One of the most popular and persuasive arguments for theism is the so-called Fine-Tuning Argument. But there is a singular fallacy in it that I think usually gets overlooked. There are of course actually many fallacies in it, but those have been well-explored (hence I’ve written on this subject often enough and there is a recent video of actual cosmologists explaining it). What I will get to today is a crucial error right at the very heart of the argument, which though I have touched on it in other articles, I have not made as explicit as I will here today.

Of course, if you weren’t already aware, the Fine-Tuning Argument goes something like this:

  1. We observe that the fundamental constants of the universe require an absurdly perfect balance to allow the possibility of living organisms naturally arising (those most commonly appealed to are the strength of the gravitational force and the strength of the electromagnetic force).
  2. This perfect balance either arose by random accident, some inevitable natural law, or intelligent design.
  3. If selected at random, such a perfect set of measures is extraordinarily unlikely; and even if some inevitable natural law would make no other outcome likely, that would still be extraordinarily unlikely; whereas such a perfect set of measures would be highly likely if selected intelligently.
  4. Therefore this perfect balance is extraordinarily more likely to have been selected intelligently.

Yes, that doesn’t get you all the way to Jesus, or Christianity, or anything the like. But that’s not the point of the argument. Most religions can be dismissed as depending on the extraordinarily improbable thesis of invisible superghosts, because most arguments for any specific religion depend on the presumption of there being a god of such means in the first place. Something like the fine-tuning argument can therefore set the stage by turning that presumption into an established premise, from which further arguments can then proceed. Though that requires it be a sound argument. So it matters whether its premises are even true, or whether the conclusion actually follows from them.

First: All Those Other Fallacies

Even Christopher Hitchens acknowledged that the Fine-Tuning Argument was the best argument theists have, although he still found it unconvincing. His reason for finding it their best argument is that it at least pointed to a real thing that needed to be explained, and did have a superficial appeal as long as you didn’t look at it too closely; whereas most of their arguments appeal to false facts or overt fallacies that can be exposed as fraudulent in a minute’s time. His reasons for still not finding it persuasive were twofold: (1) it has never passed peer review in cosmological science, which is a bloody big red flag (for if real cosmologists aren’t persuaded, neither should amateurs be); and (2) there is far more chaos, destruction, failure, and uninhabitability in the universe to actually conclude it is finely tuned. Hitchens is right on both points, but especially the latter.

I have put that point into simple Bayesian terms, following multiple corroborated peer reviewed formalizations doing the same. In short, “if chance produced this universe, we should expect it to be only barely conducive to life, indeed almost entirely lethal to it,” and that is exactly what we observe—just as Hitchens notes, both in time and space: by an absolutely extraordinary proportion of both, the universe is hostile, not amenable, to life. In fact, it is pretty much as hostile to life as any universe could be and still be able to produce any minuscule quantities of life clinging to the barest fleeting motes of survival. Which is the most probable universe that accidental life will ever observe: because a universe so hostile it never produces life will never be observed, while by far most of the random universes that could produce life will resemble ours in precisely this respect. 

Hence, what we actually observe (as Hitchens relates) is predicted on the “fine-tuning was caused by accident” hypothesis; and yet it is counter-predicted on the “fine-tuning was caused by a life-seeking engineer” hypothesis. Fine-tuning, plus the evidence of the universe as actually observed, is therefore evidence against theism, not the other way around. But this requires putting back in all the evidence the theist leaves out. They only mention fine-tuning and the tiny motes of habitability for life; they conveniently leave out the vast proportions of space and time that are no such thing. Most apologetics works that way.

There are very, very much more hospitable universes to choose from among all logically possible universes available to an omnipotent builder. In fact we will ourselves soon be building them: within a timescale of mere centuries (a trivial delay in a universe billions of years old) it’s likely we will all be living in utopian simverses that are entirely hospitable to life—and a far happier life too, provided we can overcome our own fallibilities, but gods are supposed to be infallible, so that can’t have gotten in their way. Popular theisms already entail the same conclusion. Their gods, after all, knew how to build an eternal Heaven that everyone insists will be a much better place—indeed these days that’s (almost) the entire emotional point of believing in gods in the first place. So theists admit there are vastly better designed, vastly more hospitable, vastly more suitable worlds to build. So they are facing a serious problem explaining why God didn’t make ours one of those (I made this point already in Sense and Goodness without God decades ago, in IV.4, pp. 275-77). So they leave all that out, stripping away all the actual pertinent evidence, and leave only the “Earth is hospitable to life” datum, and ask us to be “amazed” that the fundamental forces were just precisely arranged so as to produce an occasionally rare Earth in a vast deadly void of lethal garbage. 

Whereas, as I’ve put it before, we can prove it the case that “all godless universes ever observed will be finely tuned,” for it is not possible then to be alive in any other, whereas “most universes a God would be likely to make wouldn’t even have physical constants” much less any “need to tune them.” It’s like Captain Kirk said about god requesting a starship: what need does a god have of that? The request itself disproves he’s god. In like fashion, fine-tuning is what we expect to see if there is not a god; and it is not what we expect to see if there is one. The argument’s own premise thus disproves its own conclusion.

That alone dispatches the argument. Since Premise 3 in the argument makes assertions about the likelihood—how likely it is we would observe the evidence we do observe, on each of the three competing hypotheses—leaving out evidence that affects those relative probabilities is fatal to the argument’s required logic. When we include all the data, we get a reversal of probabilities: intelligent engineering flips from the most likely to in fact the least likely explanation of all the data taken together. However unlikely a random selection would be, or however remarkable a natural inevitability would be, the evidence simply supports the conclusion that one or the other most likely happened, and not the other way around.

And that’s that.

But even if you are dissatisfied with “we don’t know what the explanation is, but all evidence points to it not being God,” you can still point out that we do have perfectly viable explanations for fine-tuning, which have far more scientific backing than “God did it.” Insofar as the evidence already points to a natural rather than an intelligent cause of fine-tuning (as I just noted it does), then if the only two logically possible explanations were “our universe is part of a quasi-infinite multiverse” and “a god exists,” then fine-tuning is evidence for the multiverse hypothesis

One cannot object to this by saying a multiverse violates Ockham’s Razor “because it poses infinitely many epicycles to explain one,” not only because we can say the same thing about your god solution as well (a point I shall be getting to), but more importantly because no peer-reviewed multiverse theory in the science of cosmology today just “poses infinitely many epicycles to explain one.” To the contrary, all current multiverse theories demonstrate the inevitable appearance of endless universes from a very simple set of physical starting conditions; a much simpler “brute fact” proposal than any god theory, involving fewer assertions, and all of them (unlike anything about god) backed by scientific knowledge and precedent. Even William Lane Craig’s beloved Borde-Guth-Vilenkin theorem (and every single one of those cosmologists individually) asserts a multiverse is an inevitable causal (not just epistemic) consequence of observed facts (see Alex Vilenkin in “Before the Big Bang 9: A Multiverse from Nothing?” esp. starting at 8:21 and then 15:04; Alan Guth in “Before the Big Bang 4: Eternal Inflation & The Multiverse,” esp. starting at 34:28; and Leonard Susskind, “Was There a Beginning?” in the MIT Technology Review). Thus in any competition between explanations, “there is a multiverse” enormously prevails, by every epistemic measure and standard, over “an invisible superghost did it” (see Six Arguments That a Multiverse Is More Probable Than a God). Hence fine-tuning is evidence for multiverse theory, not for any god theory.

That’s all one need say. You can just drop mic and walk away at that point. But of course one could still challenge other components of the argument. I’ll digress only briefly on those because I think most of them are less effective points to make, although one of them is really important. And then I’ll close by hitting head-on the usual rebuttal you’ll hear to what we just demonstrated, and why it actually contains yet more seeds of the argument’s own destruction.

Those Other Minor Challenges

It has already been pointed out (by the likes of Victor Stenger and others) that naive formulations of the Fine-Tuning Argument use invalid mathematics. For example, they will hold one constant as fixed (like, gravity) and ask how many values for another constant (like, electromagnetism) would produce stars, and then planets, and thus life. But if we are selecting values at random, both values can vary. And obviously, for any doubling of the alpha constant, for example, there can be a doubling of the gravitational constant, producing exactly the same balance between those constants all over again, and thereby observationally the same outcome: stars, planets, life. It is thus easy to show that there are infinitely many pairings of G and alpha that will produce life—and infinitely many that won’t (and even if you need groups of three or more constants to be aligned, the same follows). So we know there are infinitely many possible universes that can bear life, and infinitely many that can’t. How do you calculate which is more likely, then, when both sides of the frequency equal “infinity”? 

One might try to solve this problem by finitely sampling the probability space, but that isn’t what naive Fine-Tuning Arguments usually do (or even know how to do). And it’s not clear that that would even work, as has been pointed out even by Christian apologists: in McGrew, McGrew and Vestrup, “Probabilities and the Fine-Tuning Argument: A Sceptical View,” Mind 110 (2001), it’s explained that our assumptions about what you’d get by “randomly” selecting a universe from a finite range of possibilities don’t actually hold for infinite ranges. So we actually can’t even do the math. A finite sampling of the space doesn’t in fact solve this problem, because we don’t get to do that when something actually gets selected. If you roll an infinitely-sided die, it actually does not follow that a side that appears ten times as often as another will come up ten times as often when rolled, precisely because the rules of finite arithmetic that would entail that conclusion don’t apply to infinite quantities.

However, I am assuming here that this has “been solved” somehow, and thus I am steelmanning the argument by assuming that even after we do the correct math (whatever that is), we still get Premise 1. But even then one could still challenge Premise 1 on the grounds that it is not as well established as is typically claimed. We actually have never explored (even by randomly sampling) the entire “possibility space” of comparably-outcomed collections of physical constants so as to determine they are rare. We do not have even the means to know every possible physical force that could manifest to change the equation. Published studies only assume the forces we have in this universe, not all logically possible combinations of forces, and therefore no published study actually addresses the question we need to answer in order to establish anything like Premise 1. Nor do we know every possible value each possible force could take, which is an even more important problem: we don’t even know how these constants take any value, so we cannot claim to know the range available is infinite, or that every value is equally likely. Nor do we know which combinations lead inevitably to a universe’s collapse or other event resetting the values and thus recalibrating the probabilities. 

This last point is also often overlooked. For example, if 90% of “force-constant” combinations result inevitably in universe-collapse followed by a re-randomization of the values, we can only ascertain the probability of a life-bearing universe from the remaining 10% of combinations. Likewise if 90% of that 10% of those combinations result in runaway inflation that produces the exact opposite, a Big Rip outcome, and that once again re-randomizes the values (which cosmologists admit is possible), then we can only use 1% of the original combinations possible—because all other combinations will inevitably produce one of those. Think of a six-sided die-roll whereby you have to re-roll the die every time it rolls lower than 2 or higher than 5 (hence “any extreme value results in a re-roll”): a result of a 1 or a 6 will then never occur. Because they won’t last; inevitably they will be replaced with another roll. So the final result will always be between 2 and 5, only four possibilities rather than six; and the probability of ending up with any one of those results will always be 1 in 4, not 1 in 6. And this remains true no matter how many sides the die has. A hundred-sided die that does this, will still only ever end up between 2 and 5, and the odds of rolling a 2 on such a die will still be 1 in 4, not 1 in 100. And this will be true even of an infinitely sided die: it will always only ever end up rolling 2, 3, 4, or 5, and the odds of it rolling any one of those will again be just 1 in 4, not infinity to one against.

And because we don’t know this is or isn’t what will happen with any random combinations of any randomly possible forces of any random measure, nor do we know whether some measures are more likely outcomes than others or whether they are all equally likely, nor do we know whether there is a limit to how many forces can randomly arise in combination or no such limit at all, we simply cannot calculate the probability of any one outcome. At all. Much less such as would support Premise 1. 

Nevertheless, I think steelmanning requires granting Premise 1 even in the face of this objection, since disproving Premise 1 is also impossible on present knowledge, so attacking the argument this way is harder to articulate the merit of. You would essentially have to argue, “we actually don’t know whether this is true or false; and when we don’t know something, prior odds favor the explanation is natural, not supernatural; therefore, prior odds favor the explanation of apparent fine-tuning is natural, not supernatural.” But though that is entirely true, it’s a psychologically weak rebuttal. It grants too much to the theist, making it seem they are just an inch away from proving their point, and throws you into the weeds of explaining why this isn’t just a presumption of naturalism. And though you can indeed show it’s not, by that point this all looks like you are trying to avoid an inexplicable fact rather than explaining it. 

So I don’t recommend relying on that approach (and I myself have done so only occasionally, and always in a context emphasizing the other, stronger rebuttals). Likewise, while Premise 2 as I worded it is unassailable (those three options do exhaust all logical possibilities), some presenters misformulate it, inviting easy attack. My intention here was again to steelman the argument, since a theist could just as well do so, too, and thus evade or dispatch any such rebuttal. Likewise, as I wrote it, one could also say the conclusion does not formally follow from Premise 3, as a crucial step has been omitted: a reckoning of the relative prior probabilities of the three hypotheses. To arrive at line 4 from line 3 requires assuming these priors are equal or already favor the conclusion, and that is an assumption not established in any premise I presented. One might say I’m in that case strawmanning the argument by leaving that premise out. But I left it out on purpose: because it is indeed typically overlooked by advocates of this argument; and when not overlooked, is not validly accounted. This in fact brings us to my concluding point of today.

That Most Hidden Fallacy

Even if we assume there is no multiverse that we are a small part of (even though all evidence and logic to date supports that conclusion), but instead assume that the cosmic dice were rolled only once, and just by chance they rolled one of those rare universes that would produce life rather than one of the many more possible universes that wouldn’t, it still follows that “there is a God” is no more likely a remarkable bit of luck than that random universe would be. And this is the hidden fallacy in every Fine-Tuning Argument. The theist will want to insist the prior odds on any “we just got lucky” starting point obviously favor gods over random fine-tuning, but that actually isn’t the case. A lucky God is just as improbable on no prior assumptions as a lucky universe.

In other words, what someone might resort to in defending the Fine-Tuning Argument from the points so far made is to say that the “random one-off universe” requires a remarkably lucky accident, whereas if the God they are talking about already existed, then that same fine-tuning would be 100% certain to exist, and thus not unlikely at all. I’ve already shown you what’s wrong with that second premise. Fine-tuning isn’t even a likely way a real god would make a universe, much less “100% certain” to be the way, and all the other evidence is in accord with it being in our observed case a chance accident rather than an expected outcome of an intelligent plan. But there is another hidden premise here, underlying even that: that we can just assume there would be exactly such a god.

In fact, that entails an even more incredible amount of luck. So you can say we got lucky on a random roll of a universe; or just as lucky (or even more so, honestly) on a random roll of a god. Even at best it’s a wash; and at worst, God is much less likely. Both are positing that we are incredibly lucky. And arguably, the theist is positing an even more incredible bit of luck than even the “one universe” atheist is. Because there are infinitely many combinations of force-constants that would give us some universe that would produce some kind of people somewhere. But there are very few specific gods who would explain all known facts, and those gods all have the highest specified complexity imaginable (see The Argument from Specified Complexity against Supernaturalism).

These aren’t just some randomly selected deities from among all logically possible deities (much less all logically possible things), the vast majority of whom will have no interest in creating people or even life (an interest that is hyper-specific within the infinite set of all logically possible interests). And even of those who do, the vast majority of them will be limited, fallible, and morally ambiguous (because there are more logically possible ways to be like that than to be perfect). And by far most randomly selectable gods will be far more empirically obvious—since you also need extraordinarily convoluted reasons for a god not to be. To the contrary, the God the theist needs in order to be a viable explanation for fine-tuning and every other observation is a very specific deity with infinitely convenient knowledge, abilities, and virtues, with an extraordinarily convenient nature even beyond that—they just exist for no reason, yet are still perfectly ordered and capable; and they require no body to remain so perfectly ordered and capable—as well as possessed of a very specific and convoluted set of motivations so bizarre theists will insist they are incomprehensible (see Is a Good God Logically Impossible?).

You might say, “But no one is saying God arose from a roll of cosmic dice.” But that’s not relevant to the point. It’s like a singular past-eternal universe: it could just luckily have been fine-tuned without there ever being an event of rolling cosmic dice either. A theist would say that’s just as lucky. Hence we aren’t talking about some event where some random selection was made, or what we might call “contingent luck.” There could have been such an event; it’s just not relevant to what I am saying here whether there was. We are instead talking about the “existential luck” of there just being such a thing, out of all the logically possible things there could have been instead (including nothing ever being at all, although that works out to be the least likely thing there would ever continue to be: see The Problem with Nothing). There is no way to say that “we just got lucky with this god thing” is any more likely than “we just got lucky with this universe thing.” Since even at best both probabilities are inscrutably small, they are therefore epistemically commensurate. Which means, even in the “best case” for the theist, as far as we honestly know, one is no more likely to have happened than the other.

So even if we assume fine-tuning is just a product of pure brute luck, both proposals as to how it came about entail commensurately improbable luck: lucky universe; or lucky god. Theists are thus trying to explain the fine-tuning of the universe by insisting upon an unexplained fine-tuning of the requisite God. Six of one, half a dozen of the other. So we cannot argue for one against the other by saying it was more likely. We simply have no knowledge of that being the case. By contrast, the kind of luck we would need to get one of the current multiverse theories now deemed plausible in cosmological science is demonstrably far less, as they require vastly simpler sets of brute facts than either a “single randomly fine-tuned universe” or a “randomly selected perfectly convenient god.” All their premises are rendered probable by abundant scientific evidence (unlike any attribute of the required god), and are far simpler in theoretical components, and hence in what chance accidents would be required to manifest them over alternatives (again exactly unlike the required god).

The only way to make us not wildly existentially lucky to even have such a god is if the required god is logically necessary. Of course all attempts to prove that have failed. But the important point here is that this means the Fine-Tuning Argument is really just a fig leaf for trying to pass off an Ontological Argument without ever actually defending it. If the Ontological Argument fails, so does every Fine-Tuning Argument. Because in terms of “brute luck” there is no way to distinguish the extraordinary luck of having such a god, and the extraordinary luck of having such a universe—and that’s even on a single-universe, random-constants model. So it is not actually possible to say that Fine-Tuning is more likely the product of a god as of chance accident. And remember, when we look at the actual evidence, it all favors the chance accident explanation. Which leaves us with no other conclusion to logically make.

This follows even for other “fig-leafed” Ontological Arguments like the Arguments from First Mover or Principle of Sufficient Reason. Those are just more sideways attempts to “prove” God is a logically necessary being; and they also fail (see The Argument to the Ontological Whatsit and Feser’s Five Proofs of the Existence of God; and compare Why A Neo-Aristotelian Naturalism Is Probably True). No one has ever presented a valid and sound argument to the conclusion that, really, anything must logically necessarily have always eternally existed, much less something as bizarrely hyper-specific as the requisitely convenient God; if anything has eternally existed, so far as we know, it could have been something else, and thus existential (and not contingent) luck is just “a thing we have to accept” at some level of analysis. And if that’s the case, then the only question is, which possible thing then requires the least amount of luck to be presumed? And the answer so far looks like any of the currently contending multiverse cosmologies. They require far fewer assumptions as to what “just” has to have been true by mere brute fact, than any other competing alternative, especially theism. 

Even if you are uncomfortable with existential luck and feel sure in your heart that there “must” be something at the base of everything whose existence is a logically necessary fact—even though you can’t really prove that to be so, so this is really an emotional and not a logical position to insist upon—we have ample evidence that the most credible contenders are mindless natural facts, like spacetime itself, or the initial quantum chaoses or rudimentary tunneling loops of various inflationary cosmologies and the like. As those are actually necessary (the universe we observe right now literally could never exist without them), and are vastly simpler posits (with far fewer elements, and all of them already in evidence as things that exist) than any bizarre and convoluted god concept.

Conclusion

The irony is if there were a successful Ontological Argument, no one would need the Fine-Tuning Argument. If you already could prove the needed God’s existence logically necessary (by any of the routes mentioned earlier), you wouldn’t need to appeal further to any evidence he existed at all, much less that of Fine-Tuning. Thus the only reason we have the Fine-Tuning Argument is that the Ontological Argument fails, and theists want to try and sneak it in anyway, like a magician’s trick. And they usually don’t even know they’re doing this. Like someone who thinks disembodied minds can exist because they “can imagine one” have fooled themselves—because they can only ever thus simulate one, so as to “imagine” it, with an embodied mind, thus refuting their entire conclusion as to what’s possible—so have Fine-Tuning proponents fooled themselves into thinking they can get the conclusion of the Fine-Tuning Argument on its own—when in fact they need the Ontological Argument to go through first, otherwise their conclusion collapses back into improbability again. Thus they think they have made a profound argument; and don’t realize it’s just a convoluted trick to avoid having to prove the Ontological Argument, which is the only actual way to make the Fine-Tuning Argument work.

To be honest, mindless spacetime looks like the only thing that has a good chance of being a logically necessary being (see The Argument to the Ontological Whatsit). Which is entirely consistent with the observation that Fine-Tuning is more likely a product of nature than of a god. And this all follows even if a single-universe, random-constants model is a true account of all we see, and already science has proved otherwise. Multiverse models are far more empirically supported (see Naturalism Is Not an Axiom of the Sciences but a Conclusion of Them and Six Arguments That a Multiverse Is More Probable Than a God), and require far less luck to get started from nothing. In fact, quite possibly, almost no luck at all. And like “just the right amazingly perfect and convenient god,” every multiverse theory today entails observed Fine-Tuning will be effectively 100% expected, only without anything having to be “just the right amazingly perfect and convenient.” So if both theories predict the same observation to the same probability, it falls to their relative prior probabilities to decide which is the more likely. And among all suitable things that are supposed to “just exist” for no reason, contemporary multiverse theories require vastly less existential luck than God. 

Contemporary multiverse theories therefore not only outweigh God in respect to all scientific evidence, but even more so in respect to prior probability. The God theory is thus doomed. And it was ironically the observation of fine-tuning itself that killed it.

Discover more from Richard Carrier Blogs

Subscribe now to keep reading and get access to the full archive.

Continue reading