripple pretty much any analysis of any Bayesian argument for or against theism, including his own. For instance, BB defends the anthropic argument, or the argument from psychophysical harmony. In laying it out, he makes some argument that, on naturalism, the data in question is very improbable. Now imagine a naturalist retorting, “Sorry BB. If your argument that the data is extremely unlikely under naturalism is correct, then naturalism would be, very probably, false. Or at least you’d have to be assuming a version of naturalism that is very probably false, since we very probably wouldn’t have psychophysical harmony. Therefore, under naturalism, we must reject your argument.” BB would, I think, immediately see the absurdity in such a response. Yet, his dismissal of my objections is parallel and mirrors this very absurdity.
Perhaps BB meant to suggest that the godman objection is double-counting because we should hold theistic answers to the problem of evil in the background as data. But, again, not only is this not double-counting for reasons I already made explicit, I dedicate an entire section to addressing that sort of objection in my original piece. I reject that there are any good theodicies, so I reject the demand to hold a successful theodicy in the background. Even if such a demand is accepted, I also reject, and have argued against in my original post the claim that holding theodicies in the background saves the anthropic argument from my objections. In other words, even if we update to a version of theism on which God has reasons for allowing evil,[6] God creating us is not explained. BB does not address this key point.
Perhaps BB thinks there is other data we can hold in the background that permits theism to predict the data of our specific existences without running afoul of my objections. But keep in mind such data would also be held in the background for naturalism. So if the background data is too specific it may also allow naturalism to predict your existence with a high probability. If, on the other hand, it's too general, then chances are it won't be enough to help theism predict our specific existences in light of my objections. It's BB's burden to identify data that, when held in the background, leads to a significant update in favor of theism over naturalism. A burden that I contend BB has not remotely met.
So I must conclude that BB's above response is a failure, the accusation even in it's most charitable interpretation simply does not land. As we've already seen there's another reoccurring theme in BB's reply, which goes something like this "even if X is really implausible, so long as X has a 5% or 10% probability, you should still massively update in favor of theism, with the version where X is true". We've seen that he says it in response to the concern that BB must assume certain ethical views. He also says it in response to my objection that God wouldn't create:
A consistent problem with BB's arguments, not just here but elsewhere, is that he will assign an arbitrary credence, or evidential weight to a proposition P, and say "surely it is reasonable to think P is n% probable, right?" But he doesn't actually justify why such a probability assignment holds beyond brute seemings. This is a bad methodology. First off, and I'll get into this more much later, but this is just not how credences of this sort work. I take it credences are determined relative to a background web of beliefs and many complicated relations between those beliefs, you don't assign an arbitrary credence to some proposition (e.g "God would create all possible people") based on how plausible it strikes you in a vacuum. More importantly, in standard ways of testing a hypothesis with empirical data, you start with a distribution space and a null parameter for that space[7]; you don't assign an arbitrary credence to a random possible outcome of your model based on subjective intuition. If you did, the argument would basically just amount to: "I have a subjective intuition that your existence is astronomically unlikely unless God exists. Therefore, your existence is astronomically unlikely unless God exists." But that's a terrible argument. It would be ineffective to anyone who doesn’t share your subjective intuition, could in-principle be used for any crazy hypothesis you might conjure up, and wouldn't hold up in any serious statistical analysis or model selection.
Leaving aside that the probability BB assigns to God creating you is wholly unjustified. I'll grant that for some it may seem plausible, in a vacuum, that assigning a minimum 5% credence to God creating you is reasonable. But recall that the existence of you in particular, or so BB himself would claim, has a vanishingly small a priori probability. The idea, I take it, is that there are enormously many ways the world might contain no observers, and even more ways (higher-order infinitely many) for it to not contain you, a specific observer. So the scenario where you exist is astronomically unlikely. But this means, holding that God has a 20%, or 5%, or even a 0.0000001% chance of making a scenario with you specifically, would be to hold that, the truth of theism would significantly raise the expected probability of you existing by a factor of magnitudes of infinity. It would entail that God is much more favorable to actualizing scenarios where you in particular exist, relative to the infinite upon infinite possible scenarios where you don't exist.
This is not the modest claim BB imagines it to be. In fact, it is the very point in dispute. It is a claim that demands serious motivation, which BB, in levying this objection, has not even attempted to deliver. Indeed, my 'godman' objection, if successful, (which I unsurprisingly think it is) shows that for any act of creation God performs, it's very unlikely that God would create a creature like you, since He's much more likely to create a more impressive being, or a duplicate of one, which is a positive reason to think God would, in fact, not greatly favor actualizing scenario's with beings like you, much less you in particular, relative to infinitely other possible valuable states of affairs that do not include you specifically.
Ultimately, I suspect BB’s response amounts to little more than a rhetorical sleight of hand—a tactic equally available to naturalists. Imagine a naturalist claiming, "even if there’s only a .01% chance that there’s a multiverse where all physically possible outcomes (including that you exist) are realized, or that there's a naturalistic disposition weighted towards leading to your existence, you should massively update in favor of either the naturalistic disposition or the naturalistic multiverse that leads to you existing." BB would agree that this just abuses how credences work. There’s no good reason to believe, a priori, that the version of naturalism favoring your specific existence is any more probable than 1/n, where n represents the countless versions of naturalism that do not predict you specifically. Yet, again, the move BB makes here is parallel and equally unjustified. I've given forceful reasons to think that God is antecedently very unlikely to create beings like us, and this is on top of the fact that God creating a world with us specifically over any of infinitely other possible worlds is already intrinsically very unlikely given the sheer range of possible options available to God and the a priori improbability of our specific existences. My broader dialectical point is that there isn't a convincing reason, to think that the antecedent probability of God creating us in particular is over 1/n* where n* is any of infinitely many possible worlds without us God might have moral reason to instead create, or possible sets of people of equivalent cardinality that God might have reason to instead create. The one reason BB has given for thinking God would be specifically interested in creating us (because it's good to create a person with a happy life) actually just fails to justify the claim that God would specifically make us, in the face of all the objections I've raised in my initial piece.
To make what is essentially the same point another way, if our goal is to explain the specific data that we exist and if this data is vanishingly unlikely a priori, then we will need to update all hypotheses, not just theism, to the expanded versions of those hypotheses that do indeed explain the data. In theism's case, this means updating from bare theism to an extended 'theism+' that makes it likely that we will exist. What BB fails to establish—and what is crucial—is the claim that the ratio of the 'theism+' sub-hypothesis that makes our existence likely, to the total space of bare theism is significantly greater than the proportion of extended non-theistic sub-hypotheses that make our existence likely within the space of non-theism. This would include not only versions of naturalism such as the multiverse or stalking-horse naturalism, but various non-theistic spooky or supernatural explanations such as a limited non-divine creator, or a cosmic value-directed telos etc. And, again as I pointed out, the advocates of these hypotheses can make exactly the same rhetorical move BB makes: surely there’s at least a 1% chance that cosmic teleology would be directed to create us given background moral theories?
As a final note, this is all assuming an explanation of the data cries out here. Why insist that the demand for an 'ultimate explanation' of one's specific existence that goes beyond our immediate experience and observational data, let alone a theistic explanation, is even legitimate in the first place? I myself find such a demand both unnecessary and unconvincing[8]. Now, it might be said that we're just being Bayesians here and updating in light of the evidence. But, once again, using Bayes theorem and some funny math, one can trivially support virtually any hypothesis in light of some purported extremely improbable data, provided one adjusts their priors and likelihoods creatively enough. For ex. the probability that my foot lands in this very specific place, presumably described by some set of real numbers, rather than any other place is basically 0 unless an invisible fairy with a desire to put my foot there, guided it to that location, so I update in favor of the hypothesis that an invisible fairy guided my foot.
Further, there are burden-shifting grounds that cut against the rejection of the logical possibility of R-goblins given theism. To start, the free will defense to the logical problem of evil as put forth by Alvin Plantinga, entails that, possibly, creaturely essences are transworld depraved. Further, God is orthodoxly conceived as a person that necessarily has a perfectly good nature. But if it's possible for a person to necessarily have a perfectly good nature, then, unless a relevant asymmetry is provided, we have solid grounds to think a person with a necessarily evil nature is also possible. Both are 'cross-world essences'.
There's a mention that soul theory and most views of personal identity deny that there are intrinsically evil natures. He doesn't explain why that is, but even if it's granted, it's irrelevant, as all that matters for my point is that creatures with such natures are logically possible.
Lastly, we can modify the concept of R-goblins in such a way that they are only contingently morally evil and have horrifically barbaric dispositions. It still looks pretty bad to create them, I think most people would agree with me here. So it's still a counterexample to the general thesis that it's always good to create a possible person that has a good life.
He goes on:
BB also says:
BB attacks my argument that God would not create, since the world sans creation is perfect. Saying:
Put another way: God is said to be a maximally good being, which means that He is the most excellent thing that could possibly exist[11]. But then if you say that it would be better if lesser goods exist, then you'd be saying that God is not the most excellent thing that could exist, God + lesser goods is. It would be to say that a better being than God prior to creation, would be a being that entails lesser goods, or has the relational property of existing alongside lesser good beings. But at least on my concept, the most perfect being would be a being that, alone, is the best thing that could exist, not dependent on lesser goods.
As for the architect analogy. I don't know if I even understand it. I guess the idea is that you can have a building with one perfectly good room, and 4 other bad rooms and so it wouldn't be a perfect building. But I agree with that. A perfect building would be a building for which all its rooms are perfect. There is a distinction between the perfection of a room and the perfection of the building as a whole. Where this analogy falls short is that we are talking about a world which contains a maximally good being and nothing else and asking if it can be improved. So, there's nothing else that exists to make the world less than perfect. The building has shitty rooms that make it less than perfect. Further, we aren't talking about a perfect sortal object like a building, or an island, we are talking about a perfect being, the most perfect thing that in general could exist, not the most perfect of a certain kind of object.
BB uses an analogy, from John Buck to motivate the claim that God would create:
1. If the "anti-God" would create people to torture, then God would create people and give them good lives.
2. The "anti-God" would create people to torture.
3. Therefore God would create people and give them good lives.
This argument is a valid instance of modus ponens, so the conclusion can be denied only on pain of denying one of the premises. What should we make of this? The problem is, depending on how we understand the concept of the "anti-God", either 1 is false, or 2 is false. But no understanding renders both premises true.
There are two ways we can understand the "anti-God". One way is that they are a being that exemplifies the maximal amount of axiological badness possible. They are a maximally bad being. In this case, 1 looks like it's at least on firm grounds, though it can be denied if one accepts that there is an axiological asymmetry. But suppose we grant that there is no axiological asymmetry. If a maximally bad being would create more intrinsically bad stuff, then absent any relevant asymmetry, for the same reasons, a maximally good being would plausibly create more intrinsically good stuff.
The problem is then with premise 2. If we take seriously the concept of a maximally bad being, then they would instantiate not only a maximally intrinsically evil character, but all possible evils to the worst degree imaginable, including the most horrific and maximized agony and despair possible. But that means creating more beings to suffer won't add to the world's disvalue where the anti-God already exists, since, by definition, the worst possible, and maximized despair, agony and general badness is already instantiated and so cannot be increased by the addition of more suffering beings. In fact, making conscious beings to horrifically torture would add more value to the world, since those beings would probably have some intrinsic moral value, even if their lives are as shitty as can be. So it would increase the world's value but not increase its badness, and an anti-God wouldn't do that. I concede that it seems very intuitive prima facie that a maximally bad being would create people and torture them, but that's because the notion of a maximally bad being is a truly bizarre one that is hard to conceptualize. Once we take seriously what the existence of a maximally bad being actually entails, the intuition dissipates.
What if we just take the "anti-God" to just be an all-powerful and maximally morally evil being that doesn't necessarily instantiate maximal axiological badness. In that case, 2 looks pretty plausible, such a being would create people and maximize their despair and agony. But 1 is false, because the anti-God and God are not analogous. The anti-God can make the world worse by torturing people, God can't improve the world because He already instantiates absolute perfection and maximal goodness of being. So, all and all, the argument is interesting and challenging but, I think, ultimately unsound.
I'll also note that even if the argument is sound, while it would undermine my arguments against divine creation, the analogy would actually provide motivation to my godman/duplication objection. A maximally bad or evil being would plausibly only create beings that instantiate the worst horrors imaginable and duplicates of them. So too, a maximally good being would plausibly only create beings that instantiate the best and highest person-centric goods, and duplicates of them.
I mention in my original piece that a popular theistic meta-ethic is that goodness is constituted by resemblance to God, and that even if you don't accept that view, it's a commitment of perfect being theism that things are good to the extent that they resemble God even if goodness is not identical to resemblance to God’s nature. BB says:
I also can't make sense of the latter point. Do I resemble myself less than the state of affairs containing myself and some other guys that resemble me? No, that makes no sense, I am myself, so nothing could resemble me better than myself. So too, the state of affairs containing God and some stuff that resembles God doesn't resemble God more, which means it isn't more good since what goodness is on the meta-ethical view outlined, is constituted by resemblance to God.
In any case, I gave an argument in my post to justify the claim that the weaker view is a commitment of perfect being theism, to which BB says the following;
I also don't understand how the second sentence flows from the first in a way that's relevant.
BB misapplies the double-counting charge again in response to my objection to the claim that God would create people for the sake of their good, where I note that it would be made false if there was an axiological asymmetry. I've already addressed that charge and won't repeat myself further. I'll also note that I do more than simply cite the axiological asymmetry as a response to the relevant objection, and the axiological asymmetry wasn't the most crucial point there.
He next responds to my second argument that God would not create, if creation involves any blemishes or imperfections. Saying, of the condition I provide:
In any case, the most important thing to note here is that he misunderstands the condition I laid out. It doesn't just say He'd only create evils or blemishes if they serve a greater good. It says He'd only create evils or blemishes if, for every greater good that entails those evils or blemishes, it is not the case that there's an even greater good which does not entail them. For reference here's the condition:
A morally perfect person M will actualize some state of affairs S that contains or entails some moral imperfection or evil E only if S is a greater good and it's not the case that there is a possible state of affairs S* which is a yet greater good M could choose to actualize instead of S, and S* does not contain or entail E.
The reason this matters is because, as a result of him not understanding what he has accepted, he gives a response that misses the point:
We can grant that imperfections or blemishes serve greater goods. But that's not enough to satisfy the condition. The conjunction of those evils and blemishes and free will or whatever good BB wants to appeal to may well be an overall greater good. But if there is an even greater good that does not entail those evils or blemishes, then, on the condition I've given, God wouldn't create those evils and blemishes. The problem is, in fact, there is a yet greater good that does not entail them, as I explain in my piece:
Now, this condition entails that God would not create morally imperfect or sinful creatures. This is because even if the creation of morally imperfect creatures is an overall good, it still entails moral blemishes or evils E, and per our condition, a morally perfect being would only permit E if E entails a greater good, and there is no even greater good that does not entail E. However, we can see that for any overall good that entails E, there is a greater good which does not entail E. God Himself is such a greater good, indeed, He is the greatest possible good, and God does not entail E, or any evils.
So BB misses the point. He could of course reject the condition, but then he has to address the motivations I've given for it, which he didn't do. He also says this:
Finally, he addresses my argument that God would not create from divine impassibility:
This is, however, an objection I pre-emptively address in my original post, and he doesn't address or even acknowledge my response, which is as follows:
Perhaps the idea is that while there are bad things about humans, since God can't create the best beings anyways, it's good enough for Him to create humans. But this is untenable. For one, it's morally absurd. Imagine I kill someone and justify it by saying "Sure, killing that guy was not the morally best thing I could do, but it's not the worst either and it's good enough. I can't do the morally best thing anyways". For two, it robs theism of empirical content. For any state of affairs we observe, no matter how bad, we can always say that God would create it because He can't do the best thing anyways and it might have been worse. But then theism wouldn't rule out any observable state of affairs, it would be equally consistent with any observation we could have, and would thereby be void of empirical content. That's bad because it would mean we can't test theism with any observable data, so it would be off the table as an explanation for the anthropic data. So, there clearly has to be some constraints on what God would create. The question is whether humans are within them, and I've argued they aren't.
The second problem is that it's false. The kinds of things that fix the goodness of creatures do have an intrinsic maxima. If God creates beings that are, like Him, perfectly good, then He can't make better ones. Indeed, Mark Walker (2009) in his paper on the 'anthropic argument against theism' argues that God would make ontological equivalents to Himself. A claim this strong is also not necessary, since the whole point of my friend’s more modest take on Walker’s argument was to suggest that there are certain valuable states that have a deepest qualitative level or finite set of aspects that can be instantiated, and that we fail to meet these criteria. Indeed, a large part of his post is intended precisely to address the ‘no best option’ objection BB is raising (and he told me that this is made even more explicit in the new version of that post that is pending). As a result, BB’s objection simply does not apply to a construal of godmen consistent with the very arguments I cited in my original post.
The last problem is that even if it's true, it doesn't help, in fact it makes things worse. Suppose it's true that for every being God could create, there's a better one. In that case, we can argue that God does not exist as follows[13].
1. For every possible being B, there's a morally and axiologically better possible being B*.
2. If God exists then God is morally perfect and omnipotent
3. If God is morally perfect, He would have the most impressive moral resume possible.
4. If a moral resume R1 is otherwise equivalent to another moral resume R2, but R1 includes an act of creating an axiologically and morally better being than a corresponding act of creation in R2, then R1 is a more impressive moral resume than R2.
5. If 1, 2, 3, and 4 then if God exist then for any possible being B, God would not actualize B (Since there’s a better being B* He could have actualized instead)
6. Therefore, if God exists then for any possible being B, God would not actualize B. (from 1-5)
7. If God exists, then it's not the case that there exists a being that God would actualize (from 6)
8. If 7, then if God exists, it's not the case that beings other than God exist
9. However, beings other than God do exist.
10. Therefore, God does not exist. (from 7, 8 and 9)
1 is the assumption for reductio, 2 is definitional, 3 is definitional, 5, just express a conceptual entailment from the previous 4 premises, 8 follows from God being the sole entity prior to creation, and 9 is uncontroversially true. That leaves 4, but 4 is (and I know that BB loves his intuitions) intuitively obvious! It's also a plausible conceptual truth, what it means for a moral resume to be more impressive than another moral resume is that the set of actions it involves are better all-things-considered. So the reasoning looks pretty water-tight given we accept 1, by my lights.
So that objection fails. This leaves us 0 for 4, and with one more objection to examine. It's another one I pre-empt and address, the objection being that God creates humans in addition to 'godmen'. But this time BB actually does address my responses. So let's see, he addresses my point that it's not at all obvious that God would create a world that maximizes happy people instead of a world which instantiates the best kinds of goods:
I'll also note that the appeal to the pareto principle in cases where the people don't exist yet is dubious. The pareto principle typically applies to changes affecting existing individuals, where an action is considered good if it makes at least one existing person better off without making anyone else worse off. To-be-created people aren't current stakeholders whose well-being we can measure or compare.
BB is also wrong that my objection is an appeal to anti-natalism. The options aren't 1) creating possible happy people is always all-things-considered good or 2) creating people is always all-things-considered bad. There's actually a third option, 3) creating possible happy people is sometimes all-things-considered good, and sometimes not. That's what I'm invoking.
BB challenges my argument that godmen can be duplicated without end, and so for any person God creates, He could create a duplicate instead, which He has better reason to do. His first objection was one I already refuted. But he has two more. Here’s his second objection:
In any case, BB does attempt to prove that the duplication is impossible:
Most of what BB has to say related to stalking horse naturalism is defending the intrinsic probability of perfect being theism. But I want to stress that even if one thinks his defense is convincing, he still hasn't fully engaged with my broader dialectical point. I'll note right away, that if it were the case that it were plausible that perfect being theism predicts our specific existence, without any further assumptions or built-in intentions, then I'm happy to admit that the stalking-horse objection would lose much of it's dialectical bite (which isn't to say I wouldn't still think it's cogent, just that it would be more of a dialectical uphill battle). However, as argued previously, that's not plausible.
I make the point that theism is as mysterious as Nd, arguing that, insofar as a naturalistic disposition weighted towards your existence is mysterious, so too is it mysterious for God to be motivated by certain moral reasons that makes your particular existence likely. BB replies:
Still on the topic of the mysteriousness of theism, I also point out that theism is just as uninformative with respect to the causal mechanism at play. BB replies:
So this leads us into our discussion on the priors. BB writes:
Call a U-perfection a perfection of unicorns. Here's an argument for a necessarily existing unicorn. Here are our axioms.
A1 A property is a U-perfection only if its negation is not a U-perfection.
A2 U-perfections entail only U-perfections.
A3 The property of being a necessarily existent unicorn is a U-perfection.
Now, here’s the argument:
1. Suppose being a necessarily existent unicorn is impossible
2. Since being a necessarily existent unicorn is impossible it entails not-being a necessarily existent unicorn
3. From A3, A2, and 2, the property of not-being a necessarily existent unicorn is a U-perfection
4. But this contradicts A1, if a property is a U-perfection it's negation cannot be a U-perfection
5. Therefore being a necessarily existent unicorn is not impossible
6. Being a necessarily existent unicorn is either impossible or actual
7. From 5 and 6, there is something that is a necessarily existent unicorn.
So, the GOA and its modern variants pull no weight here. It doesn't give any convincing reason, by my lights, to think our credence in theism should be anything more than 0 or arbitrarily close to 0. Instead we can focus on the contingency argument. In its strongest variants, I think the contingency argument is much more forceful. Do I think it's compelling? Not really. I can't run through all the considerations here, as the topic is dense and spawned a rather vast and technical literature. The first problem is with the PSR. Unrestricted or strong PSRs tend to lead to modal collapse, which has numerous problems, not the least of which being that it implies there's no contingent facts to explain, making the argument self-undermining. Whereas weak or restricted PSRs either end up entailing strong one's, or are too weak to support the theistic-friendly conclusion[18]. There's also the Hume-Edwards principle, which suggests that we don't need an explanation for the set of all contingent facts, instead an explanation for each fact is enough, but that means we don't need a necessary explanation outside of the set of all contingent facts[19]. But that's just stage 1, which doesn't even get you to theism yet. You also have stage 2 arguments, which attempt to bridge the gap between necessary being and God, but I think those are even more unconvincing[20]. A lot more can be said here in reply, and reply back, but it would take us off-topic. I'm not here to discuss the contingency argument.
The most crucial problem here is, once again, BB's problematic atomism[21], and abuse of credences. When assessing one's credence in the success of arguments for theism such as contingency arguments, there are several interlocking parts and background considerations. I've already mentioned considerations surrounding the PSR, and stage 2 arguments, but even that barely scratches the surface. One might reject the argument on the basis of a generally pessimistic view of metaphysics[8]. One might reject the argument on independent or Moorean grounds, e.g the argument must have gone wrong somewhere, since you have a higher credence in the proposition that God, traditionally conceived, does not exist than you do in the premises. This could be due to the problem of evil in its many forms, divine hiddenness, arguments from divine incoherence, arguments for metaphysical naturalism, arguments from low priors etc. The point being, one's credence in the success of the contingency argument as a proof for theism (which would basically just be one's credence in theism) is determined relative to a background set of beliefs and methodological commitments. So BB's demand that we consider the contingency argument on it's own, and that "surely you should have a 1% credence in the success of the contingency argument, right?" Is, quite frankly, unserious. It's a rhetorical move and nothing else. One's credence in theism will be a function of weighing one's total evidence, background beliefs, and epistemic context therein, not some arbitrary credence one assigns when considering the contingency argument alone.
To see the point another way. We can use BB's atomistic approach[21] to credences to make an argument that theism is false with a probability of 1 given BB's own views. BB thinks all possible people exist. A set that has the cardinality of, at minimum, ℶ2. Let an instance of gratuitous suffering be an instance of suffering God would not permit. Now consider the proposition that "a randomly selected member of the set of all possible people minimally has a 0.0000000000000000000000000001% chance of having experienced gratuitous suffering at any point in their existence". That seems plausible, right? I mean it doesn't seem that unlikely a priori that some random possible person experienced gratuitous suffering, and 0.0000000000000000000000000001% is a pretty modest estimation. But the problem is, if for any randomly selected member of a set of ℶ2 people, there is a 0.0000000000000000000000000001% chance they experienced gratuitous suffering, then what's the chance that, of the entire set of ℶ2 people that at least one of them experienced gratuitous suffering? Well, since a set of ℶ2 people is so large, even if the probability of one randomly selected person experiencing gratuitous suffering is vanishingly small, the probability of having at least one member of that set experiencing gratuitous suffering will turn out to be 1, or at least pretty much 1.
Let G denote "it's not the case that a randomly selected person has experienced gratuitous suffering", and Gn be the proposition that ‘nobody in the set of ℶ2 people that exist experienced gratuitous suffering’.
But ~Gn is logically equivalent to the proposition, 'someone in the set of ℶ2 people experienced gratuitous suffering'. So, ~Gn has a probability of 1 and entails that God does not exist, since God would not allow someone to undergo gratuitous suffering by definition. It follows that 'God does not exist' has a probability of 1, given we accept my constraints and BB's view that there exists ℶ2 people. Yikes![22]
BB is not without responses. He might say "Hang on a minute. Something funky is going on here. We shouldn't be assessing the probability of the proposition that some random person undergoes gratuitous suffering in isolation. If we're theists, then our beliefs entail that there is no gratuitous suffering, so conditional on theism it's a probability of 0 that anyone experiences gratuitous suffering. Even if we don't conditionalize on theism, if you think there's some remotely plausible theodicy that explains some type of gratuitous evil, that theodicy once held in the background will probably cover every token instance of that type. So, it's not like we should assign an arbitrary credence to some person experiencing gratuitous suffering and assume an independent uniform distribution of the property of 'experiencing gratuitous suffering' across the set of all people". To which I would then reply "Congratulations! You've realized the folly of your credence-abuse".
That was our first extensive detour in this section. But like true soldiers we march on! BB mentions provability, lack of arbitrary limits, and 'ontological uniqueness' and claims that those are theoretical virtues of a theory, in response to my argument that theism has a low prior in virtue of having many ways of turning out to be false. Provability is what I just discussed. I go on to address the arbitrary limits argument in my initial piece, and we'll get into arbitrary limits more later here. As for ontological uniqueness, it's not clear what BB means here. He doesn't explain what he means nor cite any literature on why we should take it to be a virtue of a theory. Were I to venture a guess, ontological uniqueness means being such that some object is utterly unlike anything else that exists. In which case, not only do I reject that such a feature is a virtue of a theory, I think it's a vice. Some object being utterly unlike anything else that exists means that it fits poorly with our background knowledge and experiences of the world. It means that we have the best inductive evidence we could have against the existence of such an object. Inductive inferences generally take the form "every F we've seen has had some property G, so, for every F, it probably has property G". But being ontologically unique means God has properties that nothing else has let alone anything we could experience. Which means many of our inductive inferences will lead us to the conclusion that there is probably no object with God's properties. Perhaps BB means something else by ontological uniqueness, but if so, I can't be faulted for this misrepresentation, as BB didn't explain what he meant.
While we're on the topic of background information. BB says the following in reply to my point that theism fits poorly with our background information relative to naturalism:
Suppose you go into your drawer, and you find that your socks are missing. Oh no! You reflect on two potential explanations for this. E1 is that your spouse took them out while doing the laundry, or perhaps misplaced them by accident. E2 is that some sock-stealing fairy stole your socks. E1 and E2 look a priori equally intrinsically simple. Both make the same amount of ontological commitments, E1 to your spouse, and E2 to some sock-stealing fairy. Both commit to the truth of a similar amount of propositions, with E1 committing to "there exists your spouse, there are some specific desires your spouse has that make it likely that they'd be responsible for your socks not being there, and they have the power to make your socks not be there among other things" and E2 commits to "there are sock-stealing fairies, they have the desire to steal socks, they have the power to steal socks". If anything E2 might even be less complex, since the desires it posits are more simple, and the powers it posits are more simple. Nonetheless, you think E1 is much more probable than E2, to the point that you think E1 is a plausible explanation, and E2 isn't even a contender. Why? Because E1 fits with your background information much better than E2. In virtue of your experiences, you know your spouse exists, you know what powers they have, you know they do your laundry (or at least you know it's common for spouses to do laundry in general), and you may even know they have a tendency to misplace things. So I don't even need to argue that fit with background information is a theoretical virtue, you already probably believe that.
More charitably, maybe BB is saying that background knowledge doesn't matter when it comes to fundamental entities. If that's his point, then for starters I see no reason whatsoever to believe that, and so BB needs to argue it. Further, I think that's implausible. Every other theory I can think of, especially scientific theories, are assessed relative to our background understanding of the world, so why think theories that invoke fundamental entities are anything special? Even when scientists assess theories about elementary particles like electrons, the best theory is one which has to fit with our existing background knowledge. In this case, wave-particle duality, spectral lines, and Heisenberg's uncertainty principle. Such a paradigm shift in our background knowledge is in part why we eventually moved from the classical model to the quantum mechanical model. Finally, even if that's right I would then just take that to be a reason to be skeptical of all theories about fundamental reality. I see no reason to think we can abstract away from, and completely go beyond our background knowledge, our conceptual frameworks, and our experiences, to a priori intuit what the best theory of ultimate reality is from the armchair. In general I see no good reason to think the criteria of success employed by a priori theoretical metaphysics is truth-conducive. Among other things, you have the problem of how our minds could be causally related to the relevant object of our belief in such a way to produce reliable knowledge.
I argue that theism is unparsimonious when compared to naturalism. BB replies as follows:
Perhaps the idea is that the simplicity and multiplication of fundamental entities are all that matter when assessing grand theories. Firstly, that seems false to me. When it comes to grand theories we are evaluating all that those theories entail, the ontological commitments and the propositions loaded into them. Theories, after-all, are sets of propositions. So theoretical content has to do with the set of all propositions entailed from a theory, not just those propositions concerning fundamental posits. Theism and naturalism are theories about all that exists, and the shape of causal reality after-all, not just what exists at the foundation. But since grand theories are theories about everything that exists, we have, if anything, more reason to evaluate grand theories based on more than the fundamental entities posited by those theories. This just seems to follow from how we'd assess priors generally, we take the conjunction of all the propositions entailed by a theory, and calculate the probability of the conjunction of those propositions being true.
Perhaps the idea is that the simplicity and multiplicity of fundamental entities is what matters most when evaluating theoretical content, it "comes first" so to speak before other considerations regarding derived entities. I think this is the most plausible version of the "Laser" being appealed to, and indeed, I understand this to be Schaffer's view. In answering objections Schaffer is careful to stress that the multiplication or simplicity of fundamental entities is not the only thing to consider[23]. This doesn't help the theist, however. I think it can be plausibly argued that theism and naturalism are at least a wash when it comes to the simplicity of their fundamental posits, if not naturalism has an outright advantage. For starters, theistic grand theories are actually committed to the fundamental entities that naturalism posits, such as elementary particles, quantum fields, or the universe if we're priority monists. Theism posits the existence of a supreme supernatural agent in addition to those fundamental physical entities, so by the naturalists' lights, the theist is indeed multiplying fundamental entities beyond necessity[24].
But let's say we just compare the most fundamental entities of both naturalism and theism. I see no good reason whatsoever to think that God is more intrinsically likely than sets of elementary particles, quantum fields or the universe. On top of the fact that the existence of elementary particles, quantum fields, and the universe are well-established empirical facts that play roles in our best scientific theories, and theism both is not and doesn't. I think there are other reasons to think theism is not simple, and so doesn't get an advantage. God is attributed numerous complex properties such as conscious and intentional agency, infinite knowledge, infinite causal power, moral perfection, and so on. Where one could think (as I do) that each of these attributes add to the overall complexity, and lack of coherence in the concept of God. Also, as we've seen, theistic explanations invoke mystery and inscrutability, making them generally less epistemically simple compared to naturalistic explanations that are grounded in observable and testable phenomena. There are some ways theists have tried to get around this, by arguing that God is simple by virtue of being an entity with the property of perfection, or an entity that lacks any limits where all of God’s properties are just entailed from that. None of those, I think, are convincing, but we'll get to that later.
So, we have reason to think that theism and naturalism are at least a wash in terms of their fundamental entities. But that means we can then assess other features of those theories to break the tie, such as total ontological commitments, and the kind of ontology in general both theories propose. Naturalism posits a monist ontology with less ontological commitments, theism posits a dualist ontology with more ontological commitments[25]. So, when we move to simplicity considerations regarding non-fundamental entities, naturalism wins that round.
BB next objects to my argument from the success of the natural sciences which rules out supernatural explanations. His reply has two very brief parts, but despite this, I will examine both of them thoroughly. BB says:
Further, when discussing the idea of methodological naturalism as a progressive research project, the primary thing I brought up was novel predictions. But none of those arguments can be plausibly thought to be novel predictions of theism, in the same way that paradigmatic scientific theories such as general relativity have generated novel predictions. For instance, the existence of us, and our harmonious faculties was known prior to our coming up with theism, and indeed theism was constructed as a causal explanation for that phenomena as God is the creator of the universe and all life. So, it's not like theism predicts those things, independent of how theism as an explanatory hypothesis was constructed. Perhaps one could say that the fragility of the constants was discovered after the construction of theism. But theism doesn't predict the fragility of the constants, it predicts life, not that God would create life within a narrow range of life-permitting conditions. Note that I'm not even saying here that novel predictions are necessary for those arguments to work, just that novel predictions are a feature of a progressive research project that theism lacks.
Finally, I'll note that I reject all those arguments. I and co-authors have discussed our gripes with the argument from psychophysical harmony and I will not repeat those gripes here. Whether there is a theistic explanation for anthropics is the very thing in contention here when assessing our considerations as to whether a naturalistic theory should be preferred, so appealing to that is something akin to question-begging. The presence of laws gets into complicated stuff about laws of nature, which could warrant an entire article in its own right, so I don't want to discuss that here. So that leaves fine-tuning, which I think is probably the strongest among the arguments listed. There are many solid blogposts the discuss fine-tuning that make, what I think are powerful objections. There are many great papers with objections against the fine-tuning argument as well. Here are just 3 problems with the FTA: (i) it assumes an independent uniform distribution across all the possible values of the constants on naturalism but there's no good reason to assume that and good reasons to reject it, (ii) it assumes that, once life is counted as evidence for naturalism, then the fragility of the constants is further evidence for theism, but given God's omnipotence (and thus ability to actualize life within any physical conditions) there's no good reason to think that, lastly (iii) versions of the design hypothesis that are so constructed or constituted with resources to predict the life-permitting constants have a correspondingly lower prior.
I could go on but once again though, this argument has spawned a great deal of literature, and I can't do justice to the dialectic here without dedicating an entire article to it. BB's quick mention of those arguments without bothering to say anything more of substance was, I think, an attractive rhetorical move that might hold up in a live discussion, but it hardly passes for rigorous and serious philosophy.
Now, BB's next comment is the following:
BB appeals to his theodicy in his last comment, and this was his third time doing so in the article. So, I guess I'll take this opportunity to discuss it. The theodicy is the 'hypothesis of indifference' theodicy, which goes as follows:
True enough, God might have reason to create a universe like ours, but the same could be said, say, about a universe composed mostly of proto-conscious fart particles (no more conscious than fart particles in our universe if you accept panpsychism). Merely giving an epistemically possible thing God might do, does nothing to affect our evidential judgements, what matters is what God is more likely to do antecedently. In fairness, some other theodicies, like the saint-making theodicy, are also pretty ad-hoc on my view. But at least saint-making connects the evils we observe to entailed future goods in the afterlife, the 'hypothesis of indifference' theodicy, in itself, just says "well, maybe God would create an indifferent universe like ours" without connecting that to any even remotely plausible story on which greater goods are entailed from it.
I'll also note that the theory (theism + indifference), so construed, is highly gerrymandered and suffers from probabilistic tension. BB does not hold after-all, that God creates a wholly indifferent universe, He creates a universe that is weighted towards value insofar as it gives rise to sentient life. The problem is, virtually all reasons to think a designer motivated by moral reasons would be indifferent towards value in creating a universe would be reasons to think the designer wouldn't make a universe weighted towards conscious life with capacity for moral agency, and virtually all reasons to think a designer motivated by moral reasons would be different to bringing about valuable states would be a reason to think said designer wouldn't create a universe that's indifferent to value after life begins to exist. In fact, I contend that, antecedently, it appears to me much more likely that God would create a universe with no life at all, than that He'd create a universe with life, but that's indifferent to suffering and well-being. The former is more intrinsically probable (it's less theoretically complex in terms of desires it posits of God and less ontologically complex in terms of the outcome of those desires), and it aligns with my moral intuitions more (it seems very bad to create a universe with horrific suffering, it doesn't seem very bad to create a universe with no life). I don't deny that the theist can tell an epistemically possible story on which God creates a universe that leads to life, but is indifferent towards well-being and suffering, but again, I see no reason whatsoever to think such a story is any more antecedently plausible than any other merely possible story you can tell.
Finally, the theory of theism + hypothesis of indifference (call it Th) is empirically vacuous. Part of the problem here, I think, is that it's not clear what is meant by "indifferent universe". Prima facie, it looks like virtually any range of possible observations we might have can be easily accommodated with God putting us in an 'indifferent universe' and eventually heaven for the sake of greater goods, in accordance with exactly what we'd expect if Th is true. Indeed, BB himself says that evil wouldn't be any evidence against such a theory, it "explains evil precisely as well as naturalism" says BB. But since theism's predictive content is constrained only by moral goodness, the only observational evidence we could have against theism could only be evidence of what God would probably not allow, which would be gratuitous evil. But since any evil we observe is no evidence against Th, it follows that there's no evidence we can observe which can be evidence against Th. Suppose we observe the most horrific evil we could observe (not hard to imagine, some evils in our world are pretty close to that). Th can explain that by God putting us in an indifferent universe. Here's the problem; In the standard Bayesian framework, a hypothesis H is confirmed when the conditional probability P(H|O) > P(H). It's a theorem of probability that for any observation O, P(H|O) > P(H) iff P(H|~O) < P(H). This means that an observation O counts as evidence for a hypothesis H if and only if ~O would count as evidence against H. So, if no observation we might have can count against H, then nothing can bolster it either. But this would mean that, since nothing we observe can be evidence against Th, it follows nothing we observe can be evidence for Th making Th a non-starter as an explanation for any observation.
Perhaps BB could say that evidence of a “hell-world” that is different towards bringing about suffering, or bad-states might be evidence against Th. But even if some observations could verify us being in a hell-world and count as evidence against Th, all this would mean is that only observations falling under that type (O1) could be evidence against Th, but no other type of observation (O2, O3, On...) or lack thereof could be evidence for or against Th. Only ¬O1 (not observing a hell-world) could provide any evidential confirmation for Th, but even then, not much, since ¬O1 is strongly predicted by ¬Th too (given it's intrinsic improbability). Any other observation would be equally likely under Th and ¬Th. Thus, Th would still lack substantive empirical content, and the kinds of observations BB cites as evidence for Th (harmonious laws, specific persons, and other goods) cannot actually be evidence for Th, since observations of their absence couldn't be evidence against Th.
On the other hand, maybe BB means something much more specific by 'indifferent universe.' For instance, maybe what is meant is that God creates goal-directed biological organisms through unguided evolutionary mechanisms like those in our world, causing these organisms to experience sensations such as pain and pleasure in response to certain stimuli. These mechanisms, together with the conditions of the world, are indifferent to moral goals or values in such a way that we'd expect the kinds of horrendous suffering we see. While that does have some empirical consequences, that would just be to build the specific data we observe into Th. That is, quite clearly, viciously ad hoc and makes his model an unvirtuous just-so-story, not anything that resembles a serious explanatory hypothesis. Furthermore, the only reason we would be loading Th with those specific assumptions rather than others is because they entail our specific observations. Those observations then cannot serve as any evidence for Th, as this would be explicitly double-counting. Finally, if that specific and complicated theoretical content is built into Th, it will then also have a correspondingly astronomically low prior probability built into it, insofar as our observations themselves are taken to be extraordinarily improbable. None of BB's motivations for his theodicy deals with this problem of empirical vacuity, which, I take it, is devastating for his model.
BB also appeals to Van Inwagen's free will theodicy, but, qua ‘theodicy’, it runs into the same problems as the regular old free will theodicy as well as many theodicies in general, (doesn't explain natural evil, accords very poorly with our moral practices and intuitions, it's logically possible for God to create agents who freely choose the good and Plantinga's TWD doesn't work as a response to that and so on.). More importantly, it's worth noting that Van Inwagen grants that the stories to which He appeals, including the one where God allows suffering to allow creatures to freely choose Him are only "true for all we know", mere defenses, and he argues that this together with his broad modal skepticism defangs the problem of evil. But his modal skepticism would also, insofar as it defangs the problem of evil, make theism off the table as an explanatory hypothesis by making theism consistent with any observations given we can appeal to some 'for all we know' story about certain goods and their modal connections to our observations. While this is no problem for Van Inwagen who rejects evidential natural theology[27], this is a huge problem for BB. So Van Inwagen’s ‘theodicy’ definitely cannot be used in the way BB wants to use it, indeed his entire approach to the problem of evil is an anathema to the way BB wants to approach it.
So that leaves only connection-building (4.) and the divine intimacy theodicy (2.). There are two problems with those theodicies (there's more actually, but that would demand more space). The first problem is they don't actually explain the data. If God permits suffering for us to foster relationships with others, we'd expect there to be empirical evidence that horrific evils do in fact generally lead to those valuable relationships. However, not only does the empirical evidence not bear this out, the evidence we do have cuts against it. If God permits suffering to foster valuable relationships with Himself, then it makes the existence of non-resistant non-belief in its various forms, as well as its geographical distribution puzzling.
The second direct problem is that they belie a total lack of imagination. God is omnipotent and so can do all things logically possible. So even if we did observe that horrific evils was generally connected to valuable relationships of that sort, there's no good reason, so far as I can tell, to think such a connection is logically necessary. But if it's logically possible for God to bring about all the same goods of those valuable relationships without forcing creatures to undergo horrific suffering, then God, being morally perfect, would do that. Even if one thinks that, logically necessarily, a world with the most valuable relationships must involve shared struggles[28], it is logically possible for there to be shared struggles without horrific evils. Consider, as just one example, the shared struggle of a sports team up against insurmountable odds. That does not entail horrific suffering. BB could of course reply that he's just telling a story on which the goods of connection-building and relationships to God are instantiated in the afterlife, and it is somehow logically necessary that we experience horrific suffering on earth to experience those goods to their fullest extent. But again, I see no reason to accept such a story as being, even remotely, antecedently plausible. I could tell the story that any natural initial state of the universe, logically necessarily, must produce us. That's just as antecedently plausible by my lights, which is to say, not at all.
I'll also note that, as I said in my initial piece, the appeal to us being in an infinite afterlife in the future, implies that us being in a situation where we aren't in the afterlife, has an epistemic probability of 0. For any t, it's infinitely likely we'll be in one of the infinite times that we are in the afterlife, and not one of the finite times on earth.
I know this was a lot of space spent proportional to the two sentences I was initially responding to. But I harped on this both because BB often does these quick responses that seem to substantively respond to what I said, but actually just gloss over various issues, and because I wanted to highlight more of my discontents with BB's approach to evaluating and testing theism as a hypothesis, why I think it's much too credulous, and doesn't stand under scrutiny. BB seems to think he can just list out some theodicies, and that's enough to take out the bite of the problem of evil, without leading to other issues afflicting his model. But it's not that simple.
He also says "This is very implausible, and the history of scientific inquiry shows us discovering many unlimited things". So, you might think he goes on to provide examples of this. He doesn't even provide a single one. Convincing, I know, I am just utterly floored.
Now, in fairness, infinites do sometimes show up in our scientific theories, but usually just as idealizations to simplify calculations or indicators of where the theories break down. For instance, the singularity in a black hole is considered an infinite density, but this is understood as a sign that general relativity is incomplete and will ultimately be replaced by a more comprehensive theory of quantum gravity. Another example involving general relativity, GR implies that space is a continuous manifold, meaning that every path in space (and time) is infinitely divisible, which allows a mathematical description of paths through spacetime to be divided into an infinite number of smaller segments. Last example, in statistical mechanics, the thermodynamic limit involves taking the number of particles (N) and the volume (V) to infinity while keeping the density (N/V) constant.
Where infinites generally do not show up, is in the causal powers of the theoretical posits of our best scientific theories. Generally, theories which have posits with unlimited causal power are considered unvirtuous. First because, as mentioned before, it hampers our ability to generate testable predictions since if the theoretical posit has unlimited causal power it becomes consistent with a wider range of possible observable outcomes rather than the specific one's we get. Second because, in general, we would only posit what is necessary to explain our observations. If some posit having limited causal power is sufficient to explain our observation, then we wouldn't say that it's unlimited if such a commitment is unnecessary to explain our observation. Unlimited causal explanations posit more causal power than needed.
The problem is actually more severe here when considering my above quoted argument, which BB does not directly address. The idea is, while multiverse naturalism perhaps does not predict the existence of ℶ2 people, it predicts the existence of a subset of people (all physically possible people) that is known to contain you, or at least someone physically exactly like you with all your evidence, in a situation empirically indistinguishable from yours, as a member[30]. So it entails the observed data, despite not entailing ℶ2 people. This does not threaten the bare SIA, which has an all else equal clause, and all else is clearly not held equal here because this is a case where we precisely know the subset of possible people that multiverse naturalism entails, and it's all those people that are jointly physically possible of which you (or someone physically identical to you with the same evidence) are a member. This also puts us in a position to once again see why whatever principle BB has to appeal to in order to undermine my argument, in this case the N-SIA, must be false, this time as a trivial matter of logic and Bayes theorem.
Call the hypothesis Hu the hypothesis that you exist and there are only 100 total observers. So, Hu is logically equivalent to the conjunctive proposition 'you exist' (call this Y) and 'there are 100 observers in total' (call this R). It's a theorem of propositional logic that, for any p and any q, p&q entails p, and p&q entails q. In other words conjunctions entail their conjuncts. So this means Hu (R&Y) entails Y. Which also means P(Y|Hu) = 1. Or Hu predicts your existence with a probability of 1. So when accounting for your existence, Hu gets as much of a boost through updating in line with Bayes rule as any other theory which entails that you exist. But Hu also entails R, that only 100 observers exist. Thus, a theory which predicts only 100 observers exists gets as much of an update, conditional on the fact that you exist, as any theory which predicts more observers and which entails your existence. Since, for any hypotheses H1 & H2, and any evidence E, if H1 and H2 both entail E, then E can't be any evidence for H1 over H2. Of course, this is not to deny that Hu is a bad theory and will have a correspondingly ridiculously low prior since it just contains what it explains, (a problem which, importantly, does not afflict multiverse naturalism. As there are plenty of reasons to think many-worlds theory is independently probable or so I've argued in my initial piece) but that doesn't change how much of a boost Hu gets when updating on the data. So, simply assuming standard propositional logic and Bayes theorem leads us to trivial counterexamples to the N-SIA, where theories which posit much less observers get just as much, if not more of an update as theories which posit more observers. Which would mean that even if the SIA entailed the N-SIA, BB's arguments in the linked articles would then not be arguments for the SIA, but a set of regrettable bullets we have to bite on pain of rejecting propositional logic or Bayes theorem. Fortunately, though, the SIA does not entail the N-SIA, so we can just reject the N-SIA with no problems. Since only the N-SIA would, if true, render my argument unsound, BB's response fails.
BB lastly makes mention of modal realism, saying it collapses induction. But that's irrelevant, as multiverse naturalism is not modal realism, and does not collapse induction. This is because, on multiverse naturalism, each universe is deterministic and obeys the laws of physics. There may be other skeptical scenarios that might be run against multiverse naturalism, however, such as Boltzmann brains. But insofar as that afflicts multiverse naturalism, so too does it afflict the version of theism that predicts all possible people. Since the supposed problem would be that there are much more observers that are Boltzmann brains, than there are typical observers like us. But that's a problem regardless of if you accept a God-created multiverse with every possible person, or a naturalistic multiverse with every physically possible person. Further the Boltzmann brain skeptical scenario can be run even without a multiverse, and just the fact that in the distant future, our universe is expected to reach a state of maximum entropy where random fluctuations can occur due to quantum effects or thermal processes. With the idea being, given enough time, it can be hypothesized that there will be much more Boltzmann brains as a result of these fluctuations than typical observers. Whatever response you have to that, is a response the multiverse naturalist can make use of as well. I must conclude then that BB's response to the multiverse naturalism objection is a failure.
Good philosophy is applying serious scrutiny, not just to your interlocutor's arguments but to your own. It's taking the time to deeply consider your interlocutor's perspective, to carefully reflect upon their objections, and why they were made. It's not rapidly sifting through their objections, and prioritizing putting a quick reply out above all else. Unfortunately, it is the latter, not the former that appears to be emblematic of BB's general approach.
As for myself, I remain just as unmoved and unimpressed by the anthropic argument for theism as I was when I wrote my initial piece. None of BB's attempted defenses of the argument from my objections are successful. By my lights, pretty much no naturalist ought to feel all that threatened by the argument. It is however, possible, albeit very unlikely, that I will change my mind and come to think of the anthropic argument as, at least, an evidential chip in theism's favor. So, I'll end this on a conciliatory piece of advice.
If BB wishes to seriously challenge my view that the anthropic argument is wholly unpersuasive, he will need to do better than a rushed reply. I encourage reading through this and my original post very carefully and thoroughly, and putting serious effort and time into understanding and engaging with my objections, and perspective. Before putting out this response, for instance, I read through BB's reply and my draft multiple times to check for errors, and even had a philosophically intelligent friend proofread my draft and give feedback. BB should do the same, or something similar enough before releasing a response. If he comes out of it still disagreeing, that's perfectly fine. But I at least hope for more diligent and rigorous engagement in the future.
Basically, you should reason as if you are a member of a universal class of all possible observers each of which are equally likely to exist. This entails that, all else equal, hypotheses or scenario's on which there are more observers are more likely conditional on your existence, because, of the observers that could exist, there is a larger share of those observers that do exist and a higher chance for you to be among the observers that are instantiated. Here is the much stronger claim that BB appears to be endorsing:
N-SIA: Necessarily, if a hypothesis H1 posits n-times as many observers as H2, then, upon updating on your existence, you ought to think H1 is n-times more likely than H2.
Notice that this isn't entailed from the SIA. The SIA has an "all else equal" clause, the N-SIA does not. The SIA just tells you how to reason about yourself as an observer among a set of possible observers (that is by taking yourself to be a random sample), the N-SIA proposes a necessary condition on which hypotheses one ought to think are more likely conditional on your existence. While the SIA does imply that scenarios with more observers are more likely when all else is held equal, this does not entail, contra BB, that it is necessarily the case that, conditional on your existence, you should think hypotheses which entail more observers are more likely.
This is a crucial distinction that I must stress, because only the N-SIA, and not the bare SIA would, if true, undermine my objection that there is no reason to update in favor of hypotheses with more observers where the total class of observers is, say, ℶ2 (an uncountable infinity that represents the power set of the real numbers). The standard SIA is perfectly consistent with my objection, in fact the objection stands on firmer ground given the SIA. The idea is, if we reason as if we are a random selection from the set of all possible observers, then no matter how many observers of the class of possible observers are instantiated, the probability that you are randomly selected to be instantiated as an observer in that set would always be 0, if it is even defined (which isn't clear to me).
I'll stress that the SIA, unlike the much more radical N-SIA, does not imply that having more observers increases the probability of your specific existence if the pool of possible observers under consideration is proportionately larger, it only implies that when all else is equal (and this would include holding equal the size of the pool of possible observers) more observers are more likely given that you exist. In fact, the problem is much worse, the standard SIA actually implies that the N-SIA is false. This is because reasoning as if one is randomly selected from a set of all possible observers would only lead one to update in favor of scenarios with more observers if it means a greater proportion of observers in the total class of possible observers which includes you exist, which isn't the case when the total class is infinite. Let me illustrate this with a simple analogy:
Suppose you have a deck of cards, where you know that within the deck is the ace of spades. There are 2 scenarios, and in each scenario the deck of cards is randomly shuffled.
1. You have a deck of 10 cards, and you know the ace of spades is in that deck. You select 1 card at random.
2. You have a full 52 deck of cards, and you know that the ace of spades is in that deck. You randomly select 4 cards from that deck.
Now, if we treat the cards as observers and reason as the SIA tells us to, you would reason as though the ace of spades is a random sample, equally likely to be any one of the individual cards. So, in scenario (1.) there's a 1 in 10 chance that the card you pull is the ace of spades. In scenario (2.) there's a 1 in 13 chance, that, of the 4 cards you pulled, one of them is the ace of spades. But, wait: That means that the 1 card you pull is more likely to be the ace of spades in scenario (1.), than any of the 4 cards you pull being the ace of spades in scenario (2.), despite the fact that you pull out 4 times as many cards in scenario (2.). But this means, if you reason as if each card is a random sample from the set of all cards in the relevant domain of quantification, it's not the case that, necessarily, having more pulled cards makes it more likely that one of those pulled cards is the ace of spades. What matters is the relative proportion of pulled cards to the total set of cards being pulled from which contains the ace of spades.
Let's now apply this analogy directly to an example scenario involving observers: suppose there are 2 random-observer-making machines, M1 and M2. M1 is programmed to select among a class of 500 possible observers, among which you are one of them, and to select 100 of those observers to create at random. M2 has a much larger class of possible observers it can create, 1 trillion let's say, among which you are one of them. M2 selects 200 million of those observers to create at random. Only one of the machines performs their function. You learn that you exist and were created by one of the machines. Which machine should you think performed its function?
The probability that you exist given that M1 performs its function and creates 100 observers, is 1 in 5 or 20%. The probability that you exist given that M2 performs its function and creates 200 million observers, is 1 in 5000, or 0.02%. So, it's 1000 times more likely, given that you exist, that M1 performed its function than that M2 did. You should therefore think you were created by M1, even though the fact that you were created by M1 implies that much less observers are created than if you were created by M2. This is completely consistent with the SIA, in fact if anything it is probably implied by how the SIA tells you to reason about your existence. The SIA says to treat yourself as a random sample in the set of all possible observers, not to treat the probability of your existence as being necessarily higher in relation to any arbitrary scenario where you happen to exist alongside more observers. Treating yourself as a random selection from the pool of possible observers, and indeed, simply appealing to standard ways of assigning epistemic probabilities in situations of ignorance such as the principle of indifference implies that there are possible albeit abstract hypotheticals, like this one, where you have more epistemic reason to update in favor of scenarios which entail less observers in light of your existence.
BB also claims that the motivations he has for the SIA motivate the claim "that more people existing is more likely" regardless of the percentage of possible people instantiated. This is false, the motivations he provides in the linked article are motivations for the regular SIA, many from Joe Carlsmith. But the SIA, as I've argued, actually implies that the percentage of people instantiated does matter. I don't want to waste too much space so I'll just address one scenario BB provides in his article where he cites Carlsmith:
Save the puppy: You wake up in a red jacket. In front of you is a puppy. Next to you is a button that will create a trillion more people, all wearing blue jackets. No one else exists. A giant boulder is rolling inexorably towards the puppy, and it will crush the puppy with very high probability. You want to save the puppy, but you can’t reach it. However, you accept SSA, and you understand the power of reference classes. So you make a firm commitment: if the boulder doesn’t swerve away from the puppy, you will press the button; otherwise, you won’t. Should you now expect the boulder to swerve, and the puppy to live?The absurd conclusion here arises if one thinks that the boulder will swerve just because if it doesn't you'll press the button creating a trillion observers and it would then be extremely unlikely that you'd be the first observer. On the SIA this concern is mitigated by considering yourself as a random sample from the set of all possible observers, rather than a random sample from the set of all actual observers. This means that while the probability of being the first observer upon pressing the button is exceedingly low, the probability of existing in a scenario with a much larger pool of observers is proportionally higher since more possible observers are actualized, and thus more 'slots' are available for you to exist, which balances out the probabilities (all-else-equal). Therefore, you have no reason to expect the boulder to swerve conditional on the fact that you'll press the button if and only if it doesn't swerve. But this solution doesn't require committing to the N-SIA, the bare SIA solves it by canceling out the update when you press the button.
I could go on addressing other motivations BB offers for the SIA. But the point is, they are motivations for, well, the SIA over rival views like the SSA (which I'm by no means committed to). If BB thinks he has motivations specifically for the much stronger N-SIA that does undermine my argument, it's his burden to provide them, a burden he has completely fallen short of meeting. The motivations present in the article linked, so far as I can tell, if accepted, don't give us any reason whatsoever to accept the N-SIA in favor of the more modest SIA. Even if they did, we should still reject the N-SIA, since as we've seen, it entails incorrect reasoning about epistemic probabilities (and we'll see later in the last section that the N-SIA can be proven false another way).
In sum, BB is wrong that my argument against updating in favor of more observers in infinite contexts depends on rejecting the SIA or the motivations BB has offered for it, they only depend on rejecting the N-SIA. However, not only has BB given no good reason to accept the N-SIA so far as I can tell, the N-SIA is plainly false, and so it's no cost that my argument depends on its falsity. I’ll also note that even if my argument did entail the falsity of the bare SIA, BB would still need to engage with my argument directly, and it’s not at all sufficient to point to his independent motivations for the SIA. Even if the SIA does better than any rival views in anthropics, such as the SSA, it may well be the best of a bad lot.
BB tries to directly address my objection with an analogy:
For another, there is an important sense in which a greater share of possible people exist if ten people out of infinity exist than five. Sure it’s still either zero or infinitesimal, but it will be, in some intuitively important sense a bigger infinitesimal. Suppose that you’re in Hilbert’s infinitely big hotel and your room is red. You’re considering between two possibilities—either googolplex rooms were painted red or only one was. You should think the first hypothesis is more likely, even though they’re both probability of either zero or infinitesimal. The math around this is tricky and there’s no agreed upon way of doing it, as TT notes, but there are still certain clear results.Now, it's true that there's a sense in which it could be rightly said that 10 is a 'greater share' of ℶ2 than 5. In the sense that if you have 10 members of that set, you have more than 5 members. That's trivially true. But there's also an important sense in which it is simply false that 10 is a 'greater share' of ℶ2 than 5, if "greater share" here is understood as 'taking up a larger fraction of ℶ2'. Both 5 and 10 take up exactly the same fraction of ℶ2 in standard set theory and cardinal arithmetic, which is 0. But it's the latter, not the former sense that's relevant to calculating the probability that, among all observers in a set of ℶ2, how likely you are to be in a particular subset of 5 or 10 observers in that set. If you assume that you are a priori equally likely to be any observer in the ℶ2 set of observers, the chance of you being in any particular subset containing 5 or 10 agents is 0, because those subsets leave out 100% of the total amount of observers in the set, ℶ2.
BB claims that if you're in Hilbert's Infinite Hotel and you learn that your room is painted red, upon updating you should think it is much more likely that googolplex rooms are red than one. But he doesn't show how this update actually bears out probabilistically, he just baldly asserts it as if it were an obvious truism. Contra BB that's not at all obvious, and the only reason to think it's obvious, so far as I can tell, would be if one inappropriately extends one's intuitions about probabilities regarding finite sets to infinite sets. In fact I contend that it is false, both hypotheses have a conditional probability of 0 given that you see yourself in one of finite red rooms and there are infinitely many non-red rooms each of which you had an equal chance of being in, since, both googolplex and one are 0% of even the smallest transfinite cardinal.
Bizarrely BB seems to concede this, but yet he still says one should still think the hypothesis that a googolplex rooms are painted red is more probable. But that's incoherent, if they both have a probability of 0, then, well, they both have the same probability, 0. Charitably, maybe he's just saying you should still think googolplex rooms being painted red is more likely, even though it's false that it's actually more likely. But that looks odd, surely on any plausible account of epistemic rationality you shouldn't believe false things. More charitably, maybe he's saying that we should use non-standard analysis so we can say a googolplex within infinity is a larger infinitesimal than one within infinity. But even if that is granted, how this would bear out probabilistically is at most highly unclear and BB has not even attempted to demonstrate that this would permit a substantial Bayesian update in favor of more red rooms. Overall, at the very least BB's claim demands serious clarification and motivation. Recall that precisely what's in dispute is whether Bayesian updating in line with the SIA works in infinite cases in a way that allows the anthropic argument for theism to succeed. Simply insisting that it does work that way is dialectically inert.
BB also says:
TT’s argument, if successful, would totally jeopardize anthropic reasoning. If SIA is correct but impossible to do if there are infinite possible people, and there are infinite possible people, that means all anthropic probabilities end up being undefined and anthropic reasoning becomes impossible. Clearly, this is wrong!But this is simply a massive overstatement to the point of being misleading. My critique only implies that an inference starting from the SIA to updating in light of your existence to more observers within an infinite universal class of observers, fails. This doesn't entail that any and all anthropic reasoning in infinite cases is doomed, at most it undermines a certain kind of inference and prompts a reevaluation of how we might apply such reasoning when dealing with infinite possibility spaces of observers. Further, anthropic reasoning is not limited solely to contexts defined by the SIA or infinite sets of observers. It's consistent with my argument that there are numerous other contexts and configurations of possible observers where anthropic reasoning is not only applicable but fruitful and serves useful functions in our epistemic practices. Finally, even if my arguments were devastating to anthropic reasoning conditional on there being infinite possible observers, all this would mean is that we should either reject that there are infinite possible observers, or we should be humble and withhold judgment about the correct approach to anthropic reasoning. What's wrong with that? BB doesn't explain why that's clearly wrong beyond simply asserting that it is.
Next, BB addresses my pre-emptive rebuttal to the claim that God would create all possible people not just the same cardinality. He says:
So TT says that God would just create ℶ2 people—assuming there are ℶ2 possible people—but not necessarily all possible people. He says that creating all possible people involves creating the same number of people as just creating ℶ2 people because it’s the same cardinality.But even if two actions both involve creating the same cardinality of people, if one of them involves creating a proper subset of the people of the other, then the other is better. If you’ve already created aleph null people, but then you can create Suzie and give her a good life, you should do so. It’s good to create a happy person—as I’ve argued elsewhere—and that doesn’t depend on whether an infinite number of people already exist.
BB here claims that if given a choice between creating 2 sets of people, even if both sets have the same amount of people given happy lives, it's better to create one that isn't a proper subset of the other. But why believe this? Having a proper subset of a set of happy people would be plausibly less valuable than having the complete set solely in virtue of the fact that the complete set has all the happy people of the proper subset and more. This would be true in the case of finite sets. But we are talking about a case where we have 2 infinite sets with the same cardinality, which means the exact same amount of happy people are created in both cases.Suppose that the future is infinite and someone will get stabbed once a day forever. That’s infinitely bad. Yet surely it would be worse—even though it would involve the same cardinality worth of pain—if they got stabbed twice a day forever. That’s because it would add extra badness, where the badness of the original is a proper subset of the original badness. In a similar way, even if God has made ℶ2 people, why in the world would he stop short of creating new people to whom he could give good lives
BB says that creating a happy person is a valuable action, and that doesn't change just because you already have infinitely other valuable people. However, I'm not at all committed to denying that. In fact, it can be granted that for any happy person that is created, there is value in creating them, regardless of how many others are already created. The problem comes when we think about God's decision regarding which set of ℶ2 people to actualize. Let's say you have S1 which contains all possible people, and S2 which contains the same amount of possible people as S1, but we take out some arbitrary number of specific elements, in this case possible people, including you and 'Suzie'. We know the act of creating a happy person for the sake of their good has intrinsic value. We can label the intrinsic value of creating a person with a good life, V. We can then have a function which maps every person in S1 or S2 with a distinct Vi, where Vi is the intrinsic value of actualizing that particular person.
Here's the kicker. Since both sets have the same cardinality that means that all elements from S1 can be put in one-to-one correspondence with all elements of S2. In other words, for every created happy person in S1, there's a corresponding created happy person in S2. This also means, for every Vi that corresponds to an element in S1, there's a Vi that corresponds to an element in S2. So, the intrinsic value of creating the people in sets S1 and S2 is equal, since every act of creating each person in S1 has a corresponding act of the same intrinsic value, of creating a person in S2. So, the real issue isn't whether there's a reason to create Suzie when you already have infinite people; it’s whether, prior to creation, God has any reason to choose to create an infinite set that includes Suzie over an infinite set that doesn’t, if both sets have the same cardinality. Each person in the set without Suzie can be mapped with a person in the set with Suzie, and each act of creating those people would confer the same intrinsic value, V. Thus, BB’s claim that God would create all possible people, even if this doesn’t imply creating a greater cardinality of people is unmotivated, the reasons he tries to give are uncompelling.
BB also uses an example of getting stabbed twice versus once for an infinite amount of days. But again, since our ways of aggregating and comparing quantities simply don't work the same way for infinites, it's unclear why we should put much stock in BB's intuition. But assuming we do hold that a person getting stabbed twice a day for infinite days is worse than them getting stabbed once for infinite days, this could not be because the total amount of suffering endured is greater (they're the same). It would have to be because for each individual day, the suffering the person undergoes is more intense and painful. But then how this analogy is supposed to apply to God's decision to actualize S1 over S2 is unclear. Each member of S1 can be mapped to a member of S2, where the intrinsic value, V, of creating each person and giving them a happy life is present equally in both cases.
Next BB says:
TT accuses me of inconsistency in that I rely on intuition at various points in the argument (for example, in arguing for USIA I claim that it seems that one’s credence and betting odds should be the same—e.g. if you’re two-thirds confident in a proposition you should bet in favor of it at up to 1:2 odds), but ignore the counterintuitiveness of my position. Yet why is my position counterintuitive? I grant that the existence of infinite people sounds weird, yet there’s a big difference between a position sounding weird and genuinely seeming wrong. To me it seems strange that we would have bare intuition about the number of people.
He continues:
TT claims this is counterintuitive because it’s unclear how we would map the actual people onto the set Beth 2. It’s true that there isn’t an obvious procedure for doing this, but they can be shown to be the same cardinality (Lewis showed the number of possible people across worlds was at least Beth 2). I don’t know exactly how you would design a procedure to map the reals onto the powerset of the naturals either, but I know there is a bijection to the two because they’re both cardinality Beth 1. He notes we can’t list the real numbers, but that doesn’t rule out a possible bijection between them and some set equal in sizeThere's two confusions here to address. First, BB claims that we can acknowledge sets like the reals and the powerset of naturals as having the same cardinality without knowing a specific mapping procedure. In fact, our understanding of their equivalent cardinalities comes precisely from our ability to prove that there is a bijection between them. For instance, each real number in the interval [0,1)[0,1) can be represented by its unique binary sequence. We can map each sequence to a subset of naturals, where each digit in the sequence (0 or 1) indicates the absence or presence of a corresponding natural number in the subset, creating a bijection between the sets.
Second, BB misconstrues my point in that section, though perhaps I could have been more clear. It wasn't that being unable to list the real numbers prevents us from having a bijection on a set as large as ℶ1 or ℶ2. Nor was it simply some argument from ignorance from our lack of knowledge of a way to map concrete people in the world to a higher-order infinite set. It's that there is a positive intuition that it is impossible to do such a mapping. For any amount of actual people, even if there were infinitely many, it seems we can conceptualize them as first, second, third, and so on, mapped to 1, 2, 3, etc., but we'd never at any point map people to irrational numbers like π or complex subsets of real numbers. There could never at any point be a 1.78657089….th person, or so it appears. It seems like a set of actual concrete people would have to be discrete, not continuous like the set of real numbers or bigger sets like ℶ2. People seem to be distinct countable entities that can, at least in principle, be listed and individually identified with no intermediate values between them. But higher-order infinite cardinalities are not listable, between any two real numbers there are infinitely other real numbers. It’s perfectly fine and sensible to think about such sets abstractly, but how could we even begin to map concrete actual people in reality to a set with such properties? I'll stress that this doesn't deductively prove that it is conceptually or logically impossible for there to be ℶ2 people. It simply illustrates that there is something counterintuitive about it. On reflection, myself and likely many others will find the notion of ℶ2 people existing far more counterintuitive than the intuitive appeals in infinite cases that BB likes to rely on.
Next BB replies to my objection that while all-else-equal the SIA will lead us to favor hypotheses with more observers, the 'total evidence' does not favor hypotheses with more observers. While more observers make your existence more likely, they also make you less likely to be you and not someone else, as there are more possible identities to assume. BB starts with the following;
I have two problems with this. First of all, it is wrong as a matter of probabilistic reasoning, for reasons I’ll explain in a moment. Second, and more importantly, it is not SIA in any real sense, and as a result, every argument given for SIA is an objection to this view. I will not repeat the arguments for SIA as I’ve already given them in a million different articles and everyone is sick of them, but all the arguments given here, for instance, are fully applicable in exactly the way they are to every alternative to SIA. To avoid the result described in section 3, one has to think that one should think the world has a lot of people given that they exist, which they shouldn’t if the update is cancelled out.BB once again claims my objection (we'll call it the TE objection short for total evidence) would be an objection to the SIA itself. Now, I've already dedicated a semi-extensive discussion to that objection, and even addressed the result in section 3 to which BB refers. Once again, it's utterly unclear to me how my TE objection at all undermines the bare SIA as formulated by Bostrom[5]. In fact, in laying out the TE objection, I explicitly assume that we are reasoning as if we are a random sample from the set of all possible observers, and that, in taking into account one's existence, we update in favor of hypotheses which entail more observers. All I do is add that once your existence is in the background, knowing further data related to your identity as a specific observer with a certain body, set of experiences, perspective etc. leads to an update in the other direction. This doesn't at all contradict the SIA, in fact it refines its application by incorporating a more detailed assessment of what Bayesian reasoning about ourselves as observers entails within large populations.
The SIA entails that all else equal taking into account the data of your existence leads to an update in favor of scenarios with more observers, it does not entail that you can’t have other evidence which leads to an update in favor of less observers. So, on its face, it's simply not true that any argument for the SIA would undercut the TE objection. The purported absurd results BB provides are ones which afflict alternatives to the SIA like the SSA, which I'm not committed to adopting. If BB thinks he has a rebutting or undercutting defeater for the TE objection, it is his job to provide it. However, he provides no such defeater, so as a result his claim is simply unmotivated.
BB however does attempt a direct attack on the TE objection:
I am confused by this, mainly because it's just unclear to me how it follows from anything BB says here that the TE objection fails as a simple matter of probability as he claims. In BB's coin-flip analogy, we can agree that if all you know is you were created in a room after a coin flip—with heads creating one person in room 1 and tails creating two people, one in each room—then tails is more likely. However, as BB himself acknowledges, this is under the condition of uncertainty about which room you're in. Contrary to this scenario, in reality, we are certain about our specific identities, bodies, experiences, and perspectives. Those details aren't vague or unknown; they are in fact, critically, a part of your "current evidence."It’s also wrong as a matter of probability. I’ve made this point before, but it’s a tricky one, so I’ll repeat it. Suppose a coin is flipped. If it comes up heads, a random person is going to be created in room 1 and if it comes up tails a random person is going to be created in room 1 and another random person is going to be created in room 2.Upon being created in a room my credence should be 2/3 in T2. That’s because I’m twice as likely to exist on tails. So now I split my credence three ways—there’s a 1/3 chance I’m created in room 1 and the coin came up heads, 1/3 room 1 and it came up tails, and 1/3 room 2 and it came up tails.If I knew I was in room 1 then it would cancel out the update. But I don’t—I might be in room 2, which straightforwardly confirms tails. Therefore, when one accounts for the uncertainty about which agent I am in which world, there’s no second update. It’s true that if there are twice as many people the odds I’d be one of the first few people are half as great, but if I don’t know if I’m one of the first few people, then there’s no update in favor of views on which I’m likelier to be one of the first few people.Because I don’t know which location I’m in, I can’t update in favor of a greater likelihood of being in the particular location I’m in—with my skin and such—without taking into account the probability of my existing with my current evidence. But that’s just determined by the number of people with my current evidence the theories predict. So there’s no update that cancels out the SIA update
If, following the analogy, you discover you're in room 1, this new evidence cancels out the initial update in favor of tails, since the scenario of being created in room 2 is ruled out and you’re only left with two possible outcomes. Thus, if BB’s analogy allows for a second update upon gaining new information about your location, it should similarly allow for a second update when taking into account evidence of your specific identity as an observer among other observers, including your unique body, experiences and properties—features that differentiate you from other identities that exist. As I argued in my original piece, this update would counterbalance the update in favor of hypotheses with more observers when taking into account your existence. So, even granting the SIA, the total evidence doesn't favor a scenario with more observers all-things-considered. Nothing BB says here remotely undermines this point.
First Recap
BB had recaps in his response so I figured I'd do the same. To quickly recap the last sections:- BB's appeal to proper classes in set theory, as collections "too large to be a set", may be coherent, but it simply doesn't help in the context of our discussion of how many possible observers we should think there are. He also points out that Grim's original Cantorian argument only explicitly attacks the possibility of a 'set’ of all truths. However, it can be formulated to apply to any class or collection of truths, or so Grim himself believes.
- BB does not adequately counter my argument that if a set of all possible people can't exist, neither can a God-actualizable state of affairs with all possible people. Though this wouldn't necessarily destroy the anthropic argument itself.
- BB claims that my objections to the applicability of the SIA in infinite contexts, and that our total evidence doesn't favor scenario's with more observers all-things-considered, actually just imply the falsity of the SIA itself. As a matter of fact, they only imply the falsity of a much stronger principle which I call the N-SIA, which is false. But even if they did, he'd still need to engage with my arguments beyond gesturing to independent motivations for the SIA. BB also falsely claims that my objections would doom anthropic reasoning all-together.
- BB's claim that we should still update in favor of scenarios with more observers, despite them taking up 0% of the possible observer space, is both unmotivated and unclear.
- BB says that even if two cases involve creating the same cardinality of people, God would still create all possible people even if it doesn't mean creating more. However, this claim is unmotivated. As I argue, if two actions involve creating the same cardinality of people, this entails they also would have the same intrinsic value.
- BB misunderstands my point concerning the counter-intuitiveness of having a set of ℶ2 people, and, consequently, fails to properly address it.
- BB proposes an analogy that he thinks counters my total evidence objection to the anthropic argument, but it just doesn't.
The Double-counting Accusation and BB's Credence-Abuse
BB addresses my point that it is far from clear that it's true that it is always, necessarily, good to create a happy person and give them a good life. As this assumes certain controversial views in population ethics. For instance, it assumes that anti-natalism and person-affecting views are false. He says:First of all, even if it did assume that, that wouldn’t be a problem because those views are extremely improbable (see my articles linked above, also Thornley’s excellent paper—I have a paper that I’ve sent to a journal that has, I think, a totally decisive argument that I’ll write about when it’s published). Second of all, as I described above, we’re here considering the relative probabilities of different views—even if you think the odds of anti-natalism are pretty high, as long as there are nonzero odds of pro-natalism, your existence will favor theism.On this first point, I'll limit myself to some brief comments, as I don't view it as central. My mention of anti-natalism and person-affecting views was to highlight that BB's argument depends on accepting controversial assumptions in population ethics. For those who find those views plausible, there's a strong antecedent reason to question the crucial premise upon which the anthropic argument depends. It's not surprising to me that BB thinks those views are extremely improbable, but I think he, at best, overstates the case against them (something BB has a tendency of doing).
With respect to his second point, the idea that "even if such and such has a low probability, you should update in light of your existence, on theism plus such and such". This is another reoccurring theme in BB's reply, which I'll say a lot about later in this section. I'll make a couple points for now. First, I'll note that the claim that your existence significantly boosts the probability of theism does not merely hinge on pro-natalism, it hinges on a more specific interpretation of divine motives. As I noted in my original piece, it assumes not just that God would make people, but that God would create indiscriminately—that it is always all-things-considered better than not to create a possible person with a happy life. But even if creating happy people is generally a good thing, it does not follow that it is all-things-considered good to create more and more possible people with happy lives without any restriction. Many possible theistic narratives, for instance, allow for a God who creates selectively, according to divine wisdom or a larger cosmic plan, prioritizing other goods over sheer numbers of happy individuals.
Second, BB assumes, without any justification, that it is methodologically sound to update our beliefs about ethical views like pro-natalism based on their utility in explaining the data when combined with theism. But I am not at all inclined to grant him that much. I take it ethical views are the kinds of things that must be independently justified, by virtue of their accordance with our moral reasoning and practices. They aren't themselves variables to be adjusted at our behest based on what best accommodates the already observed data when combined with whatever metaphysical postulate we're using to explain that data. Relatedly, I suspect there may be a looming circularity worry here. Why do we exist? God creates us. Why think God would create us? Because such-and-such ethical assumption. But why think such-and-such ethical assumption is true? Because it explains why God would create us and accounts for the improbability of our existence. Of course, this wouldn't be an issue if one holds to the ethical assumption on independent grounds, but insofar as one updates in favor of the ethical assumption for the sole purpose of explaining some otherwise improbable data regardless of it's independent plausibility, that looks to me like a vicious explanation. You don't use the same data as evidence for a theory, and as a basis to tailor the theory's assumptions. You can't use the data twice, as that would be double-counting, and as we will now see, BB acknowledges the folly in double-counting:
I also think that even if this is right, it doesn’t affect the relative force of the argument. If anti-natalism is true then theism can’t be (for God wouldn’t have created at all). So that means the probability of theism equals the probability of theism + pro-natalism. Thus, given that that’s the only plausible theistic hypothesis before one examines the argument, if the argument raises the probability of theism + pro-natalism, then it raises the overall probability of theism (this is the double-counting charge that TT objects to later).So this is yet another reoccurring theme in BB's reply. He says my critique, if true, would imply the falsity of theism, since we know that creation obtains. So, it is double-counting to appeal to it as a response to the anthropic argument. He also says this about my arguments that God probably would not create at all:
Stuff was created. So if theism is true it has to be that God would create. This means that the only type of theism that was plausible prior to thinking about the anthropic argument was one on which God would create. Thus, if theism is already committed to that, it won’t affect the degree of the update. For Truth Teller to justify his claim that the anthropic argument doesn’t raise his credence in theism one bit, he’ll have to do more than raise this objection, because this doesn’t affect how big of an update he should get in favor of theism.He also says it about my godman/duplication objection:
If this is right then theism is false. Thus, the argument—for the reason I’ve now explained several times—doesn’t affect how much the anthropic considerations should raise our credence in theism. It just provides a separate objection to theism.So, BB appears to hedge a lot on this objection. Let’s start by examining it in the context of my objection based on God and creation, which is the one place I take it, it can be charitably interpreted. Let’s also get clear on what double-counting is, because it has a specific meaning in probabilistic reasoning and theory choice. Double-counting is when you use the data to be explained in formulating, or updating a hypothesis, and then use that same data as evidence for the hypothesis. But I haven't counted the data even once, let alone twice. I'm not making the point that theism must be false given what we observe, I'm putting aside that knowledge. In Bayesian contexts, when examining conditional probabilities such as P(Y|T), we are making antecedent probability judgments, meaning we put aside our knowledge of the relevant data, including any evidence of creation and the existence of imperfect creatures. I'm also not raising an objection to theism that's completely unrelated to the anthropic argument. I'm directly attacking the central premise of the anthropic argument: that what theism, uncritically assumed, would lead us to expect, is a world populated by creatures like us.
Nevertheless, if I am being charitable, what BB means to say is the following: when considering the anthropic argument, we must hold certain data, k, in this case the existence of a universe that God created in our background knowledge. We then use k in conjunction with theism to predict the new data. Of course, this would not at all vindicate his "double-counting" charge, since failing to hold certain data in the background when evaluating what to expect from the theistic hypothesis, isn't anything close to double-counting. In any case, I concede that this move can be made in my original post, which is precisely why I introduced the duplication/godman objection. My argument concerning the imperfection of creation was to highlight that if we evaluate theism purely on antecedent grounds, without any presupposed knowledge of creation, we wouldn’t expect God to create at all, much less expect our specific existences. I never suggested that this precludes a theistic update if we assume God's act of creation as a given. Again, that is precisely what my godman/duplication objection aims to undermine.
When it comes to his accusation of double-counting in the context of the duplication objection we run into trouble, and it is hard to see how BB’s accusation could be charitably interpreted. That objection is directly about the probability that God would make a person like you, it is not something that can be held in the background when assessing the anthropic argument. True enough, it is a byproduct of my critique that the world we observe, with very unimpressive beings and blemished creation is very, very unlikely under theism, and thus that theism is probably false (at least if we only consider that specific evidence). But that's just what virtually ANY critique of the probability judgment that 'P(Y|T) is not hopelessly low' or 'P(Y|T) is higher than P(Y|N)' would entail. It certainly does not mean I am double-counting. The dialectic is simply that BB put forth an argument for theism, and I responded by directly challenging one of the premises of that argument—the assertion that 'P(Y|T) >> P(Y|N)'. We would only massively update in favor of theism if that assertion is true. This isn't at all illicit; in fact such antecedent analysis of the relative conditional probabilities is integral to how Bayesian reasoning operates.
BB's resistance to this line of critique, with his dogged insistence that we would just reject my objections conditional on the truth of theism, if taken seriously, would cripple pretty much any analysis of any Bayesian argument for or against theism, including his own. For instance, BB defends the anthropic argument, or the argument from psychophysical harmony. In laying it out, he makes some argument that, on naturalism, the data in question is very improbable. Now imagine a naturalist retorting, “Sorry BB. If your argument that the data is extremely unlikely under naturalism is correct, then naturalism would be, very probably, false. Or at least you’d have to be assuming a version of naturalism that is very probably false, since we very probably wouldn’t have psychophysical harmony. Therefore, under naturalism, we must reject your argument.” BB would, I think, immediately see the absurdity in such a response. Yet, his dismissal of my objections is parallel and mirrors this very absurdity.
Perhaps BB meant to suggest that the godman objection is double-counting because we should hold theistic answers to the problem of evil in the background as data. But, again, not only is this not double-counting for reasons I already made explicit, I dedicate an entire section to addressing that sort of objection in my original piece. I reject that there are any good theodicies, so I reject the demand to hold a successful theodicy in the background. Even if such a demand is accepted, I also reject, and have argued against in my original post the claim that holding theodicies in the background saves the anthropic argument from my objections. In other words, even if we update to a version of theism on which God has reasons for allowing evil,[6] God creating us is not explained. BB does not address this key point.
Perhaps BB thinks there is other data we can hold in the background that permits theism to predict the data of our specific existences without running afoul of my objections. But keep in mind such data would also be held in the background for naturalism. So if the background data is too specific it may also allow naturalism to predict your existence with a high probability. If, on the other hand, it's too general, then chances are it won't be enough to help theism predict our specific existences in light of my objections. It's BB's burden to identify data that, when held in the background, leads to a significant update in favor of theism over naturalism. A burden that I contend BB has not remotely met.
So I must conclude that BB's above response is a failure, the accusation even in it's most charitable interpretation simply does not land. As we've already seen there's another reoccurring theme in BB's reply, which goes something like this "even if X is really implausible, so long as X has a 5% or 10% probability, you should still massively update in favor of theism, with the version where X is true". We've seen that he says it in response to the concern that BB must assume certain ethical views. He also says it in response to my objection that God wouldn't create:
Furthermore—and this is a common theme—even if you think God probably wouldn’t create, as long as there’s a, say, 5% chance he’d create, your existence still massively favors theism. Thus, if one is at all uncertain about whether God would create, the anthropic data will still massively favor theism. I think it would be hard, on the basis of contested arguments, to get anywhere much below a 50% chance of God creating, meaning that the odds of your existence on theism is still much higher than on naturalism.And.. shocker, he also says it in response to my godman objection;
I know I’m repeating this a lot but it’s important—even if you think there’s only a, say, 20% chance he’d make us, your existence still favors theism because the odds of your existence is so minuscule on atheism.So what to make of this? Again, I take this to be a deeply unconvincing, and wrong-headed response. For starters, as a minor point, there's the potential circularity worry I raised earlier depending on how the objection is run (which, in an ironic twist, would imply that BB is the one double-counting data, not me). Further, I'll add that it's not the case that, the mere rejection of my arguments that God would create, say, alone would entail that God has a non-negligible probability of creating you specifically. It's being a 5% chance that my arguments for God not creating all fail, would not at all entail that there's a 5% chance that God would create us specifically, which is something BB appears to implicitly conflate. Perhaps what BB means to say is that there is some non-negligible, say 1%, probability that, all my arguments that God would not create fail, that my duplication objection fails, AND that God would be interested in creating all possible people with ultimately happy lives, including you in particular, in light of some background axiological view. That's what BB really needs to be true. But this brings us to our most crucial point:
A consistent problem with BB's arguments, not just here but elsewhere, is that he will assign an arbitrary credence, or evidential weight to a proposition P, and say "surely it is reasonable to think P is n% probable, right?" But he doesn't actually justify why such a probability assignment holds beyond brute seemings. This is a bad methodology. First off, and I'll get into this more much later, but this is just not how credences of this sort work. I take it credences are determined relative to a background web of beliefs and many complicated relations between those beliefs, you don't assign an arbitrary credence to some proposition (e.g "God would create all possible people") based on how plausible it strikes you in a vacuum. More importantly, in standard ways of testing a hypothesis with empirical data, you start with a distribution space and a null parameter for that space[7]; you don't assign an arbitrary credence to a random possible outcome of your model based on subjective intuition. If you did, the argument would basically just amount to: "I have a subjective intuition that your existence is astronomically unlikely unless God exists. Therefore, your existence is astronomically unlikely unless God exists." But that's a terrible argument. It would be ineffective to anyone who doesn’t share your subjective intuition, could in-principle be used for any crazy hypothesis you might conjure up, and wouldn't hold up in any serious statistical analysis or model selection.
Leaving aside that the probability BB assigns to God creating you is wholly unjustified. I'll grant that for some it may seem plausible, in a vacuum, that assigning a minimum 5% credence to God creating you is reasonable. But recall that the existence of you in particular, or so BB himself would claim, has a vanishingly small a priori probability. The idea, I take it, is that there are enormously many ways the world might contain no observers, and even more ways (higher-order infinitely many) for it to not contain you, a specific observer. So the scenario where you exist is astronomically unlikely. But this means, holding that God has a 20%, or 5%, or even a 0.0000001% chance of making a scenario with you specifically, would be to hold that, the truth of theism would significantly raise the expected probability of you existing by a factor of magnitudes of infinity. It would entail that God is much more favorable to actualizing scenarios where you in particular exist, relative to the infinite upon infinite possible scenarios where you don't exist.
This is not the modest claim BB imagines it to be. In fact, it is the very point in dispute. It is a claim that demands serious motivation, which BB, in levying this objection, has not even attempted to deliver. Indeed, my 'godman' objection, if successful, (which I unsurprisingly think it is) shows that for any act of creation God performs, it's very unlikely that God would create a creature like you, since He's much more likely to create a more impressive being, or a duplicate of one, which is a positive reason to think God would, in fact, not greatly favor actualizing scenario's with beings like you, much less you in particular, relative to infinitely other possible valuable states of affairs that do not include you specifically.
Ultimately, I suspect BB’s response amounts to little more than a rhetorical sleight of hand—a tactic equally available to naturalists. Imagine a naturalist claiming, "even if there’s only a .01% chance that there’s a multiverse where all physically possible outcomes (including that you exist) are realized, or that there's a naturalistic disposition weighted towards leading to your existence, you should massively update in favor of either the naturalistic disposition or the naturalistic multiverse that leads to you existing." BB would agree that this just abuses how credences work. There’s no good reason to believe, a priori, that the version of naturalism favoring your specific existence is any more probable than 1/n, where n represents the countless versions of naturalism that do not predict you specifically. Yet, again, the move BB makes here is parallel and equally unjustified. I've given forceful reasons to think that God is antecedently very unlikely to create beings like us, and this is on top of the fact that God creating a world with us specifically over any of infinitely other possible worlds is already intrinsically very unlikely given the sheer range of possible options available to God and the a priori improbability of our specific existences. My broader dialectical point is that there isn't a convincing reason, to think that the antecedent probability of God creating us in particular is over 1/n* where n* is any of infinitely many possible worlds without us God might have moral reason to instead create, or possible sets of people of equivalent cardinality that God might have reason to instead create. The one reason BB has given for thinking God would be specifically interested in creating us (because it's good to create a person with a happy life) actually just fails to justify the claim that God would specifically make us, in the face of all the objections I've raised in my initial piece.
To make what is essentially the same point another way, if our goal is to explain the specific data that we exist and if this data is vanishingly unlikely a priori, then we will need to update all hypotheses, not just theism, to the expanded versions of those hypotheses that do indeed explain the data. In theism's case, this means updating from bare theism to an extended 'theism+' that makes it likely that we will exist. What BB fails to establish—and what is crucial—is the claim that the ratio of the 'theism+' sub-hypothesis that makes our existence likely, to the total space of bare theism is significantly greater than the proportion of extended non-theistic sub-hypotheses that make our existence likely within the space of non-theism. This would include not only versions of naturalism such as the multiverse or stalking-horse naturalism, but various non-theistic spooky or supernatural explanations such as a limited non-divine creator, or a cosmic value-directed telos etc. And, again as I pointed out, the advocates of these hypotheses can make exactly the same rhetorical move BB makes: surely there’s at least a 1% chance that cosmic teleology would be directed to create us given background moral theories?
As a final note, this is all assuming an explanation of the data cries out here. Why insist that the demand for an 'ultimate explanation' of one's specific existence that goes beyond our immediate experience and observational data, let alone a theistic explanation, is even legitimate in the first place? I myself find such a demand both unnecessary and unconvincing[8]. Now, it might be said that we're just being Bayesians here and updating in light of the evidence. But, once again, using Bayes theorem and some funny math, one can trivially support virtually any hypothesis in light of some purported extremely improbable data, provided one adjusts their priors and likelihoods creatively enough. For ex. the probability that my foot lands in this very specific place, presumably described by some set of real numbers, rather than any other place is basically 0 unless an invisible fairy with a desire to put my foot there, guided it to that location, so I update in favor of the hypothesis that an invisible fairy guided my foot.
God and Imperfect Creation
Now we've addressed BB's most repeated objections in his article, and saw that they are devoid of merit. Fortunately, BB has direct objections to my arguments to fall back on. I'll address them one by one. With respect to my appeal to the possibility of persons God would not create, like R-Goblins, (entities that have inherently evil dispositions) BB says:I reject that there are such things as natures that constrain who a person is across worlds. Thus, I don’t think that R-goblins are conceivable. For any particular R-goblin in a world, it seems conceivable that they’d eventually come to have different dispositions. You are a soul, after all. And even views of personal identity that deny the existence of souls generally don’t hold that there are these robust evil-doing cross-world essences. So on nearly every view of personal identity, this is impossible.BB says R-goblins are inconceivable. This could mean a number of things given the many different notions of conceivability in the literature[9]. But I take it the relevant sense at play here, given that we are discussing the possible creatures God could create, is logical possibility. God is typically understood as being able to actualize any logically possible state of affairs. But there's nothing within the description of an R-goblin, at least as far as I can tell, that entails a contradiction. An R-goblin is just understood as a conscious agent whose nature is intrinsically morally evil. BB does not attempt to demonstrate a contradiction here beyond claiming it, which is a common trend of his reply.
Further, there are burden-shifting grounds that cut against the rejection of the logical possibility of R-goblins given theism. To start, the free will defense to the logical problem of evil as put forth by Alvin Plantinga, entails that, possibly, creaturely essences are transworld depraved. Further, God is orthodoxly conceived as a person that necessarily has a perfectly good nature. But if it's possible for a person to necessarily have a perfectly good nature, then, unless a relevant asymmetry is provided, we have solid grounds to think a person with a necessarily evil nature is also possible. Both are 'cross-world essences'.
There's a mention that soul theory and most views of personal identity deny that there are intrinsically evil natures. He doesn't explain why that is, but even if it's granted, it's irrelevant, as all that matters for my point is that creatures with such natures are logically possible.
Lastly, we can modify the concept of R-goblins in such a way that they are only contingently morally evil and have horrifically barbaric dispositions. It still looks pretty bad to create them, I think most people would agree with me here. So it's still a counterexample to the general thesis that it's always good to create a possible person that has a good life.
He goes on:
Furthermore, even if this is possible, it’s totally irrelevant. Perhaps if this is right not all possible people will be created, but all possible people with non-corrupt essences will be. But I don’t have a corrupt essence, and I don’t think you do either. So therefore my existence is very likely on theism and very unlikely on atheism. I’m not an R-goblin!However, this just misses the point of why I brought up R-goblins in the first place. The point is that the claim that God would create us is motivated by BB through the principle that it is always a good action to make possible people and give them good lives. But R-goblins are a counter-example to that. You can modify the principle to say that it is only good to create possible people of a certain sort. I acknowledge as much in my original post. But once you admit that there must be constraints on the kinds of people that are good to create, it becomes incumbent on the proponent of the anthropic argument to motivate the claim that humans like us are within the range of possible people that are all-things-considered worth creating, especially when it comes to an omniscient, omnipotent being who has lots of creative options to actualize instead of choosing to create us specifically. It has to be likely that a possible world with us is commensurably better than any possible world without us in particular.
BB also says:
TT notes that the “distance in moral impressiveness between us and God and the very best God could create is plausibly at least as great as the distance in moral impressiveness between us and R-goblins,” and claims that this implies that it would be wrong to make us instead of God men. But the things that make creating R-goblins bad is that their existence is bad, not that their existence is comparatively worse than other beings.But this is a distinction without a difference. I've cited examples, including those from my friend’s post[10], that show that there are multiple things that are positively bad about the decision to create humans instead of more impressive beings, indeed because there are many things about humans that are bad (cognitive defects, barbarism, selective altruism, lack of ability to experience empathy, poor moral reasoning etc.).
BB attacks my argument that God would not create, since the world sans creation is perfect. Saying:
God is the single maximally best thing but that doesn’t mean that a world with God alone is the best possible world. A world with God and other good things is better because it has extra good stuff. Having a single perfect thing in a world doesn’t make the world perfect—the world would be improved by adding more people.And:
But we have to distinguish maximal goodness in objects contained in a world from maximal goodness in a world. The world doesn’t have maximal goodness but it has a maximally good singular entity. It has the best single thing but it’s not a maximally good world because a world is improved by having more stuff good stuff. Compare: a architectural team where each architect designs one room, and there are five rooms total, might have maximal goodness in one room, but not maximal goodness overall if there’s one perfect architect and four bad ones.So, apparently BB glossed over the part in my original piece where I pre-empt and address this very point. I'll just quote myself.
It may be argued that while God is maximally good or perfect qua being and thus unimprovable, this does not mean a world with God alone is perfect, or maximally good qua world and thus unimprovable. This is the most common objection, and I must admit I find it perplexing. If the world, W1 we are talking about is a world where God exists sans creation, then W1 is coextensive with God, the very same object is being picked out when I refer to W1 as when I refer to God. So, I must ask the rhetorical question, what sense can be made of the claim that W1 can be improved, and yet God cannot? One might be tempted to argue that the possible world where God creates, W2, contains a diversity of beings whereas W1 does not, and diversity of beings is good qua world, and so that is what is meant by good qua world. However, as already argued, since God exemplifies maximal goodness and absolute perfection qua being, the existence of other good beings would not increase the goodness of the world where maximal goodness of being already exists—a point which is further compounded by the fact that things are good only to the extent that they resemble God.
The point being that there isn't actually a real distinction to be made between a world and all the stuff exemplified in that world. A world isn't an ontological item that is over and above all that is instantiated in it, it just is a collection of objects, properties, relations and true propositions regarding them. If a world W1 contains only God, then referring to W1 is referring to God, since they're co-extensive. So to refer to W1 and say of it that it can be improved, is referring to the state of affairs where God exists, and saying of it that it can be improved. It is to refer to the being God and say of Him that He can be improved.Put another way: God is said to be a maximally good being, which means that He is the most excellent thing that could possibly exist[11]. But then if you say that it would be better if lesser goods exist, then you'd be saying that God is not the most excellent thing that could exist, God + lesser goods is. It would be to say that a better being than God prior to creation, would be a being that entails lesser goods, or has the relational property of existing alongside lesser good beings. But at least on my concept, the most perfect being would be a being that, alone, is the best thing that could exist, not dependent on lesser goods.
As for the architect analogy. I don't know if I even understand it. I guess the idea is that you can have a building with one perfectly good room, and 4 other bad rooms and so it wouldn't be a perfect building. But I agree with that. A perfect building would be a building for which all its rooms are perfect. There is a distinction between the perfection of a room and the perfection of the building as a whole. Where this analogy falls short is that we are talking about a world which contains a maximally good being and nothing else and asking if it can be improved. So, there's nothing else that exists to make the world less than perfect. The building has shitty rooms that make it less than perfect. Further, we aren't talking about a perfect sortal object like a building, or an island, we are talking about a perfect being, the most perfect thing that in general could exist, not the most perfect of a certain kind of object.
BB uses an analogy, from John Buck to motivate the claim that God would create:
Here’s another analogy (credit to John Buck for it). Consider the anti-god, the most terrible possible entity. Maximally evil, pernicious, and powerful. Would a being like that create? Of course—it would create huge numbers of people to torture. Yet one could reason in the same way: the world already has the maximally evil thing, so it can’t be made worse. Any argument for why this is wrong will explain why a good God would create.This is one of the most forceful objections made in BB's response, in the sense that it has serious rhetorical bite and is not obviously wrong. I take it the argument is something like the following:
1. If the "anti-God" would create people to torture, then God would create people and give them good lives.
2. The "anti-God" would create people to torture.
3. Therefore God would create people and give them good lives.
This argument is a valid instance of modus ponens, so the conclusion can be denied only on pain of denying one of the premises. What should we make of this? The problem is, depending on how we understand the concept of the "anti-God", either 1 is false, or 2 is false. But no understanding renders both premises true.
There are two ways we can understand the "anti-God". One way is that they are a being that exemplifies the maximal amount of axiological badness possible. They are a maximally bad being. In this case, 1 looks like it's at least on firm grounds, though it can be denied if one accepts that there is an axiological asymmetry. But suppose we grant that there is no axiological asymmetry. If a maximally bad being would create more intrinsically bad stuff, then absent any relevant asymmetry, for the same reasons, a maximally good being would plausibly create more intrinsically good stuff.
The problem is then with premise 2. If we take seriously the concept of a maximally bad being, then they would instantiate not only a maximally intrinsically evil character, but all possible evils to the worst degree imaginable, including the most horrific and maximized agony and despair possible. But that means creating more beings to suffer won't add to the world's disvalue where the anti-God already exists, since, by definition, the worst possible, and maximized despair, agony and general badness is already instantiated and so cannot be increased by the addition of more suffering beings. In fact, making conscious beings to horrifically torture would add more value to the world, since those beings would probably have some intrinsic moral value, even if their lives are as shitty as can be. So it would increase the world's value but not increase its badness, and an anti-God wouldn't do that. I concede that it seems very intuitive prima facie that a maximally bad being would create people and torture them, but that's because the notion of a maximally bad being is a truly bizarre one that is hard to conceptualize. Once we take seriously what the existence of a maximally bad being actually entails, the intuition dissipates.
What if we just take the "anti-God" to just be an all-powerful and maximally morally evil being that doesn't necessarily instantiate maximal axiological badness. In that case, 2 looks pretty plausible, such a being would create people and maximize their despair and agony. But 1 is false, because the anti-God and God are not analogous. The anti-God can make the world worse by torturing people, God can't improve the world because He already instantiates absolute perfection and maximal goodness of being. So, all and all, the argument is interesting and challenging but, I think, ultimately unsound.
I'll also note that even if the argument is sound, while it would undermine my arguments against divine creation, the analogy would actually provide motivation to my godman/duplication objection. A maximally bad or evil being would plausibly only create beings that instantiate the worst horrors imaginable and duplicates of them. So too, a maximally good being would plausibly only create beings that instantiate the best and highest person-centric goods, and duplicates of them.
I mention in my original piece that a popular theistic meta-ethic is that goodness is constituted by resemblance to God, and that even if you don't accept that view, it's a commitment of perfect being theism that things are good to the extent that they resemble God even if goodness is not identical to resemblance to God’s nature. BB says:
I don’t find this view at all plausible. Even the second more limited view that is “plausibly a commitment of perfect being theism,” isn’t plausible—a spirit version of me resembles God more in that he’s immaterial but is no better. Furthermore, even if goodness=resemblance to God, the world with more things that resemble God is better, having more total goodness.But the reason God is taken to be immaterial is that immateriality is a perfection of being, since He's not limited to any spatial location or material substrate. So the immaterial version of BB would have a perfection that the embodied version lacks on the theistic view. I don't understand why BB would deny that.
I also can't make sense of the latter point. Do I resemble myself less than the state of affairs containing myself and some other guys that resemble me? No, that makes no sense, I am myself, so nothing could resemble me better than myself. So too, the state of affairs containing God and some stuff that resembles God doesn't resemble God more, which means it isn't more good since what goodness is on the meta-ethical view outlined, is constituted by resemblance to God.
In any case, I gave an argument in my post to justify the claim that the weaker view is a commitment of perfect being theism, to which BB says the following;
But such a being is metaphysically impossible. Goodness is the way it is necessarily—it can’t be contingent in that way.So his response to my argument that the greatest possible being would be a being for which all good things stand in resemblance to said being is just to baldly assert that such a being is impossible and give no argument for that. Convincing. I hadn't considered that.
I also don't understand how the second sentence flows from the first in a way that's relevant.
BB misapplies the double-counting charge again in response to my objection to the claim that God would create people for the sake of their good, where I note that it would be made false if there was an axiological asymmetry. I've already addressed that charge and won't repeat myself further. I'll also note that I do more than simply cite the axiological asymmetry as a response to the relevant objection, and the axiological asymmetry wasn't the most crucial point there.
He next responds to my second argument that God would not create, if creation involves any blemishes or imperfections. Saying, of the condition I provide:
I accept this. God would only create imperfections if they serve greater goods. But then if one thinks there’s a solution to the problem of evil they’ll think the world’s various imperfections serve greater goods. Obviously if the problem of evil kills theism then the argument wouldn’t work—but this is just double counting the problem of evil. Because theism assumes that the problem of evil doesn’t work—otherwise it would have to be false—pointing to the problem of evil doesn’t affect the relative evidential force of the argument.He repeats the 'double-counting' charge yet again, a charge I've addressed extensively both in an entire section of my original piece, (which BB doesn't really engage with) and above. So you can see why I dedicated a lot of space to that objection. Since he keeps repeating it, he clearly seems to think it's the best thing since sliced bread. However, as I've already pointed out, it's simply misguided.
In any case, the most important thing to note here is that he misunderstands the condition I laid out. It doesn't just say He'd only create evils or blemishes if they serve a greater good. It says He'd only create evils or blemishes if, for every greater good that entails those evils or blemishes, it is not the case that there's an even greater good which does not entail them. For reference here's the condition:
A morally perfect person M will actualize some state of affairs S that contains or entails some moral imperfection or evil E only if S is a greater good and it's not the case that there is a possible state of affairs S* which is a yet greater good M could choose to actualize instead of S, and S* does not contain or entail E.
The reason this matters is because, as a result of him not understanding what he has accepted, he gives a response that misses the point:
TT says that this implies God wouldn’t make imperfect creatures. But perhaps making temporarily imperfect creatures serves a greater good—say, providing free will, or putting us in a broadly indifferent universe to strengthen our relationship with God. Remember, this is just the claim that people make when they deny the problem of evil, so if TT’s objection assumes that the POE works, it will already assume theism is false, and thus not give an additional reason to reject the argument.
Now, this condition entails that God would not create morally imperfect or sinful creatures. This is because even if the creation of morally imperfect creatures is an overall good, it still entails moral blemishes or evils E, and per our condition, a morally perfect being would only permit E if E entails a greater good, and there is no even greater good that does not entail E. However, we can see that for any overall good that entails E, there is a greater good which does not entail E. God Himself is such a greater good, indeed, He is the greatest possible good, and God does not entail E, or any evils.
So BB misses the point. He could of course reject the condition, but then he has to address the motivations I've given for it, which he didn't do. He also says this:
However, I think one can reasonably reject the principle, though I lean towards it. Suppose there are an infinite range of options each better than the last. It wouldn’t be wrong for a perfect being to create one of them, even though there are better options. At some point, one simply needs to pickThis also misses the point. The point is not that God must pick the best possible option out of a range of options, it's that He wouldn't create blemishes or imperfections if they don't satisfy a certain condition.
Finally, he addresses my argument that God would not create from divine impassibility:
But it can be for the sake of someone that will exist. Just as it would be wrong for God to create a miserable person because it would harm them, it would be right for God to create a happy person because it would benefit them (see here for elaboration on the point). TT says that God in his unchanging nature can’t create without it affecting him and thus that creation conflicts with impassability. This isn’t a topic I’ve thought about at all, but worst case scenario, that just means you shouldn’t accept divine impassibility (which I already don’t). Obviously if divine impassability is incompatible with God creating then theists shouldn’t accept divine impassability, so this if the argument is right, it just gives a reason to reject divine impassability.So basically his response is to get rid of divine impassibility. I actually think that's fine, not all theists accept divine impassibility. So this is probably the best response that could be given. But many if not most orthodox theological traditions accept that God is impassible, and there are many arguments to the effect that a perfect being would be impassible[12]. So if BB is forced to move to a controversial view of theism that rejects impassibility, that's at least a cost. Antecedently, (prior to considering the data), the initial space of theism where God is impassible at least takes up a decent chunk. So it remains the case that the appeal to divine impassibility, in conjunction with my other two arguments, provide forceful considerations that count against divine creation.
The Duplication Objection
So those were BB's responses to my arguments that God would not create. I conclude that he has failed to significantly take out much of the bite of those arguments. But we're not done here yet, he also replies to my godman objection. Which, if you recall, is the objection that there are much more impressive beings that God could have created instead of humans like us, so we shouldn't expect God to create us in particular. So what does he have to say there? Well, his first two objections are his most repeated ones, and I've already explained why those are without purchase. His third objection is this:I think it’s plausible that God does make creatures like that. We are Godmen in the next life. However, as argued by the theodicy I’ve given, we have limited goodness and power temporarily in this life because it strengthens our eventual relationship with God and might achieve various other great goods
But perhaps that's not all there is to the story. Perhaps God creates unimpressive beings like us so that we may, in the afterlife, go through some deification process to become more impressive beings like godmen eternally, and it is necessary that we start out as humans in this process. But this is a very bad explanation and doesn't satisfy both (i) & (ii). If God has the choice between creating a human that becomes a godman at some point in the future or in heaven, or creating a godman from the outset, we should expect God to create a godman from the outset. This is because for any time t, it's better to be a godman at t than a human at t, so God has more reason to create a godman at t than a human at t. So prima facie at least, it looks like (i) isn't satisfied. Perhaps what's being said is that one could tell some unmotivated but logically coherent just-so story about how the most valuable world, logically necessarily, must include humans at some point in addition to godmen or that humans evolve into godmen, and so God would create that world. Certainly such a story could be told, but there are numerous other logically coherent just-so stories one can tell which rule out God creating humans, or at least rule out a world with humans at this point in time. We can tell the story that God would create a world where only cheese exists at this point if we were to observe an only-cheese-world, and that the cheese eventually evolves into godmen in line with God's ultimate purpose. Why not? That's no less antecedently plausible by my estimation. But if we can just attach any unmotivated just-so story to the theistic hypothesis, it becomes empirically vacuous and we would have no reason to think any particular story, including the one where God creates humans, is true, unless we explicitly use the fact that we exist to update on what God would do, but then we couldn't also use that data as evidence for theism. In which case (ii) fails to be satisfied. Therefore, until such time as BB provides a plausible story that satisfies (i) and (ii), this consideration provides absolutely no evidential counter to my argument.
An additional note is also worth making here. If we spend a tiny portion of our existence as humans on earth but the vast majority of it as much more impressive beings in heaven. Then this implies that, for any time t that we observe ourselves persisting, we should expect to be more impressive beings in heaven at t, since there are infinitely many times we spend as impressive beings in heaven, and a proportionally 0 amount of time spent as humans on earth according to this theodicy. So, our observation that we are not deified in heaven and are instead human beings on earth would have an epistemic probability of approximately 0 on this picture. So even if this soul-making+ theodicy did explain why God creates humans like us, it would also entail that the probability of our observations given theism and this theodicy would be pretty much 0.
Yikes! That's not a good start! If that wasn’t enough, I will even more explicitly address his particular theodicy at-length later in this post, to satiate my anti-theodicy appetite. So, we're 0 for 3 so far. But there's more. His fourth response goes as follows:If this is right, then God could never create because for each creature he could create there’s some other better creature. As a result, he just has to pick some creatures to create, even if they’re not the best creatures he could create. A morally perfect person is perfectly justified in creating a very good state of affairs, even if there’s a better one, as long as there’s an infinite continuum of options of increasing goodness.He says this again later too, with regards to my duplication point:
Let’s call Godmen+ the beings better than Godmen and Godmen++ the beings better than Godmen+. By this standard he’d never create anything, because for anything he could instead create a better being.Several glaring issues stick out. The first problem with this response is that it misconstrues the argument. The argument is not that God would create the best possible creatures. It's that there are certain constraints on the creatures God would create, and we don't fall under those constraints because there are many things about humans that are actually pretty bad relative to what could be, such as our really bad moral dispositions and reasoning, our cognitive deficiencies, our constitutive inability to access or appreciate the deepest virtuous and valuable states etc. It's consistent with the argument that God does not have to create the best creatures, the point is that He wouldn't create ones as bad as humans, and would instead, if He creates at all, create far better ones.
Perhaps the idea is that while there are bad things about humans, since God can't create the best beings anyways, it's good enough for Him to create humans. But this is untenable. For one, it's morally absurd. Imagine I kill someone and justify it by saying "Sure, killing that guy was not the morally best thing I could do, but it's not the worst either and it's good enough. I can't do the morally best thing anyways". For two, it robs theism of empirical content. For any state of affairs we observe, no matter how bad, we can always say that God would create it because He can't do the best thing anyways and it might have been worse. But then theism wouldn't rule out any observable state of affairs, it would be equally consistent with any observation we could have, and would thereby be void of empirical content. That's bad because it would mean we can't test theism with any observable data, so it would be off the table as an explanation for the anthropic data. So, there clearly has to be some constraints on what God would create. The question is whether humans are within them, and I've argued they aren't.
The second problem is that it's false. The kinds of things that fix the goodness of creatures do have an intrinsic maxima. If God creates beings that are, like Him, perfectly good, then He can't make better ones. Indeed, Mark Walker (2009) in his paper on the 'anthropic argument against theism' argues that God would make ontological equivalents to Himself. A claim this strong is also not necessary, since the whole point of my friend’s more modest take on Walker’s argument was to suggest that there are certain valuable states that have a deepest qualitative level or finite set of aspects that can be instantiated, and that we fail to meet these criteria. Indeed, a large part of his post is intended precisely to address the ‘no best option’ objection BB is raising (and he told me that this is made even more explicit in the new version of that post that is pending). As a result, BB’s objection simply does not apply to a construal of godmen consistent with the very arguments I cited in my original post.
The last problem is that even if it's true, it doesn't help, in fact it makes things worse. Suppose it's true that for every being God could create, there's a better one. In that case, we can argue that God does not exist as follows[13].
1. For every possible being B, there's a morally and axiologically better possible being B*.
2. If God exists then God is morally perfect and omnipotent
3. If God is morally perfect, He would have the most impressive moral resume possible.
4. If a moral resume R1 is otherwise equivalent to another moral resume R2, but R1 includes an act of creating an axiologically and morally better being than a corresponding act of creation in R2, then R1 is a more impressive moral resume than R2.
5. If 1, 2, 3, and 4 then if God exist then for any possible being B, God would not actualize B (Since there’s a better being B* He could have actualized instead)
6. Therefore, if God exists then for any possible being B, God would not actualize B. (from 1-5)
7. If God exists, then it's not the case that there exists a being that God would actualize (from 6)
8. If 7, then if God exists, it's not the case that beings other than God exist
9. However, beings other than God do exist.
10. Therefore, God does not exist. (from 7, 8 and 9)
1 is the assumption for reductio, 2 is definitional, 3 is definitional, 5, just express a conceptual entailment from the previous 4 premises, 8 follows from God being the sole entity prior to creation, and 9 is uncontroversially true. That leaves 4, but 4 is (and I know that BB loves his intuitions) intuitively obvious! It's also a plausible conceptual truth, what it means for a moral resume to be more impressive than another moral resume is that the set of actions it involves are better all-things-considered. So the reasoning looks pretty water-tight given we accept 1, by my lights.
So that objection fails. This leaves us 0 for 4, and with one more objection to examine. It's another one I pre-empt and address, the objection being that God creates humans in addition to 'godmen'. But this time BB actually does address my responses. So let's see, he addresses my point that it's not at all obvious that God would create a world that maximizes happy people instead of a world which instantiates the best kinds of goods:
But this violates the Pareto principle. It’s good to create humans, even if there are already Godmen, because it makes the humans better off and no one worse off. To see this, suppose you found out that there were Godmen a galaxy away. Would that make it less good to have kids, because you dilute the goodness of the Godmen universe? No, of course not! The goodness of creating a person has nothing to do with the presence of other unaffected people. And all the arguments for creating happy people apply even if there are Godmen. So TT is just repeating over and over again that you might be an anti-natalism of some sort, or an anti-natalist in world with Godmen. It’s easy to find a lot of different motivations for anti-natalism but one must argue for their plausibility rather than just state that they’re a conceivable view one might hold.BB appeals to the Pareto Principle, saying that creating humans makes humans better off and nobody worse off. But that's just false. Since humans in general, once again, have bad dispositions (cognitive defects, barbarism, selective altruism, lack of ability to experience empathy, poor moral reasoning etc.) there's a virtual guarantee that a world with humans would make some people worse off than the world with just godmen. BB has offered no good reason to think having lots of humans in addition to godmen is an overall Pareto improvement to having just godmen even if there are more total people. Further, even if no one is directly worse off, the average and overall quality of life, experiences and the deepest types of goods realized may in some important sense be lesser in the world that has humans in addition to godmen. The pareto principle does not entail that more people being well-off and no one being worse off is the only consideration when it comes to God's choice in what world to actualize, it's a consideration but not the only one. Ultimately, BB just once again assumes, without any argument, that sheer number of people who are well-off determines the value of worlds. But that's the very thing in dispute. In any case, even if we make that assumption, my duplication objection takes care of that, since given that argument God will always simply make another duplicate godman rather than instantiate a human.
I'll also note that the appeal to the pareto principle in cases where the people don't exist yet is dubious. The pareto principle typically applies to changes affecting existing individuals, where an action is considered good if it makes at least one existing person better off without making anyone else worse off. To-be-created people aren't current stakeholders whose well-being we can measure or compare.
BB is also wrong that my objection is an appeal to anti-natalism. The options aren't 1) creating possible happy people is always all-things-considered good or 2) creating people is always all-things-considered bad. There's actually a third option, 3) creating possible happy people is sometimes all-things-considered good, and sometimes not. That's what I'm invoking.
BB challenges my argument that godmen can be duplicated without end, and so for any person God creates, He could create a duplicate instead, which He has better reason to do. His first objection was one I already refuted. But he has two more. Here’s his second objection:
This assumes that one can keep cloning people indefinitely. But this is contentious. If every person is their own unique soul, he can’t create new people because there is no extra uncreated person who could be created (all the possible people exist). This will also follow if one accepts Willliams’ necessitism.This is once again an objection I pre-emptively address in my original post. I directly argue that no matter what one's view of 'souls' is, duplicates are logically possible. BB says more later, but unfortunately, he doesn't meaningfully engage with the argument. His third objection is this:
This argument assumes that creating Godmen trades off with creating people. But he could create both. He could make as many Godmen as he once and people. It’s at least not terribly improbable that he’d create beings of all good kinds in infinite numbers—if there’s even a 1% chance of that, say it with me now, that massively favors theism.I'm not entirely sure what is meant by "This argument assumes that creating Godmen trades off with creating people." If it means that I'm assuming God can't create both godmen and human beings, then that's false, I don't assume that. God could create humans and godmen, obviously. My argument entails that God would create godmen and duplicates instead of humans, not that God can’t create humans. Maybe he means that I'm assuming that, for every act of creation, God could create a duplicate of a godman instead of creating a human. But I don't just assume that, I argue in my original piece that at no point does God have to worry about running out of duplicates to create. God can make duplicates by simply endlessly creating duplicate universes with all the same godmen, where the only difference is the name he gives each duplicate universe. That's logically possible, unless BB can show that there is a contradiction entailed. The 'even if it's 1%' thing is just a straightforward abuse of credences, which I've already addressed.
In any case, BB does attempt to prove that the duplication is impossible:
TT says “Since there's no contradiction in doing so, it seems to me there isn't an upper bound on the universe's God can create and the duplicates He can put in them.” But the scenario is impossible—every possible person has been created and yet new people are being created. A person is being created who is not a possible person. But clearly impossible people can’t be created!This misses the point. The argument is not that God can create all possible people, and then create more. The argument is that for every possible person God creates, He could instead create a duplicate of a godman, and He would do that because it's the morally better action. I give a positive reason to think that, possibly, God could always create more duplicates instead of humans regardless of what view you take on souls:
To spell out the contradiction:
1. In the scenario God has created every possible person.
2. In the scenario after this God creates new persons.
3. For God to create a new person they would have to be possible.
4. So God would have to create people that he’s already created.
5. If a person is created they can’t be created again.
Either we accept that it is logically impossible for a duplicate to have the same soul as an original, in virtue of it being entailed from the concept of a soul that it is singular and cannot animate multiple beings across universes, or we don't. If the former, all this means is that if God were to make duplicates they couldn't have the same soul, but this does not entail that God cannot make qualitatively identical duplicates and put them in different universes. It just means that if there are duplicates in different universes, then this entails that those duplicates must have different souls since the same soul can't exist in two universes at once. But, unless demonstrated, there's no logical contradiction in God creating duplicates in different universes. If the latter, and duplicates could have the same soul despite being in different universes, then God could create endlessly many duplicates inhabiting different universes that share the same soul since, again, there isn't a contradiction entailed from doing so. Either way, this objection does not undermine the possibility that an omnipotent deity could create godman duplicates in different universes, nor does it diminish the eminently plausible notion that creating said duplicates would always be better than creating us. The objection therefore fails.
BB simply ignores this argument. But even if my argument was that ‘God can create more than all possible people’, his response is no good. First, the structure of the argument he gives is formally invalid. Second, and more importantly, he assumes that there is an intrinsic maxima to the possible people God can create. But depending on how broad "possible person" is understood, it looks like for any amount of possible people God creates, He could create more. Suppose you have a possible person P, and then create a copy of that person P* where all the same things are true about them except that they exist in a different universe. Is that a different possible person? If so, then God could continue to make copies of P, P*, Pn, Pn+1 and put them in different universes without any end, since there's no contradiction. So 1. is false, there's no scenario where God creates every possible person. He could always make more. If on the other hand, the copy is not a different possible person, then fine, 2. would be false, but that doesn't mean God couldn't create such a duplicate, it just means that they wouldn't be a new person.
BB finishes off by addressing my point that he assumes soul theory by pointing out that Haecceitism would work too[14]. Fair enough, I'm willing to concede that much. It still assumes some disjunction of controversial views. In any case, that wasn't the main thrust of my reply and I don't think it matters that much. He then repeats the credence-abuse I addressed earlier, and I need not repeat myself. I'll only note that it's false that 'soul theory' or other adjacent views of personal identity alone would do much to raise the expectation of your existence on theism, it's just one of the underlying assumptions BB makes in his argument.
So, while some of BB's objections to my arguments against divine creation at least might have something to them, even if I don't find them ultimately convincing. His responses to my godman/duplication objection are all abject failures.
Second Recap
Here's a recap of the last few sections.
- BB charges me with double counting in levying various objections against the claim that God would create. This misunderstands both what double-counting is, and how Bayesian reasoning works. There is at least a charitable way to understand what BB might have meant with respect to my ‘God and creation’ objection–but it’s one that I pre-empted in my original piece.
- BB repeats throughout the article that "even if you have an n% credence that God would create you, then your existence favors theism". But that's a very bad approach to hypothesis testing, it's just a rhetorical move that repeats the very claim in contention while dressing it up in modest-seeming garb. But it's not modest, and my arguments provide grounds to reject such a claim. It also might actually involve double-counting data, which is pretty ironic, considering he accuses me of doing that throughout his article.
- BB responds to my appeal to 'R-Goblins' as counter-examples to the claim that God would create all possible people, by baldly asserting that R-Goblins are not possible. That would be bad enough, but there's also grounds to think R-goblins are in fact possible given theism. His second response to my R-goblin counter-example misses the point.
- BBs first objection to my first argument against divine creation, ignores my pre-emptive reply in my initial piece. His second objection is an interesting argument from John Buck, but I contend that no understanding of "anti-God" renders both premises true. His third response makes no sense, and his last response is just a bald assertion.
- BBs response to my second argument that God would not create misses the point. His response to my final argument that God would not create is to give up divine impassibility, which is fine, but still a cost.
- BBs first two objections to my godman objection just repeats his double-counting charge and his credence misuse. His third ignores my pre-emptive response in my initial piece. His fourth misconstrues the point, is false, and would only be a disaster for theism if true.
- He addresses my duplication objection. But once again his responses are a combination of toothless claims, missing the point, and repeating the same bad objections I covered earlier.
Stalking Horse, Theoretical Virtue and BB’s Credence-Abuse Again
Let's just get right into it. The first problem with BB's reply here is the initial framing. I put forth a stalking horse naturalism, which I labeled Nd, on which we have an initial disposition which entails or is weighted towards your eventual existence. BB says the following about it:This is not correct. I do explicitly address the concern that Nd has a prior of zero. Nor is the point I make just 'such and such argument that theism is false'. What it is is that, the reason for thinking Nd has an extremely low prior (0 or arbitrarily close to 0) is a reason to think, the version of theism which entails your existence also has a correspondingly low prior. Just as naturalism has to build in a disposition which entails your existence, theism has to posit an intention of God (and no outweighing desires or intentions) which entails your existence. The naturalist has no reason to grant that building in such an intention to God is any less unparsimonious. Here is what I say:The problem is that this has a prior probability of zero. Because you’re not special, a theory where the universe picks you out has no higher of a prior than a theory that it picks out any of the other infinite possible people, so it has a prior of zero.TT doesn’t address this worry. The rest of his section is just arguing that theism is improbable. Now, you can read my response to that if you want, but that’s not a response to the anthropic argument—it’s just raising a bunch of other arguments against theism.
There might be other reasons to prefer the theistic hypothesis to the stalking horse naturalistic hypothesis, however. Perhaps the stalking horse hypothesis starts out with an astronomically low prior probability—significantly lower than the theistic hypothesis. But why think this? Perhaps the idea is that we start out giving a roughly equal credence to the bare theistic hypothesis and the bare naturalistic hypothesis. But stalking horse naturalism is a very specific sub-hypothesis of the bare naturalistic hypothesis, one which takes up a very small initial portion of the probability space. There are infinitely many equally a priori likely conflicting naturalistic sub-hypotheses that do not posit the existence of an initial state causally disposed to eventually produce you in particular. So the stalking horse naturalistic hypothesis on offer, while also being ad hoc and transparently constructed to accommodate the data, has an extremely low epistemic probability, if not 0. Something like this is plausibly right. However, a parallel problem afflicts the theistic hypothesis.
Suppose we take the bare theistic hypothesis T to be the bare proposition that there exists a supernatural agent at the foundation of reality. The bare version of T does not at all make probable the data Y of your existence. For starters, T does not specify the causal powers the supernatural agent has. For all we know, they might not have sufficient causal power to create a universe with you, let alone a vast multiverse that contains all possible people. More importantly, T does not specify the contents of the supernatural agent's desires. For all we know, the supernatural agent might not desire a universe with any people at all, let alone you. So, for this reason, T, when understood as the bare theistic hypothesis is just as uninformative with respect to Y as N. For T to predict your existence, we must appeal to a sub-hypothesis within T, call it Ti, which says that the supernatural agent at the foundation of reality has sufficient causal power to create a universe with all possible people, and that they have the desire to do so, and no outweighing reasons or desires against it. But there are infinitely many versions of T that conflict with Ti which are a priori equally probable. For example, you can imagine versions of T where the supernatural agent only desires to create an empty aesthetically beautiful universe (of which there are plausibly infinitely many), or any universe with lots of people but not you (again of which there are plausibly infinitely many). So, just as Nd takes up a small portion of the initial probability space of N, the same looks true for Ti vis-a-vis T. To put the point another way, since Ti entails T in virtue of being a subset of T, it follows that P(Ti|Y) is less than or equal to P(T|Y). Further, since we've seen that T in virtue of it's lack of specificity does not predict Y any better than N does, we can reasonably infer that P(Y|N) = P(Y|T). From this it follows that P(Ti|Y) must be less than or equal to P(N|Y).
At no point does BB acknowledge, let alone address this argument. Now, later on I address the possible objection that one can appeal to a purported 'simple' and 'elegant' characterization of theism, namely perfect being theism, and that this theism predicts the data with high probability. I argue that this doesn't work, both because there's no good reason to think such a theism predicts the data with a substantially higher probability than naturalism and plenty of reasons to think it doesn't, given what I argued in the previous section. Additionally, because, on top of the fact that naturalists have plenty of reasons to prefer naturalistic explanations to theistic ones. Insofar as the version of theism being appealed to entails a complex set of desires, and powers relative to any other set of desires and powers, we have powerful reason to think such a theism has a prior approximating 0, by parity of any reasoning that Nd has such a prior.Most of what BB has to say related to stalking horse naturalism is defending the intrinsic probability of perfect being theism. But I want to stress that even if one thinks his defense is convincing, he still hasn't fully engaged with my broader dialectical point. I'll note right away, that if it were the case that it were plausible that perfect being theism predicts our specific existence, without any further assumptions or built-in intentions, then I'm happy to admit that the stalking-horse objection would lose much of it's dialectical bite (which isn't to say I wouldn't still think it's cogent, just that it would be more of a dialectical uphill battle). However, as argued previously, that's not plausible.
I make the point that theism is as mysterious as Nd, arguing that, insofar as a naturalistic disposition weighted towards your existence is mysterious, so too is it mysterious for God to be motivated by certain moral reasons that makes your particular existence likely. BB replies:
He’s maximally motivated by moral reasons. He brings about the best possible world or, barring that, some very good world. He’d create either all possible people or some huge number of people. What other moral reasons is TT talking about? God would do the best thing or some very good thing if there is no singular best thing.Suppose we grant (as I would not, and have argued against) that an entity being maximally motivated by moral reasons predicts with high probability that we would exist. There are many other possible constitutions of a divine designer's desires that do not involve being maximally motivated by moral reasons. Maybe they, like virtually all rational agents we know, are only sometimes motivated by moral reasons and sometimes motivated by strictly prudential or self-interested reasons. Maybe they're an egoist and only motivated by self-interested reasons. Maybe they're motivated to be maximally evil. The point is, why are we so lucky that the divine beings' motivations are constituted that way, relative to any other way? Just as a naturalistic initial state being so causally disposed to bring about us seems mysterious, so too does that seem mysterious.
Still on the topic of the mysteriousness of theism, I also point out that theism is just as uninformative with respect to the causal mechanism at play. BB replies:
There’s no extra mechanism. God has the property of omnipotence meaning that which he wills happens. Once there is a being like God, there’s nothing extra needed for him to be able to create.This misses the point. I never said theism involves an 'extra mechanism' that's needed for God to be able to create. I said that theism is uninformative with respect to the causal mechanism that serves as a causal explanation for the data, and the relation between God's willing some X, and X obtaining. I do not deny that God can make X obtain, the question is 'how' He does that. Typically in causal explanations we can specify the interactions involved and the laws or regularities governing those interactions. Theism is fundamentally mysterious in that regard. Maybe BB’s point is that it is not even cogent to talk about the ‘mechanism’ of God’s creation. But if so, that only reinforces my point: overwhelmingly, explanations involve mechanisms. If it is claimed that a particular cause lacks one entirely, that is both mysterious and, I take it, ought immediately reduce our prior for that explanation. He also adds:
TT talks about things in terms of mystery. But that’s not the way to do it. Think about things in terms of conditional probabilities—the odds of theism start out decently high and theism predicts that you exist which has a very low prior. Applying Bayes theorem, we get very strong evidence for theism.This also misses the point. I was not implying that we think about things in terms of mystery instead of conditional probabilities, (as if those are even in competition) I was addressing a possible objection to Nd, which is that it is mysterious, by pointing out that theism is just as mysterious. Furthermore, informativeness is often taken to be a theoretical virtue[15], so mystery is often viewed as a theoretical vice. Theoretical virtues play at least some role in even the strictest Bayesian accounts of reasoning, so appealing to them is most certainly not missing the point[16]. That said I of course also deny that theism gets any boost over naturalism when thinking about things purely in terms of priors and likelihoods. But BB should know this since that's what takes up the majority of my discussion.
So this leads us into our discussion on the priors. BB writes:
Suppose you think there’s a 1% chance that something like the Godellian ontological argument or the contingency argument works. Well then the prior in theism can’t be too far south of 1%—before looking at the evidence one should think there’s a decent likelihood that God exists necessarily.So, we turn again to one of the many reoccurring strategies BB employs in his response to me, this one being the credence abuse. To start, I'll point out that Godelian ontological arguments and the best versions of contingency arguments are not in the same league. Contingency arguments are a serious consideration, Godelian ontological arguments (GOAs), I take it, just aren't. Myself and others have pointed out that GOAs are utterly dialectically inert, lacking any real teeth at all[17]. I won't repeat the point here. Beyond that, they're subject to parodies, here's a fun one.
Call a U-perfection a perfection of unicorns. Here's an argument for a necessarily existing unicorn. Here are our axioms.
A1 A property is a U-perfection only if its negation is not a U-perfection.
A2 U-perfections entail only U-perfections.
A3 The property of being a necessarily existent unicorn is a U-perfection.
Now, here’s the argument:
1. Suppose being a necessarily existent unicorn is impossible
2. Since being a necessarily existent unicorn is impossible it entails not-being a necessarily existent unicorn
3. From A3, A2, and 2, the property of not-being a necessarily existent unicorn is a U-perfection
4. But this contradicts A1, if a property is a U-perfection it's negation cannot be a U-perfection
5. Therefore being a necessarily existent unicorn is not impossible
6. Being a necessarily existent unicorn is either impossible or actual
7. From 5 and 6, there is something that is a necessarily existent unicorn.
So, the GOA and its modern variants pull no weight here. It doesn't give any convincing reason, by my lights, to think our credence in theism should be anything more than 0 or arbitrarily close to 0. Instead we can focus on the contingency argument. In its strongest variants, I think the contingency argument is much more forceful. Do I think it's compelling? Not really. I can't run through all the considerations here, as the topic is dense and spawned a rather vast and technical literature. The first problem is with the PSR. Unrestricted or strong PSRs tend to lead to modal collapse, which has numerous problems, not the least of which being that it implies there's no contingent facts to explain, making the argument self-undermining. Whereas weak or restricted PSRs either end up entailing strong one's, or are too weak to support the theistic-friendly conclusion[18]. There's also the Hume-Edwards principle, which suggests that we don't need an explanation for the set of all contingent facts, instead an explanation for each fact is enough, but that means we don't need a necessary explanation outside of the set of all contingent facts[19]. But that's just stage 1, which doesn't even get you to theism yet. You also have stage 2 arguments, which attempt to bridge the gap between necessary being and God, but I think those are even more unconvincing[20]. A lot more can be said here in reply, and reply back, but it would take us off-topic. I'm not here to discuss the contingency argument.
The most crucial problem here is, once again, BB's problematic atomism[21], and abuse of credences. When assessing one's credence in the success of arguments for theism such as contingency arguments, there are several interlocking parts and background considerations. I've already mentioned considerations surrounding the PSR, and stage 2 arguments, but even that barely scratches the surface. One might reject the argument on the basis of a generally pessimistic view of metaphysics[8]. One might reject the argument on independent or Moorean grounds, e.g the argument must have gone wrong somewhere, since you have a higher credence in the proposition that God, traditionally conceived, does not exist than you do in the premises. This could be due to the problem of evil in its many forms, divine hiddenness, arguments from divine incoherence, arguments for metaphysical naturalism, arguments from low priors etc. The point being, one's credence in the success of the contingency argument as a proof for theism (which would basically just be one's credence in theism) is determined relative to a background set of beliefs and methodological commitments. So BB's demand that we consider the contingency argument on it's own, and that "surely you should have a 1% credence in the success of the contingency argument, right?" Is, quite frankly, unserious. It's a rhetorical move and nothing else. One's credence in theism will be a function of weighing one's total evidence, background beliefs, and epistemic context therein, not some arbitrary credence one assigns when considering the contingency argument alone.
To see the point another way. We can use BB's atomistic approach[21] to credences to make an argument that theism is false with a probability of 1 given BB's own views. BB thinks all possible people exist. A set that has the cardinality of, at minimum, ℶ2. Let an instance of gratuitous suffering be an instance of suffering God would not permit. Now consider the proposition that "a randomly selected member of the set of all possible people minimally has a 0.0000000000000000000000000001% chance of having experienced gratuitous suffering at any point in their existence". That seems plausible, right? I mean it doesn't seem that unlikely a priori that some random possible person experienced gratuitous suffering, and 0.0000000000000000000000000001% is a pretty modest estimation. But the problem is, if for any randomly selected member of a set of ℶ2 people, there is a 0.0000000000000000000000000001% chance they experienced gratuitous suffering, then what's the chance that, of the entire set of ℶ2 people that at least one of them experienced gratuitous suffering? Well, since a set of ℶ2 people is so large, even if the probability of one randomly selected person experiencing gratuitous suffering is vanishingly small, the probability of having at least one member of that set experiencing gratuitous suffering will turn out to be 1, or at least pretty much 1.
Let G denote "it's not the case that a randomly selected person has experienced gratuitous suffering", and Gn be the proposition that ‘nobody in the set of ℶ2 people that exist experienced gratuitous suffering’.
P(G)=1−10−31 (or 0.9999999999999999999999999999)
P(Gn)=(1−10−31)ℶ2
ho
ho
However, that's 0. So:
P(~Gn)=1−0=1
But ~Gn is logically equivalent to the proposition, 'someone in the set of ℶ2 people experienced gratuitous suffering'. So, ~Gn has a probability of 1 and entails that God does not exist, since God would not allow someone to undergo gratuitous suffering by definition. It follows that 'God does not exist' has a probability of 1, given we accept my constraints and BB's view that there exists ℶ2 people. Yikes![22]
BB is not without responses. He might say "Hang on a minute. Something funky is going on here. We shouldn't be assessing the probability of the proposition that some random person undergoes gratuitous suffering in isolation. If we're theists, then our beliefs entail that there is no gratuitous suffering, so conditional on theism it's a probability of 0 that anyone experiences gratuitous suffering. Even if we don't conditionalize on theism, if you think there's some remotely plausible theodicy that explains some type of gratuitous evil, that theodicy once held in the background will probably cover every token instance of that type. So, it's not like we should assign an arbitrary credence to some person experiencing gratuitous suffering and assume an independent uniform distribution of the property of 'experiencing gratuitous suffering' across the set of all people". To which I would then reply "Congratulations! You've realized the folly of your credence-abuse".
That was our first extensive detour in this section. But like true soldiers we march on! BB mentions provability, lack of arbitrary limits, and 'ontological uniqueness' and claims that those are theoretical virtues of a theory, in response to my argument that theism has a low prior in virtue of having many ways of turning out to be false. Provability is what I just discussed. I go on to address the arbitrary limits argument in my initial piece, and we'll get into arbitrary limits more later here. As for ontological uniqueness, it's not clear what BB means here. He doesn't explain what he means nor cite any literature on why we should take it to be a virtue of a theory. Were I to venture a guess, ontological uniqueness means being such that some object is utterly unlike anything else that exists. In which case, not only do I reject that such a feature is a virtue of a theory, I think it's a vice. Some object being utterly unlike anything else that exists means that it fits poorly with our background knowledge and experiences of the world. It means that we have the best inductive evidence we could have against the existence of such an object. Inductive inferences generally take the form "every F we've seen has had some property G, so, for every F, it probably has property G". But being ontologically unique means God has properties that nothing else has let alone anything we could experience. Which means many of our inductive inferences will lead us to the conclusion that there is probably no object with God's properties. Perhaps BB means something else by ontological uniqueness, but if so, I can't be faulted for this misrepresentation, as BB didn't explain what he meant.
While we're on the topic of background information. BB says the following in reply to my point that theism fits poorly with our background information relative to naturalism:
I don’t think that just looking around and seeing if beings around you have some property tells you very much about whether the foundation of all of reality has that property. This is especially so if, as I’ve argued, we have a plausible story of why that view of ultimate reality would produce conditions like the ones we observe.What BB's point is here is not entirely clear to me. If it's that fit with background information about the kinds of beings there are isn't a virtue of a theory, then that's simply ridiculous. First off, it would imply that many of our ordinary inductive inferences are undermined, since, as stated above, inductive inferences generally take a form that looks something like "every F we've seen has had some property G, so, for every F, it probably has property G". But this same general form is identical to the kinds of inductive inferences you could make against theism in virtue of it's poor fit with our background knowledge, e.g "every person we've seen has been embodied, so, for every person, they are probably embodied". Second, when we think about examples, it will become very obvious that fit with background knowledge is a virtue of a theory.
Suppose you go into your drawer, and you find that your socks are missing. Oh no! You reflect on two potential explanations for this. E1 is that your spouse took them out while doing the laundry, or perhaps misplaced them by accident. E2 is that some sock-stealing fairy stole your socks. E1 and E2 look a priori equally intrinsically simple. Both make the same amount of ontological commitments, E1 to your spouse, and E2 to some sock-stealing fairy. Both commit to the truth of a similar amount of propositions, with E1 committing to "there exists your spouse, there are some specific desires your spouse has that make it likely that they'd be responsible for your socks not being there, and they have the power to make your socks not be there among other things" and E2 commits to "there are sock-stealing fairies, they have the desire to steal socks, they have the power to steal socks". If anything E2 might even be less complex, since the desires it posits are more simple, and the powers it posits are more simple. Nonetheless, you think E1 is much more probable than E2, to the point that you think E1 is a plausible explanation, and E2 isn't even a contender. Why? Because E1 fits with your background information much better than E2. In virtue of your experiences, you know your spouse exists, you know what powers they have, you know they do your laundry (or at least you know it's common for spouses to do laundry in general), and you may even know they have a tendency to misplace things. So I don't even need to argue that fit with background information is a theoretical virtue, you already probably believe that.
More charitably, maybe BB is saying that background knowledge doesn't matter when it comes to fundamental entities. If that's his point, then for starters I see no reason whatsoever to believe that, and so BB needs to argue it. Further, I think that's implausible. Every other theory I can think of, especially scientific theories, are assessed relative to our background understanding of the world, so why think theories that invoke fundamental entities are anything special? Even when scientists assess theories about elementary particles like electrons, the best theory is one which has to fit with our existing background knowledge. In this case, wave-particle duality, spectral lines, and Heisenberg's uncertainty principle. Such a paradigm shift in our background knowledge is in part why we eventually moved from the classical model to the quantum mechanical model. Finally, even if that's right I would then just take that to be a reason to be skeptical of all theories about fundamental reality. I see no reason to think we can abstract away from, and completely go beyond our background knowledge, our conceptual frameworks, and our experiences, to a priori intuit what the best theory of ultimate reality is from the armchair. In general I see no good reason to think the criteria of success employed by a priori theoretical metaphysics is truth-conducive. Among other things, you have the problem of how our minds could be causally related to the relevant object of our belief in such a way to produce reliable knowledge.
I argue that theism is unparsimonious when compared to naturalism. BB replies as follows:
So, BB claims that simplicity has only to do with fundamental entities. That's a controversial view, sometimes known as Schaffer's Laser[23], and so I am not sure why he simply asserts it as if it’s uncontroversial or even widely accepted. BB also does not even attempt to argue for it nor does he even cite literature arguing for it. Further, I contend that such a view cannot be true in total generality. Here's a trivial example similar to one I used in a different post. Contrast the theory of natural selection NS, with the theory of natural selection + the proposition that there are invisible undetectable fairies, NS&F. Both theories deal with non-fundamental posits. Further, both have the exact same empirical consequences, so evidential considerations can't break the tie. But NS is clearly a better theory than NS&F, because it's more parsimonious and thus more likely to be true intrinsically. So long as F has a probability of less than 1 and is independent of NS, something that's pretty much a given, it will follow that NS&F is less probable than NS alone. Since you'd multiply the probability of NS by the probability of F to get a lower probability.TT worries that theism is less simple than naturalism because “naturalism only commits to natural concrete entities, and causal relations between those entities, and theism posits a supernatural entity.”Simplicity, however, is about fundamental entities. It would be absurd to say one should think God wouldn’t create because that would be simpler. Theism, then, does very well, positing just one fundamental entity with one unlimited property—perfection.
Perhaps the idea is that the simplicity and multiplication of fundamental entities are all that matter when assessing grand theories. Firstly, that seems false to me. When it comes to grand theories we are evaluating all that those theories entail, the ontological commitments and the propositions loaded into them. Theories, after-all, are sets of propositions. So theoretical content has to do with the set of all propositions entailed from a theory, not just those propositions concerning fundamental posits. Theism and naturalism are theories about all that exists, and the shape of causal reality after-all, not just what exists at the foundation. But since grand theories are theories about everything that exists, we have, if anything, more reason to evaluate grand theories based on more than the fundamental entities posited by those theories. This just seems to follow from how we'd assess priors generally, we take the conjunction of all the propositions entailed by a theory, and calculate the probability of the conjunction of those propositions being true.
Perhaps the idea is that the simplicity and multiplicity of fundamental entities is what matters most when evaluating theoretical content, it "comes first" so to speak before other considerations regarding derived entities. I think this is the most plausible version of the "Laser" being appealed to, and indeed, I understand this to be Schaffer's view. In answering objections Schaffer is careful to stress that the multiplication or simplicity of fundamental entities is not the only thing to consider[23]. This doesn't help the theist, however. I think it can be plausibly argued that theism and naturalism are at least a wash when it comes to the simplicity of their fundamental posits, if not naturalism has an outright advantage. For starters, theistic grand theories are actually committed to the fundamental entities that naturalism posits, such as elementary particles, quantum fields, or the universe if we're priority monists. Theism posits the existence of a supreme supernatural agent in addition to those fundamental physical entities, so by the naturalists' lights, the theist is indeed multiplying fundamental entities beyond necessity[24].
But let's say we just compare the most fundamental entities of both naturalism and theism. I see no good reason whatsoever to think that God is more intrinsically likely than sets of elementary particles, quantum fields or the universe. On top of the fact that the existence of elementary particles, quantum fields, and the universe are well-established empirical facts that play roles in our best scientific theories, and theism both is not and doesn't. I think there are other reasons to think theism is not simple, and so doesn't get an advantage. God is attributed numerous complex properties such as conscious and intentional agency, infinite knowledge, infinite causal power, moral perfection, and so on. Where one could think (as I do) that each of these attributes add to the overall complexity, and lack of coherence in the concept of God. Also, as we've seen, theistic explanations invoke mystery and inscrutability, making them generally less epistemically simple compared to naturalistic explanations that are grounded in observable and testable phenomena. There are some ways theists have tried to get around this, by arguing that God is simple by virtue of being an entity with the property of perfection, or an entity that lacks any limits where all of God’s properties are just entailed from that. None of those, I think, are convincing, but we'll get to that later.
So, we have reason to think that theism and naturalism are at least a wash in terms of their fundamental entities. But that means we can then assess other features of those theories to break the tie, such as total ontological commitments, and the kind of ontology in general both theories propose. Naturalism posits a monist ontology with less ontological commitments, theism posits a dualist ontology with more ontological commitments[25]. So, when we move to simplicity considerations regarding non-fundamental entities, naturalism wins that round.
BB next objects to my argument from the success of the natural sciences which rules out supernatural explanations. His reply has two very brief parts, but despite this, I will examine both of them thoroughly. BB says:
It’s also replete with us discovering things that point to the supernatural, like fine-tuning, the presence of laws, psychophysical harmony, and anthropics.So, BB's first response is to appeal to philosophical Bayesian arguments for the existence of God. First of all, this misses the point. The point is that, when only considering the background of our best scientific theories that are uncontested among both intellectually serious theists and naturalists, the content of those scientific theories, do not invoke supernatural entities, and the history of science has involved replacing explanations thought to involve supernatural phenomena with natural phenomena. Of course this isn't to deny that there are also Bayesian philosophical arguments that theists have for the existence of God, but those are highly contested, and weren't what I was talking about.
Further, when discussing the idea of methodological naturalism as a progressive research project, the primary thing I brought up was novel predictions. But none of those arguments can be plausibly thought to be novel predictions of theism, in the same way that paradigmatic scientific theories such as general relativity have generated novel predictions. For instance, the existence of us, and our harmonious faculties was known prior to our coming up with theism, and indeed theism was constructed as a causal explanation for that phenomena as God is the creator of the universe and all life. So, it's not like theism predicts those things, independent of how theism as an explanatory hypothesis was constructed. Perhaps one could say that the fragility of the constants was discovered after the construction of theism. But theism doesn't predict the fragility of the constants, it predicts life, not that God would create life within a narrow range of life-permitting conditions. Note that I'm not even saying here that novel predictions are necessary for those arguments to work, just that novel predictions are a feature of a progressive research project that theism lacks.
Finally, I'll note that I reject all those arguments. I and co-authors have discussed our gripes with the argument from psychophysical harmony and I will not repeat those gripes here. Whether there is a theistic explanation for anthropics is the very thing in contention here when assessing our considerations as to whether a naturalistic theory should be preferred, so appealing to that is something akin to question-begging. The presence of laws gets into complicated stuff about laws of nature, which could warrant an entire article in its own right, so I don't want to discuss that here. So that leaves fine-tuning, which I think is probably the strongest among the arguments listed. There are many solid blogposts the discuss fine-tuning that make, what I think are powerful objections. There are many great papers with objections against the fine-tuning argument as well. Here are just 3 problems with the FTA: (i) it assumes an independent uniform distribution across all the possible values of the constants on naturalism but there's no good reason to assume that and good reasons to reject it, (ii) it assumes that, once life is counted as evidence for naturalism, then the fragility of the constants is further evidence for theism, but given God's omnipotence (and thus ability to actualize life within any physical conditions) there's no good reason to think that, lastly (iii) versions of the design hypothesis that are so constructed or constituted with resources to predict the life-permitting constants have a correspondingly lower prior.
I could go on but once again though, this argument has spawned a great deal of literature, and I can't do justice to the dialectic here without dedicating an entire article to it. BB's quick mention of those arguments without bothering to say anything more of substance was, I think, an attractive rhetorical move that might hold up in a live discussion, but it hardly passes for rigorous and serious philosophy.
Now, BB's next comment is the following:
Furthermore, the theodicy I’ve given explains why that would be—God puts us in a broadly indifferent universe. God supernaturally intervening all the time would be the byproduct of poor design.
However, this doesn't actually engage with what I said at all, which is that naturalistic explanations fit within a progressive research project and theistic or more broadly supernatural explanations don't. Maybe God has some reasons to not intervene, but that doesn't affect my point.A Detour on BB's Theodicy
BB appeals to his theodicy in his last comment, and this was his third time doing so in the article. So, I guess I'll take this opportunity to discuss it. The theodicy is the 'hypothesis of indifference' theodicy, which goes as follows:
The basic idea of the theodicy is as follows: God places us in an indifferent universe that resembles the typical indifferent universe where we are capable of understanding the broad features of the world. An indifferent universe is one with features that are not themselves affected by the value of the world. Just as naturalism predicts evil by the hypothesis of indifference, by saying the universe doesn’t care about us, this theodicy says God would put us (for some time) in a universe that doesn’t care about us.Now, on its own (you'll see why I say 'on its own' shortly), this theodicy is, I think comfortably, the worst and most unconvincing theodicy I've seen. The story BB proposes is just that God would create a universe that gives rise to observers, but then He'd just be indifferent to the suffering and well-being of those observers for some period of time. The problem is, antecedently, we have no reason to think that God would create a universe like that over any of infinitely other possible universes He might create instead. Suspiciously, BB’s use of the phrase ‘indifferent universe’ appears to basically be a stand in for ‘our universe or one very similar to it’, but in that case, BB is just saying ‘God might have a reason to create a universe like ours’, but that’s trivial.
Note, on such a picture, we won’t always be in an indifferent universe. We will one day be in heaven and things will be very good. But the theodicy claims that there is something valuable about our temporary stay in an indifferent world.
Because it incorporates the hypothesis of indifference, it explains evil precisely as well as naturalism does. It says that God puts us in an indifferent universe, so the evils that you’d expect are exactly those that you’d expect on theism. For this reason, evil does not favor naturalism over theism + this theodicy.
True enough, God might have reason to create a universe like ours, but the same could be said, say, about a universe composed mostly of proto-conscious fart particles (no more conscious than fart particles in our universe if you accept panpsychism). Merely giving an epistemically possible thing God might do, does nothing to affect our evidential judgements, what matters is what God is more likely to do antecedently. In fairness, some other theodicies, like the saint-making theodicy, are also pretty ad-hoc on my view. But at least saint-making connects the evils we observe to entailed future goods in the afterlife, the 'hypothesis of indifference' theodicy, in itself, just says "well, maybe God would create an indifferent universe like ours" without connecting that to any even remotely plausible story on which greater goods are entailed from it.
I'll also note that the theory (theism + indifference), so construed, is highly gerrymandered and suffers from probabilistic tension. BB does not hold after-all, that God creates a wholly indifferent universe, He creates a universe that is weighted towards value insofar as it gives rise to sentient life. The problem is, virtually all reasons to think a designer motivated by moral reasons would be indifferent towards value in creating a universe would be reasons to think the designer wouldn't make a universe weighted towards conscious life with capacity for moral agency, and virtually all reasons to think a designer motivated by moral reasons would be different to bringing about valuable states would be a reason to think said designer wouldn't create a universe that's indifferent to value after life begins to exist. In fact, I contend that, antecedently, it appears to me much more likely that God would create a universe with no life at all, than that He'd create a universe with life, but that's indifferent to suffering and well-being. The former is more intrinsically probable (it's less theoretically complex in terms of desires it posits of God and less ontologically complex in terms of the outcome of those desires), and it aligns with my moral intuitions more (it seems very bad to create a universe with horrific suffering, it doesn't seem very bad to create a universe with no life). I don't deny that the theist can tell an epistemically possible story on which God creates a universe that leads to life, but is indifferent towards well-being and suffering, but again, I see no reason whatsoever to think such a story is any more antecedently plausible than any other merely possible story you can tell.
Finally, the theory of theism + hypothesis of indifference (call it Th) is empirically vacuous. Part of the problem here, I think, is that it's not clear what is meant by "indifferent universe". Prima facie, it looks like virtually any range of possible observations we might have can be easily accommodated with God putting us in an 'indifferent universe' and eventually heaven for the sake of greater goods, in accordance with exactly what we'd expect if Th is true. Indeed, BB himself says that evil wouldn't be any evidence against such a theory, it "explains evil precisely as well as naturalism" says BB. But since theism's predictive content is constrained only by moral goodness, the only observational evidence we could have against theism could only be evidence of what God would probably not allow, which would be gratuitous evil. But since any evil we observe is no evidence against Th, it follows that there's no evidence we can observe which can be evidence against Th. Suppose we observe the most horrific evil we could observe (not hard to imagine, some evils in our world are pretty close to that). Th can explain that by God putting us in an indifferent universe. Here's the problem; In the standard Bayesian framework, a hypothesis H is confirmed when the conditional probability P(H|O) > P(H). It's a theorem of probability that for any observation O, P(H|O) > P(H) iff P(H|~O) < P(H). This means that an observation O counts as evidence for a hypothesis H if and only if ~O would count as evidence against H. So, if no observation we might have can count against H, then nothing can bolster it either. But this would mean that, since nothing we observe can be evidence against Th, it follows nothing we observe can be evidence for Th making Th a non-starter as an explanation for any observation.
Perhaps BB could say that evidence of a “hell-world” that is different towards bringing about suffering, or bad-states might be evidence against Th. But even if some observations could verify us being in a hell-world and count as evidence against Th, all this would mean is that only observations falling under that type (O1) could be evidence against Th, but no other type of observation (O2, O3, On...) or lack thereof could be evidence for or against Th. Only ¬O1 (not observing a hell-world) could provide any evidential confirmation for Th, but even then, not much, since ¬O1 is strongly predicted by ¬Th too (given it's intrinsic improbability). Any other observation would be equally likely under Th and ¬Th. Thus, Th would still lack substantive empirical content, and the kinds of observations BB cites as evidence for Th (harmonious laws, specific persons, and other goods) cannot actually be evidence for Th, since observations of their absence couldn't be evidence against Th.
On the other hand, maybe BB means something much more specific by 'indifferent universe.' For instance, maybe what is meant is that God creates goal-directed biological organisms through unguided evolutionary mechanisms like those in our world, causing these organisms to experience sensations such as pain and pleasure in response to certain stimuli. These mechanisms, together with the conditions of the world, are indifferent to moral goals or values in such a way that we'd expect the kinds of horrendous suffering we see. While that does have some empirical consequences, that would just be to build the specific data we observe into Th. That is, quite clearly, viciously ad hoc and makes his model an unvirtuous just-so-story, not anything that resembles a serious explanatory hypothesis. Furthermore, the only reason we would be loading Th with those specific assumptions rather than others is because they entail our specific observations. Those observations then cannot serve as any evidence for Th, as this would be explicitly double-counting. Finally, if that specific and complicated theoretical content is built into Th, it will then also have a correspondingly astronomically low prior probability built into it, insofar as our observations themselves are taken to be extraordinarily improbable. None of BB's motivations for his theodicy deals with this problem of empirical vacuity, which, I take it, is devastating for his model.
So far, I have evaluated BB’s theodicy in a vacuum, but could conjoining it with some other assumptions make it more plausible? Here are the motivations he gives for it from this article:
But the problem is this is just appealing to other theodicies such as free will, soul-making, connection-building etc. There's not a meaningful difference between using those theodicies to explain why God allows seemingly gratuitous evil and using those theodicies to justify why God makes an indifferent universe. The theistic narratives being appealed to are the same (God permits evil because it fosters soul-building, divine intimacy, connections etc.) and the data they're used to explain is the same (why God allows horrendous evil). It's fine to appeal to those theodicies, but then they're what's pulling all the weight, not the 'hypothesis of indifference' theodicy (if it could even be properly called a theodicy in its own right). BB has not explained what his theodicy adds above and beyond the disjunction of what is above here, and the data they explain does not obviously entail an ‘indifferent universe like ours’.
Also, without wasting too much more space, none of those, I think, are compelling at all. I've addressed the soul-making and participation theodicies (points 3 and 5 respectively) elsewhere extensively and I won't repeat myself here. The first point is just "perhaps there are various unknown reasons"[26]. But that's no good, as you could say that about basically anything, to make theism predict virtually any outcome antecedently. We can run parodies of the reasons BB gives. Maybe God making a universe with only an unimpressive proto-consciousness, or non-conscious entities but disposed to produce consciousness, or simply no consciousness or dispositions to produce consciousness at this point in time at all, but rest assured, where He will create consciousness at some time, is conducive to various unknown afterlife goods. Maybe it leads to better unknown consequences somewhere down the line.
1. There might be various unknown reasons. Perhaps putting us in a universe like this is valuable because it is required for various hard-to-guess afterlife goods. Perhaps there’s some consequential decision that we’ll make in 5 quadrillion years that God knows will be positively shaped by us being in an indifferent universe. Given that we only witness 70 years of our lives or so, we’re not in a position to guess whether there are such great afterlife goods. This theodicy also avoids the problems with skeptical theism, because it only requires positing an unknown reason for one thing—making an indifferent universe. But surely thinking that God has some unknown reasons to do one particular act doesn’t produce global skepticism or any of the other untenable results of skeptical theism—we already know that we’re wrong sometimes!2. Perhaps being in an indifferent world, one whose features resemble the typical godless world that contains us and where we can know about the broad features of the world, strengthens our relationship with God. To consider an analogy: your relationship with your eventual spouse might be strengthened by the fact that you spent time without them. Having time without someone might strengthen your relationship with them. Similarly, time spent in a world apart from God might strengthen one’s eventual relationship with God, a relationship that is of infinite value.3. Perhaps struggling through an indifferent world, not being micromanaged by God, is uniquely valuable for soul-building. Just going through a narrowly tailored set of challenges doesn’t give one the knowledge that they can overcome hardships in the same way that overcoming a random suite of challenges does. The benefits of soul-building, as of the other benefits on this list, last forever.4. Perhaps being in an indifferent world builds our connections with others. An indifferent world might do this in at least three ways, relative to one where God intervenes. First, the fact that our lives have some element of randomness, where everything isn’t set by God makes it so that our relationships may be more valuable. God designing the world specifically to be such that specific people form maximally valuable relationships might rob the relationships of some of their value, in a way akin to arranged marriage. Second, as Robin Collins notes, going through hardships together, helping people out through hardship, and forgiving others through wrongdoing strengthens people’s relationships. An indifferent universe offers the chance for all of this. Finally, were we to be in a relationship with God, it might be infinitely more intense and thus crowd out our other relationships. Thus, just as there might be something valuable about a person making close friends before they get married, because a marriage crowds out their friend-making ability, the same might be true of us and God. This also seems to preclude God specifically intervening to maximize connection building. This is worth all the world’s evils, because the value of relationships lasts forever into the afterlife.5. There might be something valuable about us working to better the world alongside God or, more broadly, working with God to achieve things. But, as Collins notes, that requires God to be in some way apart from the problems of the world, so that we truly work with him, rather than him constructing sham problems that we together work to solve. For A and B to work together to solve problem C in a valuable way, plausibly neither A nor B has to be fully in control of all the workings of C.6. As Van Inwagen suggests, it’s valuable for us to freely choose God. But that requires that we be apart from God in some broad sense, and have the opportunity to choose him. That requires that he’s not ordaining the goings-on in our world. But that means that our world would be expected to look like a typical Godless world, except in the ways required for us to understand God’s nature, the nature of the world, and for us to exist at all.
But the problem is this is just appealing to other theodicies such as free will, soul-making, connection-building etc. There's not a meaningful difference between using those theodicies to explain why God allows seemingly gratuitous evil and using those theodicies to justify why God makes an indifferent universe. The theistic narratives being appealed to are the same (God permits evil because it fosters soul-building, divine intimacy, connections etc.) and the data they're used to explain is the same (why God allows horrendous evil). It's fine to appeal to those theodicies, but then they're what's pulling all the weight, not the 'hypothesis of indifference' theodicy (if it could even be properly called a theodicy in its own right). BB has not explained what his theodicy adds above and beyond the disjunction of what is above here, and the data they explain does not obviously entail an ‘indifferent universe like ours’.
Also, without wasting too much more space, none of those, I think, are compelling at all. I've addressed the soul-making and participation theodicies (points 3 and 5 respectively) elsewhere extensively and I won't repeat myself here. The first point is just "perhaps there are various unknown reasons"[26]. But that's no good, as you could say that about basically anything, to make theism predict virtually any outcome antecedently. We can run parodies of the reasons BB gives. Maybe God making a universe with only an unimpressive proto-consciousness, or non-conscious entities but disposed to produce consciousness, or simply no consciousness or dispositions to produce consciousness at this point in time at all, but rest assured, where He will create consciousness at some time, is conducive to various unknown afterlife goods. Maybe it leads to better unknown consequences somewhere down the line.
BB also appeals to Van Inwagen's free will theodicy, but, qua ‘theodicy’, it runs into the same problems as the regular old free will theodicy as well as many theodicies in general, (doesn't explain natural evil, accords very poorly with our moral practices and intuitions, it's logically possible for God to create agents who freely choose the good and Plantinga's TWD doesn't work as a response to that and so on.). More importantly, it's worth noting that Van Inwagen grants that the stories to which He appeals, including the one where God allows suffering to allow creatures to freely choose Him are only "true for all we know", mere defenses, and he argues that this together with his broad modal skepticism defangs the problem of evil. But his modal skepticism would also, insofar as it defangs the problem of evil, make theism off the table as an explanatory hypothesis by making theism consistent with any observations given we can appeal to some 'for all we know' story about certain goods and their modal connections to our observations. While this is no problem for Van Inwagen who rejects evidential natural theology[27], this is a huge problem for BB. So Van Inwagen’s ‘theodicy’ definitely cannot be used in the way BB wants to use it, indeed his entire approach to the problem of evil is an anathema to the way BB wants to approach it.
So that leaves only connection-building (4.) and the divine intimacy theodicy (2.). There are two problems with those theodicies (there's more actually, but that would demand more space). The first problem is they don't actually explain the data. If God permits suffering for us to foster relationships with others, we'd expect there to be empirical evidence that horrific evils do in fact generally lead to those valuable relationships. However, not only does the empirical evidence not bear this out, the evidence we do have cuts against it. If God permits suffering to foster valuable relationships with Himself, then it makes the existence of non-resistant non-belief in its various forms, as well as its geographical distribution puzzling.
The second direct problem is that they belie a total lack of imagination. God is omnipotent and so can do all things logically possible. So even if we did observe that horrific evils was generally connected to valuable relationships of that sort, there's no good reason, so far as I can tell, to think such a connection is logically necessary. But if it's logically possible for God to bring about all the same goods of those valuable relationships without forcing creatures to undergo horrific suffering, then God, being morally perfect, would do that. Even if one thinks that, logically necessarily, a world with the most valuable relationships must involve shared struggles[28], it is logically possible for there to be shared struggles without horrific evils. Consider, as just one example, the shared struggle of a sports team up against insurmountable odds. That does not entail horrific suffering. BB could of course reply that he's just telling a story on which the goods of connection-building and relationships to God are instantiated in the afterlife, and it is somehow logically necessary that we experience horrific suffering on earth to experience those goods to their fullest extent. But again, I see no reason to accept such a story as being, even remotely, antecedently plausible. I could tell the story that any natural initial state of the universe, logically necessarily, must produce us. That's just as antecedently plausible by my lights, which is to say, not at all.
I'll also note that, as I said in my initial piece, the appeal to us being in an infinite afterlife in the future, implies that us being in a situation where we aren't in the afterlife, has an epistemic probability of 0. For any t, it's infinitely likely we'll be in one of the infinite times that we are in the afterlife, and not one of the finite times on earth.
I know this was a lot of space spent proportional to the two sentences I was initially responding to. But I harped on this both because BB often does these quick responses that seem to substantively respond to what I said, but actually just gloss over various issues, and because I wanted to highlight more of my discontents with BB's approach to evaluating and testing theism as a hypothesis, why I think it's much too credulous, and doesn't stand under scrutiny. BB seems to think he can just list out some theodicies, and that's enough to take out the bite of the problem of evil, without leading to other issues afflicting his model. But it's not that simple.
Arbitrary Limits and The Parsimony of Theism
Next we have BB attempting to refute my skepticism of arguments for the parsimony of theism. For this, I'll remind the reader that the dialectical burden falls squarely on BB's shoulders to prove that theism is intrinsically likely. So all I'll do here, and all I'm obligated to do, is explain why I'm just not convinced by the arguments BB makes. Starting with arbitrary limits. BB says:TT objects to the idea that arbitrary limits is a vice of a theory, saying that he doesn’t share the intuition. By this standard, then, the speed of light being unlimited starts out no more probable than it being, say, 100,000,323 miles per hour and there’s nothing wrong with thinking the universe stops somewhere for no reason. This is very implausible, and the history of scientific inquiry shows us discovering many unlimited things.BB mentions the speed of light, but I don't think he could have picked a worse example for his point. Not only do we know that the speed of light does in fact have a limit, (approximately 299,792 kilometers per second) the speed of light having a limit is not considered a theoretical defect at all. For one, the speed of light having a certain limit allows for more precise, testable predictions by providing a clear boundary condition within which the laws of physics could operate. Theories which posit that the speed of light has no limit would lack predictive power in that regard. For two, the finite speed of light introduces a symmetry in the equations of special relativity, leading to a more elegant and unified mathematical structure. Finally, one might even think theories with a well-defined limit on the speed of light can be seen as more aesthetically pleasing and elegant because it imposes a natural order and structure to the universe. Of course, it would be vicious to randomly stipulate just some arbitrary number to the limit of the speed of light, but the same thing goes for stipulating that the speed of light is unlimited arbitrarily.
He also says "This is very implausible, and the history of scientific inquiry shows us discovering many unlimited things". So, you might think he goes on to provide examples of this. He doesn't even provide a single one. Convincing, I know, I am just utterly floored.
Now, in fairness, infinites do sometimes show up in our scientific theories, but usually just as idealizations to simplify calculations or indicators of where the theories break down. For instance, the singularity in a black hole is considered an infinite density, but this is understood as a sign that general relativity is incomplete and will ultimately be replaced by a more comprehensive theory of quantum gravity. Another example involving general relativity, GR implies that space is a continuous manifold, meaning that every path in space (and time) is infinitely divisible, which allows a mathematical description of paths through spacetime to be divided into an infinite number of smaller segments. Last example, in statistical mechanics, the thermodynamic limit involves taking the number of particles (N) and the volume (V) to infinity while keeping the density (N/V) constant.
Where infinites generally do not show up, is in the causal powers of the theoretical posits of our best scientific theories. Generally, theories which have posits with unlimited causal power are considered unvirtuous. First because, as mentioned before, it hampers our ability to generate testable predictions since if the theoretical posit has unlimited causal power it becomes consistent with a wider range of possible observable outcomes rather than the specific one's we get. Second because, in general, we would only posit what is necessary to explain our observations. If some posit having limited causal power is sufficient to explain our observation, then we wouldn't say that it's unlimited if such a commitment is unnecessary to explain our observation. Unlimited causal explanations posit more causal power than needed.
In my original post I next argue that even if unlimited entities are more likely than limited one's this doesn't show that unlimited being theism is more likely than the disjunction of all possible limited being theisms, so limited being theism is more modest. BB replies:
In any case, he also gives an analogy from Dustin Crummett, to the laws of nature being uniform. The idea being, the laws of nature being uniform is much more likely than them being non-uniform even though there are many more ways the laws of nature could be non-uniform. That said, I'm not really convinced by this. First of all, because I accept that the laws of nature are uniform on the basis of a posteriori reasons, not because I think it's more likely a priori. We've seen such-and-such interactions are constantly conjoined with such-and-such outcomes, so by induction we infer that for any time where we have such-and-such interactions there will be such-and-such outcomes. One can reply that this kind of induction could only work if one assumes that the laws of nature are uniform, but then that's just the problem of induction. So any solution to the problem of induction one accepts will cover that. I'm pretty much agnostic on whether the laws of nature being uniform is more probable a priori, than them being non-uniform[29].
Further, even if I did think that the laws being uniform is a priori more likely, the cases are not analogous. There are infinitely many ways for the laws of nature to be non-uniform, true enough, but there are also, plausibly, infinitely many ways for them to be uniform. Uniformity does not imply a single set of possible laws; rather, it means that whatever the laws are, they are coherent and apply universally. But just as you can have many sets of laws that are coherent and apply universally, so too you can have many sets of laws that are unpredictable and not universal (if they would even be laws at that point, if not replace laws with laws*). Are there more ways for them to be uniform or non-uniform? Who knows. So other considerations than modesty can break the tie. By contrast, there's only one way for an agent to be unlimited, (they have to have no limits in power, knowledge, or goodness to any degree n) but there are infinite ways for an agent to be limited (by having some limit or other, in power or knowledge or goodness, to any degree n).
So that's arbitrary limits. Next we turn to discuss whether 'perfection' is a simple property. BB says the following:
As for the modal realism analogy, there are several glaring issues. First off, I probably don't even share the intuition here. My credence in modal realism is basically zero, or so close to zero that it makes no practical difference. I suspect many others, upon reflection, would share this view (though I'll admit this is a mere suspicion rather than an empirically supported claim). Modal realism is far from a simple theory—it's about as ontologically extravagant as a theory can get. BB seems to conflate simplicity with unification here. Secondly, even if my credence in modal realism wasn't just zero, the analogy still collapses for two reasons.
First, some philosophers are attracted to modal realism largely because it offers an explanatory framework for modal facts by reducing them to what occurs in concrete possible worlds. But for this to work, all concrete possible worlds must exist. In contrast, I take it that any explanatory role that perfect being theism plays, such as serving as a causal explanation for the existence of the universe or life, can be equally filled by some version of limited being theism that possesses sufficient causal power.
Second, there's also probabilistic tension: if we imagine a version of modal realism where all but some possible worlds exist, the theory suffers. If all but a few possible worlds exist, we'd have strong inductive evidence that those arbitrary few also exist. This doesn't apply to limited theisms. A limited god can have their power capped at a certain level or degree n without this issue, as all their powers would be downstream of this limited causal ability. Further, the vast majority of limited gods have numerous, if not infinite things they can't do, so the induction would cut both ways and ultimately cancel-out.
I give an analogy in my initial piece that invokes a hypothetical species of aliens, who have the concept of 'being Y-limited', which entails a complex set of limits. Where such a species could argue that 'being Y-limited' is a simple property making use of the same arguments that 'being perfect' is a simple property. BB says:
He next worries that even if it’s somewhat of a virtue, that doesn’t mean it’s enough of a virtue to outweigh the combined probabilities of all the limited theisms. But theories that have extra arbitrary limits and are less simple tend to be much worse than theories that don’t, not just a bit. That’s wh a being that knows all things is way more likely than the combination of all infinite beings, each of whom knows every fact but one. Dustin Crummett’s given a helpful analogy: the odds that nature is uniform is much higher than the odds that it’s randomly disuniform in one small region because, though there are many more ways for it to be disuniform, each of those have arbitrary limits and are less simple.So, BB starts by simply claiming that unlimited being theism does match the probability of the disjunction of all limited being theisms. But of course, that's just reasserting the very claim in contention without giving an argument. Am I supposed to be convinced by that?
In any case, he also gives an analogy from Dustin Crummett, to the laws of nature being uniform. The idea being, the laws of nature being uniform is much more likely than them being non-uniform even though there are many more ways the laws of nature could be non-uniform. That said, I'm not really convinced by this. First of all, because I accept that the laws of nature are uniform on the basis of a posteriori reasons, not because I think it's more likely a priori. We've seen such-and-such interactions are constantly conjoined with such-and-such outcomes, so by induction we infer that for any time where we have such-and-such interactions there will be such-and-such outcomes. One can reply that this kind of induction could only work if one assumes that the laws of nature are uniform, but then that's just the problem of induction. So any solution to the problem of induction one accepts will cover that. I'm pretty much agnostic on whether the laws of nature being uniform is more probable a priori, than them being non-uniform[29].
Further, even if I did think that the laws being uniform is a priori more likely, the cases are not analogous. There are infinitely many ways for the laws of nature to be non-uniform, true enough, but there are also, plausibly, infinitely many ways for them to be uniform. Uniformity does not imply a single set of possible laws; rather, it means that whatever the laws are, they are coherent and apply universally. But just as you can have many sets of laws that are coherent and apply universally, so too you can have many sets of laws that are unpredictable and not universal (if they would even be laws at that point, if not replace laws with laws*). Are there more ways for them to be uniform or non-uniform? Who knows. So other considerations than modesty can break the tie. By contrast, there's only one way for an agent to be unlimited, (they have to have no limits in power, knowledge, or goodness to any degree n) but there are infinite ways for an agent to be limited (by having some limit or other, in power or knowledge or goodness, to any degree n).
So that's arbitrary limits. Next we turn to discuss whether 'perfection' is a simple property. BB says the following:
TT objects that perfection isn’t simple because it assumes the existence of various disunified parts. One is perfect in part because they have infinite power, so perfection isn’t simple. I’ve objected to that in various places. Consider an analogy: modal realism says every possible world exists. That’s a simple theory, even though modal realism depends on lots of different worlds existing. Modal realism serves to explain why the worlds exist, which is why it’s more probable than a theory according to which a bunch of random worlds just happen to exist. Perfection is similar.BB cites an article where he discusses perfection being a simple property. I already dedicated a side tangent to a separate article BB wrote, only because I wanted to stress the problems with BB's methodology and approach to theodicy. I won't be doing that again for the sake of space. I'll note that I've already had a back and forth with BB on the topic and you can see that I haven't been convinced by anything BB has so far had to say on the matter.
As for the modal realism analogy, there are several glaring issues. First off, I probably don't even share the intuition here. My credence in modal realism is basically zero, or so close to zero that it makes no practical difference. I suspect many others, upon reflection, would share this view (though I'll admit this is a mere suspicion rather than an empirically supported claim). Modal realism is far from a simple theory—it's about as ontologically extravagant as a theory can get. BB seems to conflate simplicity with unification here. Secondly, even if my credence in modal realism wasn't just zero, the analogy still collapses for two reasons.
First, some philosophers are attracted to modal realism largely because it offers an explanatory framework for modal facts by reducing them to what occurs in concrete possible worlds. But for this to work, all concrete possible worlds must exist. In contrast, I take it that any explanatory role that perfect being theism plays, such as serving as a causal explanation for the existence of the universe or life, can be equally filled by some version of limited being theism that possesses sufficient causal power.
Second, there's also probabilistic tension: if we imagine a version of modal realism where all but some possible worlds exist, the theory suffers. If all but a few possible worlds exist, we'd have strong inductive evidence that those arbitrary few also exist. This doesn't apply to limited theisms. A limited god can have their power capped at a certain level or degree n without this issue, as all their powers would be downstream of this limited causal ability. Further, the vast majority of limited gods have numerous, if not infinite things they can't do, so the induction would cut both ways and ultimately cancel-out.
I give an analogy in my initial piece that invokes a hypothetical species of aliens, who have the concept of 'being Y-limited', which entails a complex set of limits. Where such a species could argue that 'being Y-limited' is a simple property making use of the same arguments that 'being perfect' is a simple property. BB says:
TT worries next that perfection isn’t simple. It’s just one word in English, but a single word can express a complicated concept. But perfection just means unlimited goodness, and goodness is simple and fundamental, so perfection is simple. The fact that something can be described easily in English doesn’t mean it’s simple, but the fact that it has an unlimited amount of some fundamental property does
This doesn't entirely construe the spirit of the point I was making correctly, it's not just that perfection is a word in English which can have the content of a complicated concept. Though, I'll admit I could have benefited from being more clear. I was pre-empting and responding to the point that 'perfection' is simple, as once you stipulate that an entity instantiates the concept of absolute perfection, which would include perfection of goodness, power, and knowledge, all the complicated entailments can come for free from this simple conceptualization. My point is, you can imagine a species of aliens, with the concept of 'Y-limited' making the same move, once you stipulate that an entity has a limit to degree n of goodness, power, and knowledge, all the complicated entailments come for free from this simple conceptualization. It's hard to see how you could give a non-question-begging response to those aliens, to convince them that we should prefer the property of perfection to Y-limitedness. You can appeal to intuition, perhaps, but the whole point is that the aliens do not share the intuition.
BB says that goodness is simple and fundamental and perfection is just unlimited goodness. The claim that goodness is a simple and fundamental property, is, of course, highly controversial, and would be rejected by moral naturalists, anti-realists, and constructivists, but we can just grant it. Suppose the aliens agree that goodness is fundamental and simple. What they will not agree with, ex hypothesi, is that unlimited goodness, power and knowledge is more simple than limited goodness, power and knowledge. It doesn't follow from goodness being fundamental that unlimited goodness is simpler than limited goodness, or at least, BB has not argued that it does. Indeed, unlimited goodness being more simple is the very thing in dispute, the aliens think that being Y-limited is more simple. So we can see that BB has simply begged the question against the hypothetical aliens. Now, of course, the aliens themselves also can't provide non-question-begging reasons to think their concept is more simple than perfection, but what I take this to show is that preferring one view over the other is arbitrary. At best, one might appeal to intuition, but if one doesn't share the intuition, as I don't, then that won't be effective.
I'll also note that BB's response here ends up just reducing to the arbitrary limits argument at the end of the day. Since apparently, what is virtuous about perfection relative to Y-limitedness is its lack of arbitrary limits. But in that case, whether one accepts the argument will be a function of whether one is convinced of the argument from arbitrary limits, there isn't an extra consideration here. Since I'm not convinced by the argument from arbitrary limits, of course, this does not move me.
Multiverse Naturalism
So we’ve concluded BB's arguments for the intrinsic parsimony of theism. None of his defenses are compelling. This also concludes our discussion on stalking horse naturalism. So all we have left to discuss is multiverse naturalism, which I've appealed to as a naturalistic explanation of our existence that's better than a mere stalking horse. BB unfortunately doesn't say much here, and what little he does say is false or misses the point. For context, in my original article I pre-empt the point that multiverse naturalism won't predict your existence, since you won't get enough people, since there are ℶ2 possible people. I say the following:One might be tempted to object that there are not enough physically realizable outcomes to lead to the existence of ℶ2 people, and yet there are ℶ2 possible people. But even if that's right, it doesn't follow that multiverse naturalism doesn't make your existence likely. You are a physical body (perhaps with some soul or spooky ectoplasm that supervenes on those physical states) and on multiverse naturalism, we should expect all physically possible outcomes to be realized. Given that an outcome involving your physical body is physically possible, we should expect it to be realized in some universe given many-worlds theory. Sure, it may not predict all possible people, since perhaps some of them are not physically possible, but it still predicts that you, a physically possible person, would exist in some universe.
BB replies:TT claims that every physically possible outcome gets realized. He notes that if this is so then you exist. But on SIA, the number of people created matters. Any other view violates the conservation of evidence and Bayesian updating, for reasons I’ve described here and here (the general arguments for SIA given here or here can be made into objections to the argument). The thing that matters isn’t the share of experiences had but the number of experiences had.So he makes the same objection we covered all the way back in the first section. You know, the one which says that the objection I made is just an objection to the SIA itself. Just as BB was wrong the first time, he's wrong yet again for the same reasons. I'm not committed to denying the SIA in making this response, just the N-SIA. But that's not a problem since the N-SIA is false.
The problem is actually more severe here when considering my above quoted argument, which BB does not directly address. The idea is, while multiverse naturalism perhaps does not predict the existence of ℶ2 people, it predicts the existence of a subset of people (all physically possible people) that is known to contain you, or at least someone physically exactly like you with all your evidence, in a situation empirically indistinguishable from yours, as a member[30]. So it entails the observed data, despite not entailing ℶ2 people. This does not threaten the bare SIA, which has an all else equal clause, and all else is clearly not held equal here because this is a case where we precisely know the subset of possible people that multiverse naturalism entails, and it's all those people that are jointly physically possible of which you (or someone physically identical to you with the same evidence) are a member. This also puts us in a position to once again see why whatever principle BB has to appeal to in order to undermine my argument, in this case the N-SIA, must be false, this time as a trivial matter of logic and Bayes theorem.
Call the hypothesis Hu the hypothesis that you exist and there are only 100 total observers. So, Hu is logically equivalent to the conjunctive proposition 'you exist' (call this Y) and 'there are 100 observers in total' (call this R). It's a theorem of propositional logic that, for any p and any q, p&q entails p, and p&q entails q. In other words conjunctions entail their conjuncts. So this means Hu (R&Y) entails Y. Which also means P(Y|Hu) = 1. Or Hu predicts your existence with a probability of 1. So when accounting for your existence, Hu gets as much of a boost through updating in line with Bayes rule as any other theory which entails that you exist. But Hu also entails R, that only 100 observers exist. Thus, a theory which predicts only 100 observers exists gets as much of an update, conditional on the fact that you exist, as any theory which predicts more observers and which entails your existence. Since, for any hypotheses H1 & H2, and any evidence E, if H1 and H2 both entail E, then E can't be any evidence for H1 over H2. Of course, this is not to deny that Hu is a bad theory and will have a correspondingly ridiculously low prior since it just contains what it explains, (a problem which, importantly, does not afflict multiverse naturalism. As there are plenty of reasons to think many-worlds theory is independently probable or so I've argued in my initial piece) but that doesn't change how much of a boost Hu gets when updating on the data. So, simply assuming standard propositional logic and Bayes theorem leads us to trivial counterexamples to the N-SIA, where theories which posit much less observers get just as much, if not more of an update as theories which posit more observers. Which would mean that even if the SIA entailed the N-SIA, BB's arguments in the linked articles would then not be arguments for the SIA, but a set of regrettable bullets we have to bite on pain of rejecting propositional logic or Bayes theorem. Fortunately, though, the SIA does not entail the N-SIA, so we can just reject the N-SIA with no problems. Since only the N-SIA would, if true, render my argument unsound, BB's response fails.
BB lastly makes mention of modal realism, saying it collapses induction. But that's irrelevant, as multiverse naturalism is not modal realism, and does not collapse induction. This is because, on multiverse naturalism, each universe is deterministic and obeys the laws of physics. There may be other skeptical scenarios that might be run against multiverse naturalism, however, such as Boltzmann brains. But insofar as that afflicts multiverse naturalism, so too does it afflict the version of theism that predicts all possible people. Since the supposed problem would be that there are much more observers that are Boltzmann brains, than there are typical observers like us. But that's a problem regardless of if you accept a God-created multiverse with every possible person, or a naturalistic multiverse with every physically possible person. Further the Boltzmann brain skeptical scenario can be run even without a multiverse, and just the fact that in the distant future, our universe is expected to reach a state of maximum entropy where random fluctuations can occur due to quantum effects or thermal processes. With the idea being, given enough time, it can be hypothesized that there will be much more Boltzmann brains as a result of these fluctuations than typical observers. Whatever response you have to that, is a response the multiverse naturalist can make use of as well. I must conclude then that BB's response to the multiverse naturalism objection is a failure.
Final Recap
At last, we are at the end, and this is our final recap.- BB starts by misframing the discussion, suggesting that I did not address the issue of stalking-horse naturalism having a prior of zero. As a matter of fact, I did address it, and it's BB that has failed to engage with my broader dialectical point.
- BB responds to some of my points regarding the mysteriousness of theism, but his objections all miss the point.
- BB argues that theism must have a prior of 1% given that the GOA and contingency arguments might be successful. This is just a rhetorical tactic that abuses how credences work.
- BB's response to my argument that theism fits poorly with our background knowledge is not clear, but I consider two possible interpretations and argue that they both fail.
- BB claims that simplicity only concerns fundamental entities. This is a controversial view, but even then, I argue that the most plausible version of this view is of no help to him.
- BB makes mention of other Bayesian arguments for theism in response to my point about the success of natural science. This both misses the point, and skims over the depths.
- BB mentions his hypothesis of indifference theodicy, so I do an extensive side-tangent to address it, highlighting my problems with BB's broader methodology exemplified in his approach to theodicy.
- BB next tries to argue for the parsimony of theism, mainly defending the arbitrary limits argument. I argue that none of his brief defenses are convincing.
- BB lastly replies to my appeal to multiverse naturalism, his first reply is the same false claim that it assumes the falsity of the SIA which I take myself to have refuted in the first section. His other reply is that modal realism implies inductive skepticism, which even if true, is completely irrelevant.
Conclusion
In sum, BB's replies to my objections to the anthropic argument have all the force of a whisper in a hurricane. Virtually every substantive counter he makes either 1) misses the point 2) repeats an objection I pre-empt and address in my original piece without mentioning my response 3) is flat out wrong 4) just makes toothless claims that do little more than simply reassert the very point in contention or 5) fails to realize the extent of his burden in defending the anthropic argument in a way that belies, frankly, an uncareful, and uncritical approach to probabilistic thinking, theory selection, and hypothesis testing. This is not surprising to me given that his response was clearly rushed, and in my view, sloppily put together as a result. This may seem harsh, but I'm not saying this to be a sanctimonious prick. Nor am I saying it because I bear ill will towards BB, it’s his bad arguments and lackluster methodology that I have a problem with. I'm saying this because I and others have noticed that this is a consistent pattern in BB's approach to philosophical engagement and I think the philosophical bar should be raised.Good philosophy is applying serious scrutiny, not just to your interlocutor's arguments but to your own. It's taking the time to deeply consider your interlocutor's perspective, to carefully reflect upon their objections, and why they were made. It's not rapidly sifting through their objections, and prioritizing putting a quick reply out above all else. Unfortunately, it is the latter, not the former that appears to be emblematic of BB's general approach.
As for myself, I remain just as unmoved and unimpressed by the anthropic argument for theism as I was when I wrote my initial piece. None of BB's attempted defenses of the argument from my objections are successful. By my lights, pretty much no naturalist ought to feel all that threatened by the argument. It is however, possible, albeit very unlikely, that I will change my mind and come to think of the anthropic argument as, at least, an evidential chip in theism's favor. So, I'll end this on a conciliatory piece of advice.
If BB wishes to seriously challenge my view that the anthropic argument is wholly unpersuasive, he will need to do better than a rushed reply. I encourage reading through this and my original post very carefully and thoroughly, and putting serious effort and time into understanding and engaging with my objections, and perspective. Before putting out this response, for instance, I read through BB's reply and my draft multiple times to check for errors, and even had a philosophically intelligent friend proofread my draft and give feedback. BB should do the same, or something similar enough before releasing a response. If he comes out of it still disagreeing, that's perfectly fine. But I at least hope for more diligent and rigorous engagement in the future.
Endnotes
[1] Some extensions to ZFC set theory such as Neumann–Bernays–Gödel (NBG) and Morse–Kelley (MK) class theories can be taken to extend the notion of “cardinality” to include proper classes. But cardinality here isn’t used in the conventional sense–they aren’t assigned a cardinal number and so it would still be false, strictly speaking, to describe proper classes as a “number of things”. All that is really being said is that you can have injective or surjective functions that map elements between proper classes.
[2] It might be pointed out yet again that under set theories such as MK or NGB proper classes are to some extent ‘treated’ as formal objects. But I take it that even if one is a platonist about mathematical objects, proper classes don’t have ontological status in the way that sets do. For starters, there are issues in the neighborhood of reference and unrestricted quantification. It's impossible to refer to ‘the class of all sets’ or ‘all ordinals’, as a whole collection, say, in a way that unrestrictedly captures ‘every’ set or ‘every’ ordinal, since you can always generate more. Further, if we treat them like actual objects with a robust ontological status, I suspect we’d just run into the very paradoxes which the construction of proper classes are designed to avoid. Consider the properties that proper classes would have if they were objects. It looks like the class of all sets would have the property of containing all properties which do not have the property of containing themselves (since for any non-self-containing property, there is some set containing that property as an element). Is that property self-containing? If it is it isn't, if it isn't it is.
[3] The N-SIA is actually quite similar to the U-SIA that BB has argued for in a separate article. Without taking as far afield, I’ll just say that I don’t think any of his defenses of the U-SIA in favor of the regular SIA are compelling, and even if they were, it doesn't matter because we have decisive reasons to think the N-SIA is false which would apply equally to the U-SIA insofar as it has the same entailments.
[4] I have a suspicion that this may be a problem for an intuitionist epistemology in general, not just BB. Though, perhaps while intuitionists can appeal to a plausible story for which intuitions count as justifiers, BB has done no such thing to show that his intuitions count but the ones I appeal to don’t.
[5] In fairness, BB does say that, if the update cancels out in the way I suggest it does, this would allow us to derive the absurd results of the SSA. But why that is, is unclear, and BB hasn’t argued for it.
[6] Such an approach is, however, itself, not without substantial costs. As I’ve argued elsewhere.
[7] See this for a rundown of the basics.
[8] See Bas Van Fraassen’s critique of the prominent explanation by postulation in metaphysics (2002)
[9] For some examples, you have David Chalmers’s primary and secondary conceivability, as well as prima facie and ideal conceivability (2009). Stephen Yablo’s notion of conceivability (1993). And a bunch more (Spencer 2018).
[10] See also Mark Walkers paper for some good intuition pumps.
[11] Not long after writing that, BB would happen to express to me in a direct correspondence that he doesn’t accept such a view of God. This would be to reject Anselmian theism. However, such a move is not without its costs as it also strips away many of the features BB finds attractive about perfect being theism. To start provability, ontological arguments, including contemporary Godelian ones would prove an Anselmian maximally great being exists if successful. The same is true for many cosmological and contingency arguments which if successful, proves the truth of classical theism. The same problems for divine creation and more can be run against classical theism. It also undercuts BBs arguments for the simplicity and elegance of theism such as lack of arbitrary limits. Being unable to instantiate a perfect existence unless one has the relational property of existing alongside other good things is an arbitrary limit, so far as I can tell, on any plausible account of arbitrary limits.
[12] See here and in particular (Dolezal 2023).
[13] See also (Rowe 1994) for an argument along similar lines.
[14] BB also mentions Timothy Williamson’s necessitism. I don’t know much about that view but from what I do know its relevance is opaque to me.
[15] See Gregory Dawes' Theism and Explanation for a rundown of 'informativeness'. [16] See here.
[17] See my post, this video by Joe Schmid, and this paper by Peter Van Inwagen.
[19] Pruss has a response to this but it's unconvincing. See my friends discussion here.
[20] The main one's I know about are the arbitrary limits argument and some arguments Alexander Pruss gives here. Joe Schmid has videos where he discusses those, giving objections that I think are good.
[21] See section 1.3 and 1.4 here for a more thorough critique of BB's atomistic approach
[22] Credit to a guy on discord, user "Reductio" for giving me this argument.
[23] See here.
[24] The theist can respond that the fundamental physical entities posited by1naturalism aren’t fundamental on their view and hence they aren't technically multiplying fundamental entities, but this strikes me as a shallow victory. The theist is still committing to the same fundamental physical entities that the naturalist is committed to, in addition to positing a non-physical yet-more-fundamental entity.
[25] Some idealist theists propose a monist ontology, but I think that will suffer from other issues in the neighborhood of content-externalism and explanatory power.
[26] I'll note that insofar as one takes this to undermine 'noseeum inferences', that's just skeptical theism. BB implies that the problematic entailments charged of skeptical theism come from there being 'unknown reasons for more than one thing'. But that's not a correct account of the relevant literature. The problems with skeptical theism, such as radical skepticism, come from our inability to reason from God's known reasons to His total reasons.
[27] See here. Also see here for Van Inwagen actually accepting that theism is empirically untestable.
[28] I take this to be pretty implausible. The value of shared struggles in a relationship seems to me, instrumentally valuable. Suppose shared struggles didn't lead to strengthened bonds and a deeper appreciation of each-other in the face of adversity, then we wouldn't think those struggles are good qua relationship. Even if a husband and wife have a better relationship in this world in part due to their shared struggles, that gives us no reason to think their relationship couldn't be just as valuable if not more, absent those struggles in an adjacent logically possible world.
[29] I'll also note that it's weird that BB would think that the laws of nature being non-uniform is a priori unlikely considering he endorses the nomological harmony argument, which I take it is committed to taking the uniformity of laws of nature to be a surprising fact in need of explanation.
[30] In my original piece I explain why we can't update in favor of you not being a person in a situation empirically indistinguishable from yours, as your evidence would be exactly the same.
Good response, I warned him not to do a response super quickly because important things would have been missed which is what happened (at least somewhat), good philosophy takes time to write and when doing responses to criticism it's important to take into account the underpinnings of the opponents arguments.
ReplyDeleteRegardless I wanted to comment because Me and Howard have done a *ton* of research into the simplicity of theism & there is simply no good reason at this point to think that theism is positing a supernatural entity. The work of Joshua Sijuwade has showed that we can understand God in terms of being a maximally powerful module trope, and tropes are within naturalized metaphysics and thus there is at least one view of theism that is naturalized. I don't know if Matthew would hold the same exact view but me and Howard are theistic naturalist and so do not think that ontological Simplicity is a problem for theism. I know that was a small section and not relevant to your main criticisms to Matthew (in fact it's not relevant to the anthropic argument at all so I'm confused as to why you mentioned it), but there is zero reason to think that theism has more qualitative baggage than naturalism. Whether naturalism can successfully parot theism is a *different concern*, but here i think you and Matthew can share in being "ontological naturalist" even if Matthew happens to be a "theistic naturalist" and you a "indifference naturalist"
Thanks for the comment.
DeleteI guess I don't understand how God being understood as a trope solves anything. It doesn't follow from God being a trope that theories which posit the existence of God aren't more ontologically committing. Someone could accept a metaphysical view on which ghosts, goblins, or spiderman are just bundles of tropes if they exist. Would it be well and good for a ghost-believer to say "Ah, but you see, I'm not bloating my ontology by positing the existence of ghosts. I'm just saying there are a collection of tropes. Tropes already fit well within our best metaphysics". Surely not! Even if God is a trope there's still a sense in which you could still be committing to the existence of a wholly other kind of trope with specific features or entailments, that I'm not committing to.
I'm also not sure how interpreting God as a trope makes this a naturalistic view. While tropes themselves can be consistent with naturalism, I take it there is much more loaded in the concept of God than just "trope". If God is non-embodied, not located in space and time, and is a mind at the foundation of reality, causally or ontologically prior to our universe, then this isn't really a naturalistic metaphysic in any meaningful sense that I'd accept.
Hi TT, thanks for your response, I won't say much beyond what was discussed here but I highly recommend you read into Sijuwades work. I know the book is $180 but trust me it gets into the details better than I can explain here.
DeleteGod can be understood as fundamental "physical" trope, acting from a basis of normative goodness. This conception bridges the ostensible gap between naturalism and theism at least in principle. Furthermore we can consider goblins, or spiderman as "physical" also. In fact if you read the work of David Lewis his argument for modal realism was using ontological Simplicity in that he didn't need to posit anymore "kinds" of entities beyond what we consider to be "concrete", so goblins or Spiderman *would* technically exist on his framework without jeopardizing at least ontological Simplicity. Regardless though the reason we do not typically accept the existence of ghosts, goblins, or spiderman is because they go against our background knowledge, it has very little to do with concerns of ontological Simplicity
Moreover, a lawless interpretation of wave function monism proposes a fundamentally physical reality endowed with purposive power and a propensity for goodness. Such a thing would not be a different "kind" of entity beyond what we see the wave function do in physics, the only difference is the behavior of this wave function as acting towards what is good. The "kind" of things this is would already be consistent with contemporary quantum theory. Also, if you read the work of what tropes actually are and how they function they are the simplest kind of irreducible entities that we know of in contemporary metaphysics. Of course one could say that a 'personal' trope not fitting with established background knowledge may be a challenge, this can be countered by emphasizing the significance of the type/token distinction. This distinction allows for the acknowledgment of new instances within a recognized class of entities, thereby aligning the concept of God within the accepted ontological structures of contemporary metaphysics.
Considering background knowledge more broadly, if we define it as encompassing all our worldviews, we risk diverging notions of what constitutes this knowledge. Whether we view background knowledge as a holistic compilation of relevant metaphysical fields or as sequential evidence accumulation, the alignment of Theism with such knowledge remains unchallenged.
I haven't read Sijuwade, but I have a few points to address.
DeleteIt's not clear what you mean by "physical" when you say God is a "physical trope" or what this adds. Are you suggesting that God is spatiotemporally located? Can we, in principle, causally interact with or detect God?
I genuinely don't follow the wave-function monism point. A wave function is just a mathematical description that gives the probability of outcomes of a quantum measurement related to a particle or system of particles' position, spin, or momentum. Saying a wave function acts towards what is good seems like a category error—applying reason-based action to a mathematical function doesn't make sense.
Pointing out that modal realism only posits one 'type' of entity, where the type is as broad as "concrete entity," doesn't really affect 'ontological simplicity' in the sense I care about. Every view posits only one 'type' of entity in some broad sense, namely the type 'entity.' If ontological simplicity is so malleable, it becomes useless as a metric of theory-choice.
'Types' do matter in some sense. Positing one more atom in the universe isn't really a cost because we already have very good reasons to posit quadrillions of atoms. Also atoms accord well with our existing knowledge, many have been detected, observed, and interacted with, etc. Thus, 'background knowledge' and 'ontological economy' intersect.
Again, you can accept a metaphysical view where everything is tropes and where tropes are simple and irreducible. However, just because the 'type' trope is already posited in contemporary metaphysics, it doesn't follow that there's no additional cost incurred from having any collection of tropes in your ontology. Fairies might be a collection of tropes if they exist, but there are still good reasons to exclude fairies from your ontology.