"A limitless mind is simple—a mind with all sorts of arbitrary desires and powers is horrendously complicated".
I don't understand why you think we can just assume this. You seem to think a limitless mind is simple just because you can explain it in a few words. In truth we have no idea what a limitless mind would look like and it is likely that any set of laws that could allow such a mind to exist in a given realm would be extremely complicated.
By contrast, we can be pretty certain that merely powerful (not omniscient) minds are possible under the same physical laws in our universe. Given stuff like the problem of evil, wouldn't it make way more sense if God was extremely powerful but not infinitely so? And that he tried to make a universe with lots of conscious, happy entities but did not fully realize that vision?
Hard agree. Imagine making this statement about something slightly analogous to a mind: "a limitless computer program is simple--a computer program with all sorts of arbitrary capabilities and constraints is horrendously complicated". This strikes me as... less than obvious.
Why minds should be different I think deserves a serious explanation.
For better and worse, the argument there isn't persuasive because it makes untenable claims that the universe's physical laws could be modeled by a Turing machine well enough for an Invariance Theorem type argument about its K-complexity to hold. (Not that this makes BB's argument more persuasive, it just makes the whole debate more of a tarpit.)
But then that just means BB’s argument about God being “simpler” are meaningless and we have no idea of a universe with God is simpler (in a sense that affects its plausibility) than one without!
As I said, it doesn't make BB's argument more persuasive. To me the whole thing seems to be just a tarpit where people are misusing concepts and principles willy-nilly on all sides.
I disagree. By your account BB is the only one misusing concepts. The people he’s arguing with are using the concept of complexity correctly; you just don’t agree with their premises (and even then I think you might be misstating their premise; they don’t necessarily argue that you COULD program the universe into a Turing machine, they’re just positing that if you could, building a god into it adds complexity because then you have to know how to program a god). But BB is misusing the concept of simplicity and then acting as though it actually impacts the plausibility of theism.
Yes, I knew you disagreed, and by my account both sides are misusing concepts (I said so pretty explicitly though I didn't go into depth on the nature of the misuse). I do not agree that the people BB's arguing with are using complexity correctly. They correctly point out that BB uses it incorrectly, then also use it incorrectly themselves. There are many, many ways to be incorrect.
What I see going on is that two incompetent tribes are cargo-culting the language of statistics and computation oppositionally to one another while both either shouting down or ignoring the people who try to point out the errors on both sides.
I don't think it because it can be said in few english words. I think it because I'm a dualist--I think minds are fundamental and irreducible. So instantiating a fundamental thing without any limits strikes me as very simple and non-arbitrary.
So a non-dualist could reject this proposition? Many atheists - such as myself - are non-dualists.
I would also say that the idea of something with no limits does not sound that coherent to me. No limits in every domain? In which case, this mind would not be limited even by its own nature. It would be able to - for example - put in place limitations on itself. And why wouldn't it do random things, like creating and then destroying universes and then doing cruel things and then doing kind things? Its thoughts would either have to be random or it would have no thoughts at all, surely, as any structure would be a limitation - indeed, any thought would be a limitation?
I don't have a non-trivial credence in dualism. I think dualism is inherently contradictory, and so cannot possibly be true.
I also feel this way about a good deity. The animal suffering that you talk a lot about, for example - and with which I agree with you on on many points - is just something that I would not expect to see a priori if the universe was created by an entity that cared. And it seems that you would agree, as you - as a caring entity - want to reduce this suffering. That implies that you care more than this supposed perfectly good deity. Which is impossible, as this deity is perfect. That implies that either your caring is actually the opposite of good or the perfectly good deity does not exist.
Perhaps it is more accurate to call it incoherence, but I start from the claim that the non-physical can interact with the physical. That seems to completely remove the distinction between two.
The claim that a limitless mind is simple comes from theism’s metaphysical framework, where simplicity does not refer to cognitive or structural complexity, but to the absence of parts or internal composition. A limitless mind (God) is thought of as a unity, indivisible, and without contingent or arbitrary features.
I don’t think the concept of God is simple because it can be explains in less words, just that a limitless God would not have arbitrary limits that a creator that isn’t created wouldn’t have.
‘New Atheists’ aren’t really against Theism,but against Christianity,it’s why Dillahunty’s commenters were mocking virgin birth,YEC,etc. in the live chat instead of engaging with your arguments. Their self assurance comes from their confidence in Christianity’s falsehood,particularly fundamentalist / Calvinist (Baptist) Christianity (which I’m nearly certain is true)
I think it would do you a lot better to elaborate on your view of god instead of your argument for god in debates with non-intellectuals & their fans,who would then be more open to hearing you out
Ah yes, another patient who insists the X-rays are wrong because they don’t feel any pain. Let’s scrub through this plaque-ridden rebuttal, shall we?
🦷 “‘New Atheists’ aren’t really against Theism, but against Christianity.”
That’s not bizarre — it’s just diagnostic. If a dentist notices most of the cavities are in molars, it doesn’t mean incisors are immune — just that decay tends to show up where people chew the hardest. Likewise, New Atheists may claim to reject all theism, but their dental tools are almost always aimed squarely at Christianity. The most vigorous scraping happens where the buildup is culturally thickest.
🪥 “Their self-assurance comes from their confidence in Christianity’s falsehood.”
Which is like saying someone brushing their teeth aggressively must be right about fluoride conspiracy theories. Confidence isn’t the issue — basis is. The point is that their self-assurance is often directed at one tradition, despite claiming to represent a universal solvent. It’s like offering mouthwash for a full-body illness.
🦷 “Should they feel embarrassed of their beliefs when they conclude Christianity is bogus?”
No, but they should stop acting like their diagnosis is peer-reviewed when all they’ve done is poke a tongue depressor at theology. You don’t get to play periodontist just because you own a toothbrush. If they’re confident, fine — but let’s not confuse snarky certainty with sound philosophy.
🪥 “It would do you better to elaborate on your view of God rather than argue for God in debates with non-intellectuals.”
This one’s got the bite force of someone chewing aluminum foil. The suggestion is that unless your opponent has academic credentials, you should ditch argumentation and just share feelings. That’s not how any kind of inquiry — medical, philosophical, or dental — works. When explaining periodontal disease, I don’t say “well, I just feel like gums are important.” I show evidence. Likewise, debates aren’t therapy sessions; they’re exams.
And yes — calling someone a non-intellectual in that context is like accusing your hygienist of being a plumber because they didn’t quote Nietzsche during your cleaning. It's condescending, shallow, and completely misaligned with the spirit of rational inquiry.
Let’s clean this up and get you fitted with a new retainer — because this whole argument needs structural support.
Perhaps because certain people tend to take everything ChatGPT says at face value and readily accept it without doing any thinking of their own, and there's a reaction to these kinds of people by others who are so disgusted by the notion of just treating ChatGPT as a prophet that they fail to consider that it can be still be used to generate arguments that should be seriously examined (we can verify whether the math proof generated by ChatGPT about the probabilities was correct or not, rather than just accepting the result given by it).
I think the only way to react this way from the way it was mentioned in the debate is if Dillahunty wasn't really paying serious attention to anything said and was just hunting for stuff to respond to (which is a common and effective tactic for internet content creators despite its lack of argumentative validity). In context it's immensely obvious BB is saying to ask ChatGPT in the same way as saying "it's trivial, just Google it and verify that everyone agrees that this is how the math shakes out instead of wasting everyone's time making an argument against Bayes' theorem that you clearly haven't thought about prior to this conversation".
I’m really interested to see how chatGPT affects online arguments. On its face, it’s a godsend: ChatGPT is always willing to read an argument, no matter how stupid, tedious, or petty the argument is, and it can often tell you which side of the argument is stronger. Manipulating it seems difficult, so long as everybody can see the prompt. But obviously ChatGPT is not infallible - and maybe that’s enough to keep it from being seen as a good arbitrator by most.
I found the fact that his audience was so receptive to it to be quite depressing, given how transparently it was just chum for social media content and entirely immaterial to anything being discussed
Do you think there are objective facts about what the correct prior for a particular position is? You speak of God having a high prior probability but how is this established? The typical way I think of priors is as descriptive facts about a person's level of confidence in the truth of a proposition. A prior could be anything from 0 to 1. I don't generally think of there being a fact of the matter about what one's prior ought to be nor do I think that there are facts about what the prior of a given proposition are independent of any particular agent's stance on it.
You could argue that priors can also be informed by objective considerations, such as simplicity, explanatory power, and coherence with known facts. These considerations allow us to reason about what priors "ought" to be, even if there is no single objectively correct prior.
It’s possible that atheists don’t engage with theist arguments because, often, the arguments seem vague and loose, likely incorrect, but hard to argue because of insufficient specificity. For example, you say here: “it’s likely that God exists because God is a mind of the very simplest sort—one totally without limits”. An atheist would say “there is nothing ‘simple’ about an all powerful mind. We have no concept of a plan how such a mind would be constituted nor how it could exist nor even what ‘all powerful’ entails in terms of specific powers.” Consequently, when you start your argument that way, you lose the people that don’t already presuppose the existence of god.
Empty map doesn't imply empty territory. Being ignorant about parts doesn't mean that something is partless. And trying to come up with a notion of something that couldn't possibly be constructed from any parts, doesn't necessary give you something simple, more likely than not you are getting something impossible and infinitely complex.
But we’re not talking about a being that might have parts. We’re talking about a being that necessarily doesn’t have parts.
In this context, ‘simple’ is a technical term that comes from mereology. ‘Simple’ just means ‘no parts.’ Now, you might say that doesn’t track ordinary usage, but it’s not trying to.
> We’re talking about a being that necessarily doesn’t have parts.
I've already answered it here: "And trying to come up with a notion of something that couldn't possibly be constructed from any parts, doesn't necessary give you something simple, more likely than not you are getting something impossible and infinitely complex."
But let's go a bit more into details. What we are dealing with is God being simple "by definition". Like we define God as:
1. Omnipotent, omnibenevalent, omniscient and omnipresent mind
2. Necessary doesn't have any parts
Which may seem as a clever trick untill you realise that you can't change reality by picking a different definition. Anything fitting this definition would be simple in the sense of not having any parts - that's true. But now, as mathematicians do, you need to prove existence - that there is at least one entity that fits, that you definition is not itself self-contradictory. Otherwise, you have logically pinpointed something impossible, and therefore infinitely complex.
Well, I don’t see how ‘impossible’ entails ‘infinitely complex.’ I think it is impossible for there to be a smallest physical part, and yet a physical simple wouldn’t be infinitely complex—it would be simple!
Hi Robert - thanks, your "partless" comment is an excellent illustration of my point: this argument is "vague and loose, likely incorrect, but hard to argue because of insufficient specificity". What specifically is a "partless mind"? How specifically does a disembodied "mind" effect "power" in the universe? How specifically does partlessness yet all-powerfulness imply simplicity and a high prior probability? If your answer to any of these questions is that "god is beyond human understanding", then what is the justification to claim that something this sophisticated is "partless" and "simple" and "high prior"?
I’m not defending the argument, since it’s not my argument! The complaint was that ‘simple’ isn’t a well-defined notion. But giving a formal definition of simplicity is very easy in classical extensional mereology, and so I don’t think the complaint that ‘simple’ is ‘likely incoherent’ is a very strong argument.
Hi Robert - thanks for the link. I've read it, and read many of BB's other articles. I enjoy them all. However, as a skeptic, I find baseless (to me) assertions, large and small, throughout. Sometimes I comment on those issues.
The goal of BB's arguments, I assume, is some combination of a) convincing non-believers of god's existence; b) reinforcing the beliefs of already-believers; and c) monetizing the topic. I don't doubt or begrudge the success of (b) or (c).
My original and continued point here is that BB is failing on (a). If you do not already believe in god, you won't be much persuaded by "it is self-evident that god is a limitless all-good partless mind and is so simple that it must have the highest of all priors". I have not yet read in BB's many fine articles, any rationale for why I should take this as true.
To be clear: I'm not arguing AGAINST the existence of god. I'm suggesting that this argument FOR god's existence is not at all effective with non-believers, and for good reason.
Lastly, of course one can define "partless" via mereology. That's not the issue. It's the "god is partless limitless all-powerful simplicity" that needs some justification. It's just not self-evident nor convincing to anyone who is not already a believer.
What is a "part" in the context of a mind? How does that play with "limitless"? In the cases where I can think of well-defined simpleness-as-partlessness like finite dimensional algebras, the resulting simple objects don't strike me as in any way "limitless"--the matrix algebras M_n aren't in any way "less limited" than direct sums of them.
If anything, in the case of minds, the opposite strikes me as more likely: a limitless mind should be able to think about multiple things at once; to model multiple other minds; or to do all sorts of other potentially "parallel" processes and it seems pretty natural to imagine that the different thinking pieces of a limitless mind decompose it into parts: the part that's calculating the fall of a single sparrow; the part that's experiencing suffering as an embodied person on the cross; etc.
I agree with you on the merits of your arguments. I think it'd be better if you didn't include so many petty insults though. Standards of discourse are also important.
I was a regular viewer of The Atheist Experience back when I was a young atheist in the early 2010s. Mostly it was just a source of entertainment, seeing Matt effortlessly destroy all the clueless Christians who would call in. But it also reassured me that my worldview was correct and its opponents were dumb. He’s much like a televangelist that way.
It’s no surprise then that when an actually intelligent theist shows up the reaction from Matt and the audience is not so much curiosity as annoyance that their usual song and dance routine has been so rudely interrupted by arguments that have to be seriously contended with.
Great review! You did well in the debate, and I enjoyed reading this follow-up as well. That rant about the new atheist disdain for rationality was seriously based -- I agree with you entirely.
Man, I wish I had theist friends who were philosophically and mathematically inclined like this. It’s difficult to find people you can openly discuss arguments like this that doesn’t fall into them getting stuck on a random objection that’s about them not understanding the argument. There’s definitely force to the arguments from consciousness and anthropic arguments, imo, though I’m still an atheist.
Plus, with friends you can spend time having good faith discussions about what the premises and arguments say and don’t say before starting the argument — a debate like this only really works when both sides understand the other’s arguments well, so it’ll only work between people who are either very smart, well read, and done a little research on people’s position, otherwise it’ll be a blowout like this was.
If you debate Scott Alexander, I will be there instantly.
Really entertaining debate. It's kind of nuts that he has been debating these things for like 20 years and still hasn't managed to get basic epistemology concepts right.
I think BB has something like the Solomonoff prior in mind here, where your sample space is finite-length binary sequences representing different hypotheses, with the Kolmogorov complexity of each sequence relative to some Turing machine as the prior probability. That is, you imagine each hypothesis as a class of "computer programs" that can run on some universal Turing machine, and the length of the shortest program that implements the hypothesis is related to its prior probability.
So for example, roughly speaking a hypothesis of the form, "this is a sequence of all 1s of length 10" is more probable a prior than "this is a sequence of length 10 of the form 0111001110" because you can write "print '1'*10" for the first one, but you can't do too much better than "print '0111001110'" for the second.
I do, however, think that encoding concepts like "God", "mind", "limitless" (in this context) are... of dubious simplicity if actually converted into this formalism. I also am really not sure why it's at all clear that "mind without limit" is a good definition of God. Or, for that matter, whether the God hypothesis can be computed by a Turing machine.
I think BB is applying Solomonoff-type reasoning in a situation where it's really not clear if it applies, and if so, how to model it.
Combine these objections with the uncomputability of the Solomonoff prior, and I agree that we're not really doing anything usefully described as "probability" here.
Why think he's referring to the Solomonoff prior? After looking it up, it doesn't appear reminiscent of anything I've seen BB say before, and it seems like this would introduce commitments that BB wouldn't want to accept.
I think he's done so before and will doubtless do so again, but in this instance I think he wanted a chance to show the more lucid parts of Matt Dillahunty's large and somewhat dogmatic audience that there are non-stupid arguments for theism, which they may have low exposure to.
I think it's a noble pursuit but unlikely to be effective. The snide comment made by someone (who I can't recall now) that atheism is a Christian heresy is I think about the fact that atheism as a popular movement is primarily about being in conversation with Christianity and specifically a very narrow kind of low church protestantism that the adherents of the Atheist movement grew up around. The question of whether this religion is false is, to them, the only interesting question about this matter at all. They are not sincerely interested in the question of whether God exists as a philosophical or academic matter, but actually as a battle in social politics against the groups they identify with theism.
So many things are wrong with this article in terms of what actually happened, but I'll just point this one out for anyone confused. It was clear to me that Dillahunty went "ballistics" when Matthew asked him to consult ChatGPT because it came off as smug and (obliquely) intellectually derogatory. The two had just disagreed on their probability arguments when Dillahunty started to type and Matthew said something to the effect of, "yeah, go ahead and ask your Chat GPT", which carried the suggestion that Dillahunty wasn't smart enough to understand Matthew's gigabrained contentions.
Unless Matthew is just bad at reading the room, it's obvious that this was what upset Dillahunty. Asked at a different time, it might have come off differently. But if we have just disagreed and I'm typing on my computer and you, without knowing what I'm actually doing, say, go ahead and ask A.I, I'll assume you a.) think I'm not smart enough to understand your arguments, and/or b.) don't think I'm confident enough in what I'm saying to actually defend it upon attack without consulting A.I.
Not that there's anything wrong with consulting A.I for literally anything. But this context is important.
I've made a few critical comments about the Solomonoff-style prior argument for God, but I also want to register (one of) my objection(s) to the fine tuning argument.
Something like the cosmological constant, which is related to the vacuum energy, and presumably may be implicated in any future unified theory, or theory of quantum gravity, or whatever, can't be uncomplicatedly treated as a free parameter.
Imagine making a fine tuning argument from the value of pi: if pi were too much smaller, the volume of stars would be too small and all stars would collapse into black holes. Putting aside all the other problems, it's really doubtful whether you can reason this way about *mathematical* facts; and I think one ought to be *very* suspicious of Bayesian reasoning that treats the value of mathematical constants as either "observations" or "evidence".
Now, I don't *know* that the value of the cosmological constant will turn out to be a mathematical consequence of some other facts about the universe, or if so which ones, and whether there will be a canonical way to think of some of those parameters as "free parameters"; but if that were to happen we might well learn that the value of the cosmological constant is not actually a free parameter at all. It is more like the value of pi, a mathematical fact, or at least, a mathematical fact conditioned on some other physical facts.
Now, this doesn't mean that we can refute fine-tuning arguments by just *asserting* that actually, the value of Lambda is just as mathematically determined as the value of pi, but I think Bayesian reasoning about the difference between these views is complicated by the fact that almost all of the disagreement is in model uncertainty: *if* Lambda is a free parameter, then the value of Lambda is good evidence of fine-tuning, but if it's not, then it's not. So the probability for your fine tuning argument has to factor through the probability of whether we should model Lambda as a free parameter, which is a much more thorny question. It's, essentially, a question about "how should we model the process that generates the underlying laws of physics", which is not something I think we have a good handle on.
I don't know exactly what the takeaway is here; like I say I don't think this refutes fine tuning.
But I think it points to the fact that a lot of the difference between the theistic worldview and the atheistic worldview probably comes down to questions about "how to model how the physical laws were generated", and it's *much* harder to know how to get evidence for that question.
I agree that Matt did horribly (as 100% expected). Regarding the error about Bayesian evidence, I'd have to go back and review the transcript, but I remember it being possible while I watched it live that he made a semantic confusion. He was saying at one point something of the form "Just because a theory T predicts evidence E more than ~T, the evidence doesn't make the theory more probable." But this could be parsed in two ways: either that it doesn't make T more posterior-probable than it was before learning E (which is false), or that it doesn't make T more posterior-probable than ~T overall (true). But he might've said extra stuff in the conversation indicating the first, unfavorable interpretation. I honestly kind of tuned out during a lot of his talking because it was hard to listen to. So I don't know.
>Matt asked what it means to be all knowing. Answer: to be all knowing means to know every truth.
Not that Matt brought up this objection, but it's not actually clear this definition is even logically consistent! Consider the knower paradox[1] applied to God, i.e., the sentence "God does not know this sentence is true." Naively, it's easy to prove that God doesn't know it, so it's true, so there's a true statement God doesn't know.
You might want to say this sentence actually lacks a truth value, perhaps via failing to pick out an actual proposition because of self-referentiality, and that omniscience only refers to knowing true propositions. OK, fine. Call sentences that successfully pick out propositions "proposition-apt," and call those that successfully pick out *true* propositions "true-proposition-apt." But it seems like *the sentence that a particular other sentence is or isn't known by God to be true-proposition-apt* ought to be the sort of thing that's determinately true or false, i.e., it's the sort of sentence that ought itself to be proposition-apt.
Then by a (very slightly figurative) usage of Gödel's diagonal lemma, we can cook up some sentence φ which is provably equivalent to the sentence ψ = "it's not the case that that God knows that ⌜φ⌝ is true-proposition-apt."[2] Since this new sentence ψ is asserting something about God's knowledge of true-proposition-aptness of a specific sentence, it is itself proposition-apt, hence either determinately true or false. Since it's provably equivalent to φ, we should also get that φ is proposition-apt, hence either true or false. *Now* we can run an analogue of the knower paradox to generate a problem.[3]
You can next try to escape this paradox further by abandoning classical logic, but this gets very technical, and the exotic thing you get at the end of this process won't obviously allow you to quantify over all determinate truths such that it will make any sense to talk about knowing "all" of them.
[2] Here, ⌜φ⌝ represents taking the actual content of the sentence we've denoted by "φ" and putting it all in quotes.
[3] If ψ is false, then God knows that φ picks out a true proposition, so the sentence φ is true. But then the equivalent sentence ψ is also true, a contradiction to our assumption. So ψ is in fact true. But then the equivalent sentence φ is also true, and so ⌜φ⌝ is in fact true-proposition-apt. So ψ is truthfully asserting God's failure to know a true fact. (And we can do these moves because we're guaranteed up-front that both φ and ψ are proposition-apt, so we can ignore indeterminacy.)
Haven't taken the mental effort to digest this particular paradox of omniscience, but I think in pretty much every case you can get precise analogues by replacing "it's true that" with "God knows that."
Yes, and due to the semantic paradoxes, it's in fact it's highly controversial that you can both coherently (= there is a unified global truth predicate which can be applied to all sentences, as opposed to an infinite nested hierarchy of distinct truth predicates) and consistently (= there are no contradictions with doing so) talk about "all truths," as I tried to get at at the end there.
Although I don't expect you to have time to read or digest this paper, here[1] is an overview of recent technical work on the semantic paradoxes that goes over several approaches and IMO implies that all seemingly fall prey to one of those two pitfalls.
To be clear, I didn't mean to suggest that there's no chance any paradox of omniscience works. I just meant that brazenly asking what omniscience means isn't a convincing objection to theism.
I agree, and of course Matt Dillahunty didn't have any interesting objections to omniscience to raise, so I didn't mean to imply you said anything egregious during the debate. Just voicing some personal side thoughts.
Yes, this is exactly right: BB implicitly models various physical parameters as free parameters that can take on any value in a range that is physically meaningful given our current understanding of physics. But whether or not this is a good way to think about how the universe came about is more or less the issue under discussion! I don't think you necessarily need to see other instances to calibrate, but you need to have some idea of what *what the space of possible values looks like*, and I think the best answer is, we don't know.
I'd like to especially highlight that the bounded asymmetry argument another commenter gives elsewhere as exactly analogous as an argument against God: let a, b, c be some measures of God's potence, science, and benevolence--we have now parameterized God, and if we assume that those values can have taken any value in their range, then we can immediately start constructing fine-tuning arguments against God: why does God have maximal benevolence, when if we pick a real number from an interval uniformly at random, the probability of it achieving its maximum is literally 0??
But the obvious answer from theists is: this is a dumb way to model God, like potential beings all come from a factory with preset knobs for how powerful, knowledgeable, and good they are, and then the dials are twisted at random. The theists' argument for God conceptualizes the probability space in a completely different way. But atheists have the same objection against modeling universes in this way.
And that's fine, but it means most of the argument should be about *what is the probability that each side's way of modeling this is better*? Anything that doesn't address that is not going to be persuasive to the other side; it's essentially saying "conditional on my caricature of your worldview being correct, your worldview is dumb". All the work is being done in the initial conditionalizing.
"A limitless mind is simple—a mind with all sorts of arbitrary desires and powers is horrendously complicated".
I don't understand why you think we can just assume this. You seem to think a limitless mind is simple just because you can explain it in a few words. In truth we have no idea what a limitless mind would look like and it is likely that any set of laws that could allow such a mind to exist in a given realm would be extremely complicated.
By contrast, we can be pretty certain that merely powerful (not omniscient) minds are possible under the same physical laws in our universe. Given stuff like the problem of evil, wouldn't it make way more sense if God was extremely powerful but not infinitely so? And that he tried to make a universe with lots of conscious, happy entities but did not fully realize that vision?
Hard agree. Imagine making this statement about something slightly analogous to a mind: "a limitless computer program is simple--a computer program with all sorts of arbitrary capabilities and constraints is horrendously complicated". This strikes me as... less than obvious.
Why minds should be different I think deserves a serious explanation.
This has been pointed out to him! A shame that he doesn’t seem to have absorbed it. https://open.substack.com/pub/onemanynone/p/benthams-bulldog-doesnt-understand?r=7ednr&utm_medium=ios
For better and worse, the argument there isn't persuasive because it makes untenable claims that the universe's physical laws could be modeled by a Turing machine well enough for an Invariance Theorem type argument about its K-complexity to hold. (Not that this makes BB's argument more persuasive, it just makes the whole debate more of a tarpit.)
But then that just means BB’s argument about God being “simpler” are meaningless and we have no idea of a universe with God is simpler (in a sense that affects its plausibility) than one without!
As I said, it doesn't make BB's argument more persuasive. To me the whole thing seems to be just a tarpit where people are misusing concepts and principles willy-nilly on all sides.
I disagree. By your account BB is the only one misusing concepts. The people he’s arguing with are using the concept of complexity correctly; you just don’t agree with their premises (and even then I think you might be misstating their premise; they don’t necessarily argue that you COULD program the universe into a Turing machine, they’re just positing that if you could, building a god into it adds complexity because then you have to know how to program a god). But BB is misusing the concept of simplicity and then acting as though it actually impacts the plausibility of theism.
Yes, I knew you disagreed, and by my account both sides are misusing concepts (I said so pretty explicitly though I didn't go into depth on the nature of the misuse). I do not agree that the people BB's arguing with are using complexity correctly. They correctly point out that BB uses it incorrectly, then also use it incorrectly themselves. There are many, many ways to be incorrect.
What I see going on is that two incompetent tribes are cargo-culting the language of statistics and computation oppositionally to one another while both either shouting down or ignoring the people who try to point out the errors on both sides.
I don't think it because it can be said in few english words. I think it because I'm a dualist--I think minds are fundamental and irreducible. So instantiating a fundamental thing without any limits strikes me as very simple and non-arbitrary.
So a non-dualist could reject this proposition? Many atheists - such as myself - are non-dualists.
I would also say that the idea of something with no limits does not sound that coherent to me. No limits in every domain? In which case, this mind would not be limited even by its own nature. It would be able to - for example - put in place limitations on itself. And why wouldn't it do random things, like creating and then destroying universes and then doing cruel things and then doing kind things? Its thoughts would either have to be random or it would have no thoughts at all, surely, as any structure would be a limitation - indeed, any thought would be a limitation?
All you'd need is a non-trivial credence in dualism. You shouldn't treat it as an either/or thing but instead a way to anchor priors.
I don't have a non-trivial credence in dualism. I think dualism is inherently contradictory, and so cannot possibly be true.
I also feel this way about a good deity. The animal suffering that you talk a lot about, for example - and with which I agree with you on on many points - is just something that I would not expect to see a priori if the universe was created by an entity that cared. And it seems that you would agree, as you - as a caring entity - want to reduce this suffering. That implies that you care more than this supposed perfectly good deity. Which is impossible, as this deity is perfect. That implies that either your caring is actually the opposite of good or the perfectly good deity does not exist.
What contradiction is entailed by dualism?
Perhaps it is more accurate to call it incoherence, but I start from the claim that the non-physical can interact with the physical. That seems to completely remove the distinction between two.
I guess I would need a better understanding of what you include in the set of 'limitations'.
The claim that a limitless mind is simple comes from theism’s metaphysical framework, where simplicity does not refer to cognitive or structural complexity, but to the absence of parts or internal composition. A limitless mind (God) is thought of as a unity, indivisible, and without contingent or arbitrary features.
I don’t think the concept of God is simple because it can be explains in less words, just that a limitless God would not have arbitrary limits that a creator that isn’t created wouldn’t have.
Why is having a limit more arbitrary than being limitless?
For the same reason being infinite is less arbitrary than being a specific finite value - begging the question
‘New Atheists’ aren’t really against Theism,but against Christianity,it’s why Dillahunty’s commenters were mocking virgin birth,YEC,etc. in the live chat instead of engaging with your arguments. Their self assurance comes from their confidence in Christianity’s falsehood,particularly fundamentalist / Calvinist (Baptist) Christianity (which I’m nearly certain is true)
I think it would do you a lot better to elaborate on your view of god instead of your argument for god in debates with non-intellectuals & their fans,who would then be more open to hearing you out
They are also against Islam, in fact I think they really became prominent after 9/11.
Ah yes, another patient who insists the X-rays are wrong because they don’t feel any pain. Let’s scrub through this plaque-ridden rebuttal, shall we?
🦷 “‘New Atheists’ aren’t really against Theism, but against Christianity.”
That’s not bizarre — it’s just diagnostic. If a dentist notices most of the cavities are in molars, it doesn’t mean incisors are immune — just that decay tends to show up where people chew the hardest. Likewise, New Atheists may claim to reject all theism, but their dental tools are almost always aimed squarely at Christianity. The most vigorous scraping happens where the buildup is culturally thickest.
🪥 “Their self-assurance comes from their confidence in Christianity’s falsehood.”
Which is like saying someone brushing their teeth aggressively must be right about fluoride conspiracy theories. Confidence isn’t the issue — basis is. The point is that their self-assurance is often directed at one tradition, despite claiming to represent a universal solvent. It’s like offering mouthwash for a full-body illness.
🦷 “Should they feel embarrassed of their beliefs when they conclude Christianity is bogus?”
No, but they should stop acting like their diagnosis is peer-reviewed when all they’ve done is poke a tongue depressor at theology. You don’t get to play periodontist just because you own a toothbrush. If they’re confident, fine — but let’s not confuse snarky certainty with sound philosophy.
🪥 “It would do you better to elaborate on your view of God rather than argue for God in debates with non-intellectuals.”
This one’s got the bite force of someone chewing aluminum foil. The suggestion is that unless your opponent has academic credentials, you should ditch argumentation and just share feelings. That’s not how any kind of inquiry — medical, philosophical, or dental — works. When explaining periodontal disease, I don’t say “well, I just feel like gums are important.” I show evidence. Likewise, debates aren’t therapy sessions; they’re exams.
And yes — calling someone a non-intellectual in that context is like accusing your hygienist of being a plumber because they didn’t quote Nietzsche during your cleaning. It's condescending, shallow, and completely misaligned with the spirit of rational inquiry.
Let’s clean this up and get you fitted with a new retainer — because this whole argument needs structural support.
The hyper fixation on the chatgpt stuff was really funny
I read the ChatGPT comment as saying: "look, this is such a basic principle of probability that even ChatGPT can reliably explain it to you".
Perhaps because certain people tend to take everything ChatGPT says at face value and readily accept it without doing any thinking of their own, and there's a reaction to these kinds of people by others who are so disgusted by the notion of just treating ChatGPT as a prophet that they fail to consider that it can be still be used to generate arguments that should be seriously examined (we can verify whether the math proof generated by ChatGPT about the probabilities was correct or not, rather than just accepting the result given by it).
I think the only way to react this way from the way it was mentioned in the debate is if Dillahunty wasn't really paying serious attention to anything said and was just hunting for stuff to respond to (which is a common and effective tactic for internet content creators despite its lack of argumentative validity). In context it's immensely obvious BB is saying to ask ChatGPT in the same way as saying "it's trivial, just Google it and verify that everyone agrees that this is how the math shakes out instead of wasting everyone's time making an argument against Bayes' theorem that you clearly haven't thought about prior to this conversation".
I’m really interested to see how chatGPT affects online arguments. On its face, it’s a godsend: ChatGPT is always willing to read an argument, no matter how stupid, tedious, or petty the argument is, and it can often tell you which side of the argument is stronger. Manipulating it seems difficult, so long as everybody can see the prompt. But obviously ChatGPT is not infallible - and maybe that’s enough to keep it from being seen as a good arbitrator by most.
I found the fact that his audience was so receptive to it to be quite depressing, given how transparently it was just chum for social media content and entirely immaterial to anything being discussed
Do you think there are objective facts about what the correct prior for a particular position is? You speak of God having a high prior probability but how is this established? The typical way I think of priors is as descriptive facts about a person's level of confidence in the truth of a proposition. A prior could be anything from 0 to 1. I don't generally think of there being a fact of the matter about what one's prior ought to be nor do I think that there are facts about what the prior of a given proposition are independent of any particular agent's stance on it.
The reference to "simplicity" makes me think BB has something Solomonoff-like in mind.
Yeah, Solomonoff reasoning gets cargo-culted a lot.
The irony, of course it's that on Solomonoff prior anything limitless/infinite has infinitesimal probability
I don't think that's literally true, since you can encode a for loop in presumably a small number of bits, but I suspect it is true in this case
You could argue that priors can also be informed by objective considerations, such as simplicity, explanatory power, and coherence with known facts. These considerations allow us to reason about what priors "ought" to be, even if there is no single objectively correct prior.
Not sure I got your objection. How does probability being subjective or objective impact BB's argument?
He is most likely relating it to fine-tuning argument.
It’s possible that atheists don’t engage with theist arguments because, often, the arguments seem vague and loose, likely incorrect, but hard to argue because of insufficient specificity. For example, you say here: “it’s likely that God exists because God is a mind of the very simplest sort—one totally without limits”. An atheist would say “there is nothing ‘simple’ about an all powerful mind. We have no concept of a plan how such a mind would be constituted nor how it could exist nor even what ‘all powerful’ entails in terms of specific powers.” Consequently, when you start your argument that way, you lose the people that don’t already presuppose the existence of god.
I think it's pretty clear that 'simple' means 'partless' here.
Empty map doesn't imply empty territory. Being ignorant about parts doesn't mean that something is partless. And trying to come up with a notion of something that couldn't possibly be constructed from any parts, doesn't necessary give you something simple, more likely than not you are getting something impossible and infinitely complex.
But we’re not talking about a being that might have parts. We’re talking about a being that necessarily doesn’t have parts.
In this context, ‘simple’ is a technical term that comes from mereology. ‘Simple’ just means ‘no parts.’ Now, you might say that doesn’t track ordinary usage, but it’s not trying to.
> We’re talking about a being that necessarily doesn’t have parts.
I've already answered it here: "And trying to come up with a notion of something that couldn't possibly be constructed from any parts, doesn't necessary give you something simple, more likely than not you are getting something impossible and infinitely complex."
But let's go a bit more into details. What we are dealing with is God being simple "by definition". Like we define God as:
1. Omnipotent, omnibenevalent, omniscient and omnipresent mind
2. Necessary doesn't have any parts
Which may seem as a clever trick untill you realise that you can't change reality by picking a different definition. Anything fitting this definition would be simple in the sense of not having any parts - that's true. But now, as mathematicians do, you need to prove existence - that there is at least one entity that fits, that you definition is not itself self-contradictory. Otherwise, you have logically pinpointed something impossible, and therefore infinitely complex.
Well, I don’t see how ‘impossible’ entails ‘infinitely complex.’ I think it is impossible for there to be a smallest physical part, and yet a physical simple wouldn’t be infinitely complex—it would be simple!
Hi Robert - thanks, your "partless" comment is an excellent illustration of my point: this argument is "vague and loose, likely incorrect, but hard to argue because of insufficient specificity". What specifically is a "partless mind"? How specifically does a disembodied "mind" effect "power" in the universe? How specifically does partlessness yet all-powerfulness imply simplicity and a high prior probability? If your answer to any of these questions is that "god is beyond human understanding", then what is the justification to claim that something this sophisticated is "partless" and "simple" and "high prior"?
I think this is where you go and read the argument: https://benthams.substack.com/p/theism-has-a-super-high-prior-probability.
I’m not defending the argument, since it’s not my argument! The complaint was that ‘simple’ isn’t a well-defined notion. But giving a formal definition of simplicity is very easy in classical extensional mereology, and so I don’t think the complaint that ‘simple’ is ‘likely incoherent’ is a very strong argument.
Hi Robert - thanks for the link. I've read it, and read many of BB's other articles. I enjoy them all. However, as a skeptic, I find baseless (to me) assertions, large and small, throughout. Sometimes I comment on those issues.
The goal of BB's arguments, I assume, is some combination of a) convincing non-believers of god's existence; b) reinforcing the beliefs of already-believers; and c) monetizing the topic. I don't doubt or begrudge the success of (b) or (c).
My original and continued point here is that BB is failing on (a). If you do not already believe in god, you won't be much persuaded by "it is self-evident that god is a limitless all-good partless mind and is so simple that it must have the highest of all priors". I have not yet read in BB's many fine articles, any rationale for why I should take this as true.
To be clear: I'm not arguing AGAINST the existence of god. I'm suggesting that this argument FOR god's existence is not at all effective with non-believers, and for good reason.
Lastly, of course one can define "partless" via mereology. That's not the issue. It's the "god is partless limitless all-powerful simplicity" that needs some justification. It's just not self-evident nor convincing to anyone who is not already a believer.
What is a "part" in the context of a mind? How does that play with "limitless"? In the cases where I can think of well-defined simpleness-as-partlessness like finite dimensional algebras, the resulting simple objects don't strike me as in any way "limitless"--the matrix algebras M_n aren't in any way "less limited" than direct sums of them.
If anything, in the case of minds, the opposite strikes me as more likely: a limitless mind should be able to think about multiple things at once; to model multiple other minds; or to do all sorts of other potentially "parallel" processes and it seems pretty natural to imagine that the different thinking pieces of a limitless mind decompose it into parts: the part that's calculating the fall of a single sparrow; the part that's experiencing suffering as an embodied person on the cross; etc.
I agree with you on the merits of your arguments. I think it'd be better if you didn't include so many petty insults though. Standards of discourse are also important.
Yes, especially when he complains about other people ”dunking”.
I was a regular viewer of The Atheist Experience back when I was a young atheist in the early 2010s. Mostly it was just a source of entertainment, seeing Matt effortlessly destroy all the clueless Christians who would call in. But it also reassured me that my worldview was correct and its opponents were dumb. He’s much like a televangelist that way.
It’s no surprise then that when an actually intelligent theist shows up the reaction from Matt and the audience is not so much curiosity as annoyance that their usual song and dance routine has been so rudely interrupted by arguments that have to be seriously contended with.
Great review! You did well in the debate, and I enjoyed reading this follow-up as well. That rant about the new atheist disdain for rationality was seriously based -- I agree with you entirely.
Man, I wish I had theist friends who were philosophically and mathematically inclined like this. It’s difficult to find people you can openly discuss arguments like this that doesn’t fall into them getting stuck on a random objection that’s about them not understanding the argument. There’s definitely force to the arguments from consciousness and anthropic arguments, imo, though I’m still an atheist.
Plus, with friends you can spend time having good faith discussions about what the premises and arguments say and don’t say before starting the argument — a debate like this only really works when both sides understand the other’s arguments well, so it’ll only work between people who are either very smart, well read, and done a little research on people’s position, otherwise it’ll be a blowout like this was.
If you debate Scott Alexander, I will be there instantly.
Really entertaining debate. It's kind of nuts that he has been debating these things for like 20 years and still hasn't managed to get basic epistemology concepts right.
I think BB has something like the Solomonoff prior in mind here, where your sample space is finite-length binary sequences representing different hypotheses, with the Kolmogorov complexity of each sequence relative to some Turing machine as the prior probability. That is, you imagine each hypothesis as a class of "computer programs" that can run on some universal Turing machine, and the length of the shortest program that implements the hypothesis is related to its prior probability.
So for example, roughly speaking a hypothesis of the form, "this is a sequence of all 1s of length 10" is more probable a prior than "this is a sequence of length 10 of the form 0111001110" because you can write "print '1'*10" for the first one, but you can't do too much better than "print '0111001110'" for the second.
I do, however, think that encoding concepts like "God", "mind", "limitless" (in this context) are... of dubious simplicity if actually converted into this formalism. I also am really not sure why it's at all clear that "mind without limit" is a good definition of God. Or, for that matter, whether the God hypothesis can be computed by a Turing machine.
I think BB is applying Solomonoff-type reasoning in a situation where it's really not clear if it applies, and if so, how to model it.
Combine these objections with the uncomputability of the Solomonoff prior, and I agree that we're not really doing anything usefully described as "probability" here.
Hmm this was meant as a reply to Online Watcher but I seem to have screwed that up, sorry.
Why think he's referring to the Solomonoff prior? After looking it up, it doesn't appear reminiscent of anything I've seen BB say before, and it seems like this would introduce commitments that BB wouldn't want to accept.
Why not have a debate with a thoughtful, reasonable atheist?
I've done that too! Some chance I'll debate Scott Alexander and/or Richard Chappell.
A debate with Scott Alexander would be fantastic. I'm a long-time fan. I didn't know he was an atheist (though it doesn't surprise me).
I would pay money to see a debate between you and Chappell.
I think he's done so before and will doubtless do so again, but in this instance I think he wanted a chance to show the more lucid parts of Matt Dillahunty's large and somewhat dogmatic audience that there are non-stupid arguments for theism, which they may have low exposure to.
I think it's a noble pursuit but unlikely to be effective. The snide comment made by someone (who I can't recall now) that atheism is a Christian heresy is I think about the fact that atheism as a popular movement is primarily about being in conversation with Christianity and specifically a very narrow kind of low church protestantism that the adherents of the Atheist movement grew up around. The question of whether this religion is false is, to them, the only interesting question about this matter at all. They are not sincerely interested in the question of whether God exists as a philosophical or academic matter, but actually as a battle in social politics against the groups they identify with theism.
So many things are wrong with this article in terms of what actually happened, but I'll just point this one out for anyone confused. It was clear to me that Dillahunty went "ballistics" when Matthew asked him to consult ChatGPT because it came off as smug and (obliquely) intellectually derogatory. The two had just disagreed on their probability arguments when Dillahunty started to type and Matthew said something to the effect of, "yeah, go ahead and ask your Chat GPT", which carried the suggestion that Dillahunty wasn't smart enough to understand Matthew's gigabrained contentions.
Unless Matthew is just bad at reading the room, it's obvious that this was what upset Dillahunty. Asked at a different time, it might have come off differently. But if we have just disagreed and I'm typing on my computer and you, without knowing what I'm actually doing, say, go ahead and ask A.I, I'll assume you a.) think I'm not smart enough to understand your arguments, and/or b.) don't think I'm confident enough in what I'm saying to actually defend it upon attack without consulting A.I.
Not that there's anything wrong with consulting A.I for literally anything. But this context is important.
I've made a few critical comments about the Solomonoff-style prior argument for God, but I also want to register (one of) my objection(s) to the fine tuning argument.
Something like the cosmological constant, which is related to the vacuum energy, and presumably may be implicated in any future unified theory, or theory of quantum gravity, or whatever, can't be uncomplicatedly treated as a free parameter.
Imagine making a fine tuning argument from the value of pi: if pi were too much smaller, the volume of stars would be too small and all stars would collapse into black holes. Putting aside all the other problems, it's really doubtful whether you can reason this way about *mathematical* facts; and I think one ought to be *very* suspicious of Bayesian reasoning that treats the value of mathematical constants as either "observations" or "evidence".
Now, I don't *know* that the value of the cosmological constant will turn out to be a mathematical consequence of some other facts about the universe, or if so which ones, and whether there will be a canonical way to think of some of those parameters as "free parameters"; but if that were to happen we might well learn that the value of the cosmological constant is not actually a free parameter at all. It is more like the value of pi, a mathematical fact, or at least, a mathematical fact conditioned on some other physical facts.
Now, this doesn't mean that we can refute fine-tuning arguments by just *asserting* that actually, the value of Lambda is just as mathematically determined as the value of pi, but I think Bayesian reasoning about the difference between these views is complicated by the fact that almost all of the disagreement is in model uncertainty: *if* Lambda is a free parameter, then the value of Lambda is good evidence of fine-tuning, but if it's not, then it's not. So the probability for your fine tuning argument has to factor through the probability of whether we should model Lambda as a free parameter, which is a much more thorny question. It's, essentially, a question about "how should we model the process that generates the underlying laws of physics", which is not something I think we have a good handle on.
I don't know exactly what the takeaway is here; like I say I don't think this refutes fine tuning.
But I think it points to the fact that a lot of the difference between the theistic worldview and the atheistic worldview probably comes down to questions about "how to model how the physical laws were generated", and it's *much* harder to know how to get evidence for that question.
I agree that Matt did horribly (as 100% expected). Regarding the error about Bayesian evidence, I'd have to go back and review the transcript, but I remember it being possible while I watched it live that he made a semantic confusion. He was saying at one point something of the form "Just because a theory T predicts evidence E more than ~T, the evidence doesn't make the theory more probable." But this could be parsed in two ways: either that it doesn't make T more posterior-probable than it was before learning E (which is false), or that it doesn't make T more posterior-probable than ~T overall (true). But he might've said extra stuff in the conversation indicating the first, unfavorable interpretation. I honestly kind of tuned out during a lot of his talking because it was hard to listen to. So I don't know.
>Matt asked what it means to be all knowing. Answer: to be all knowing means to know every truth.
Not that Matt brought up this objection, but it's not actually clear this definition is even logically consistent! Consider the knower paradox[1] applied to God, i.e., the sentence "God does not know this sentence is true." Naively, it's easy to prove that God doesn't know it, so it's true, so there's a true statement God doesn't know.
You might want to say this sentence actually lacks a truth value, perhaps via failing to pick out an actual proposition because of self-referentiality, and that omniscience only refers to knowing true propositions. OK, fine. Call sentences that successfully pick out propositions "proposition-apt," and call those that successfully pick out *true* propositions "true-proposition-apt." But it seems like *the sentence that a particular other sentence is or isn't known by God to be true-proposition-apt* ought to be the sort of thing that's determinately true or false, i.e., it's the sort of sentence that ought itself to be proposition-apt.
Then by a (very slightly figurative) usage of Gödel's diagonal lemma, we can cook up some sentence φ which is provably equivalent to the sentence ψ = "it's not the case that that God knows that ⌜φ⌝ is true-proposition-apt."[2] Since this new sentence ψ is asserting something about God's knowledge of true-proposition-aptness of a specific sentence, it is itself proposition-apt, hence either determinately true or false. Since it's provably equivalent to φ, we should also get that φ is proposition-apt, hence either true or false. *Now* we can run an analogue of the knower paradox to generate a problem.[3]
You can next try to escape this paradox further by abandoning classical logic, but this gets very technical, and the exotic thing you get at the end of this process won't obviously allow you to quantify over all determinate truths such that it will make any sense to talk about knowing "all" of them.
[1] https://en.wikipedia.org/wiki/Knower_paradox
[2] Here, ⌜φ⌝ represents taking the actual content of the sentence we've denoted by "φ" and putting it all in quotes.
[3] If ψ is false, then God knows that φ picks out a true proposition, so the sentence φ is true. But then the equivalent sentence ψ is also true, a contradiction to our assumption. So ψ is in fact true. But then the equivalent sentence φ is also true, and so ⌜φ⌝ is in fact true-proposition-apt. So ψ is truthfully asserting God's failure to know a true fact. (And we can do these moves because we're guaranteed up-front that both φ and ψ are proposition-apt, so we can ignore indeterminacy.)
Haven't taken the mental effort to digest this particular paradox of omniscience, but I think in pretty much every case you can get precise analogues by replacing "it's true that" with "God knows that."
Yes, and due to the semantic paradoxes, it's in fact it's highly controversial that you can both coherently (= there is a unified global truth predicate which can be applied to all sentences, as opposed to an infinite nested hierarchy of distinct truth predicates) and consistently (= there are no contradictions with doing so) talk about "all truths," as I tried to get at at the end there.
Although I don't expect you to have time to read or digest this paper, here[1] is an overview of recent technical work on the semantic paradoxes that goes over several approaches and IMO implies that all seemingly fall prey to one of those two pitfalls.
[1] https://www.academia.edu/84053122/Revenge
To be clear, I didn't mean to suggest that there's no chance any paradox of omniscience works. I just meant that brazenly asking what omniscience means isn't a convincing objection to theism.
I agree, and of course Matt Dillahunty didn't have any interesting objections to omniscience to raise, so I didn't mean to imply you said anything egregious during the debate. Just voicing some personal side thoughts.
Thanks!
Yes, this is exactly right: BB implicitly models various physical parameters as free parameters that can take on any value in a range that is physically meaningful given our current understanding of physics. But whether or not this is a good way to think about how the universe came about is more or less the issue under discussion! I don't think you necessarily need to see other instances to calibrate, but you need to have some idea of what *what the space of possible values looks like*, and I think the best answer is, we don't know.
I'd like to especially highlight that the bounded asymmetry argument another commenter gives elsewhere as exactly analogous as an argument against God: let a, b, c be some measures of God's potence, science, and benevolence--we have now parameterized God, and if we assume that those values can have taken any value in their range, then we can immediately start constructing fine-tuning arguments against God: why does God have maximal benevolence, when if we pick a real number from an interval uniformly at random, the probability of it achieving its maximum is literally 0??
But the obvious answer from theists is: this is a dumb way to model God, like potential beings all come from a factory with preset knobs for how powerful, knowledgeable, and good they are, and then the dials are twisted at random. The theists' argument for God conceptualizes the probability space in a completely different way. But atheists have the same objection against modeling universes in this way.
And that's fine, but it means most of the argument should be about *what is the probability that each side's way of modeling this is better*? Anything that doesn't address that is not going to be persuasive to the other side; it's essentially saying "conditional on my caricature of your worldview being correct, your worldview is dumb". All the work is being done in the initial conditionalizing.
Ugh once again my comment has detached from where I meant it to go; this should be a reply to Citrit.