Expanded Reply to Joe Folley on the FTA

Longer written version of my response to Joe Folley's objections to the fine-tuning argument.

Joe Folley of the Unsolicited Advice YouTube channel published a video entitled, ‘Why the BEST Argument for God Fails,’ a critique of the fine-tuning argument (FTA). Now, my area of study for the last few years has been the FTA, and I think that Joe's objections do not succeed. Here is why. We shall go through them one by one. I even made a video responding to his criticisms. What follows is a written version of that response, with expansions and clarifications where I have deemed fit.

But first, a brief overview of the FTA: certain features of the laws of physics, like the strength of gravity or the rate of expansion of the very early universe, are enormously sensitive to small changes such that if you shifted the values of these features even a teeny, tiny bit, the universe would be incapable of supporting complex life - life could not evolve anywhere. If, for instance, you made the cosmological constant a tiny bit stronger, the universe would have collapsed back in on itself long before life could have evolved. Just take the fact of fine-tuning for granted. The question then becomes: why do these features of physics take on values that allow for life? It seems like a massive coincidence that, of all the possible magnitudes they could have taken, they all fall within the narrow range ‘life-permitting’ range. The FTA then claims that the best explanation for this fine-tuning design: a cosmic agent of some kind set the laws of physics in place so that life could evolve.

Okay, so that's the context; onward to objection #1.

#1 Of Puddles and People

There are really two objections that Joe runs together here, and I will reply to both.

Life is not special

Joe first makes the point that the combinations of values that allow for life are no more special or significant than any other. I take this to be the implication of the ‘puddle analogy,’ a thought-experiment meant to show that we do not need to offer an explanation for cosmic fine-tuning. We are to imagine a sentient puddle observing the hole within which it finds itself and remarking, “Isn’t it incredible that I exist within a hole that perfectly matches my shape and boundaries?” The puddle would misfire if it sought out some special explanation for this correspondence between hole and shape, because no particular puddle shape is more significant than any other, and some puddle or other had to obtain. But suppose the puddle found itself in the exact shape of a hexagon within its hexagon-shaped hole. In that case, an inference to design seems far more plausible. Similarly, if life is objectively valuable, then the combination of constant values that allow for life lends itself to a design inference in a way that no other combination of values does. More on this below.

A self-selection effect

Second sub-objection: if the universe were not fine-tuned for life, we wouldn't be here to have this conversation about fine-tuning. Here Joe appeals to philosopher Elliot Sober who maintains that because we could not observe a non-life permitting universe, our fine-tuned universe cannot adjudicate between chance and design hypotheses. Sober recounts Eddington's fish analogy as an intuition pump: you take your boat onto a lake and fish all day, catching ten large fish. You consider two explanations: (1) there are mostly large fish in the lake and (2) there are mostly small fish in the lake. You conclude that your ten catches are good evidence for ‘large fish’ over ‘small fish’. But suppose you pull the net out of the water and discover that it has large holes, such that any small fish you caught within would have been able to escape. In that case, your catch of fish is no longer good evidence for ‘large fish’ over ‘small fish’. In general, we have to take account of how we collected our evidence when we evaluate different hypotheses against that evidence. Sober claims that the lesson of Eddington's fish is that the manner in which we collect evidence must be taken as background information - we hold that information as fixed, as given, when we then ask what the various hypotheses predict about our observations. Thus, given that the net has large holes, even the small fish hypothesis leads us to expect catching large fish only, to the same degree as on the large fish hypothesis.

But then, turning to the FTA, if we include in the background information something akin to, ‘I exist at t to observe a fine-tuned universe,’ then the likelihood of fine-tuning given theism or naturalism is the same, because now the question becomes: given that I exist to make observations, what do naturalism and theism predict about a fine-tuned universe? Well, given that I exist, both hypotheses predict fine-tuning to the same degree - namely, 1. If I exist, it follows automatically that the universe is such that observers can exist within it!

Most people will suspect that Sober’s reasoning has gone off the rails somewhere - and I agree. Matthew Kotzen in particular has a lovely article pointing out where Sober's reasoning fails. He concedes that while it “is a truism of scientific and philosophical methodology that, when we collect evidence, it can matter how that evidence was collected, and that ignoring `background information' about how our evidence was collected can lead to irrational conclusions” (p. 825), nevertheless he disagrees with the way in which Sober suggests we take account of that information.

Kotzen asks us to “[s]uppose that I know the following: Just after the egg that was to become me was fertilized, doctors took that fertilized egg and placed it in a petri dish. It's not known what happened after that - either the doctors stored the fertilized egg carefully so as to make sure that it wouldn't be harmed [the design hypothesis] or they threw it randomly around the lab in a way that made it very likely that the egg would be seriously damaged [the chance hypothesis]. Again, it's quite intuitive that my existence at a time t well after my birth [the evidence] is evidence for [design] over [chance] because [the likelihood of the evidence given design is higher than that likelihood given chance]” (p. 832).

But if Sober were correct, and we took as given that I exist at some time t to see that I am alive, then we could make no inference whatsoever between the design and chance hypotheses in this case, which is silly.

Still, where does Sober's mistake lie? With Kotzen, I suggest that Sober mistakenly includes information about the relevant method of observation in the background information, rather than as additional evidence. We ought to take none of it as given and instead just include all the claims as different lines of evidence that each hypothesis must explain. In the article, Kotzen shows how this analysis reproduces the right results in the Eddington fishing example. In general, Kotzen makes the point that only when the relevant method of observation is probabilistically independent of the hypothesis can you take it as given. So, in the fish case, the likelihood of you using a net with large holes is independent of what kind of fish are in the lake - and so you can rightly take that fact about the net as background facts. But when the method of observation is not independent, then you get all sorts of silly inferential mistakes if you fix facts about method as given. So, in the case of deciding whether your current existence is evidence that your doctors took great care of the fertilized egg that would become you, the relevant fact of ‘I exist at a time t to make observations’ is not independent of which hypothesis, chance or design, is true. The probability of your very existence at a time t to make observations is itself very low if the doctors haphazardly tossed the fertilized egg to and fro, and it is much higher if instead they took care to make sure that you developed under controlled conditions.

In the case of fine-tuning, the evidence then becomes, ‘I observe a fine-tuned universe,’ a piece of evidence that will have a very low likelihood given chance, and a much higher likelihood given theism in parallel to the fertilized egg thought-experiment. Whether you call the evidence, ‘a fine-tuned universe exists’, or ‘I exist to make observations of a fine-tuned universe’ makes no difference.

How does Joe reply to thought experiments that bring out the counterintuitive implications of Sober's analysis? His basic point is that we have no reason to think that the likelihood of fine-tuning given chance is low. We have never observed the creation of a universe and so have no idea what the probabilities involved are, whereas we have observed games of Poker, for instance, and so know the odds of different card hands. But this is a different objection to the FTA. Now the problem is that we have no idea what the odds are of a fine-tuned universe given chance. But that issue is independent of Sober's anthropic concerns - it just concedes that the original objection fails and offers a new one. The first reply therefore collapses into the second one that Joe considers, to which we now turn.

#2 The Art of the Possible

Once again, we have a few arguments under this heading. So, off to the first one.

Probability 0

Joe repeats Mark Colyvan’s objection that if there are an infinite number of possible universes, then the prior probability of any universe will be 0 or infinitesimal.¹ But if the prior probability of a fine-tuned universe is 0, then it doesn’t matter how much more likely theism makes fine-tuning than naturalism, because any number multiplied by 0 is still 0. This is a very rough statement of the issues, but the details need not detain us, because the objection is in any case badly confused.² A proposition can have a likelihood of 0 given some hypothesis while having a non-zero likelihood given a different hypothesis. What is the probability of God's raising Jesus from the dead given naturalism? 0. What is that probability given theism? Some non-zero value, even if one thinks that it is small. Similarly, let us say that there are an infinite number of possible universes and that all are equally likely if naturalism is true. Then, what is the likelihood of a life-permitting universe given naturalism? 0. But what is the likelihood of a life-permitting universe if theism is true? Not 0, because we do not treat all possible universes as equally likely if theism is true; rather, we privilege life-permitting universes with a greater share of the probability space, because God is more likely to create a universe with life than one without.

I haven't actually read Colyvan's paper; I'm relying on Joe here.

To be possible or to not be possible.

Joe’s second argument is that we cannot know if other values for the constants are really possible. The FTA needs a stronger sense of possibility than logical possibility, what Joe calls real possibility. Okay; but why can't we say anything about the real likelihood or possibility of a non-life-permitting universe? This is where Joe’s point about our not having observed universe creation becomes relevant: we know that it was really possible to make a cup of coffee instead of drinking green tea, for example, because we have observed people drinking green tea all the time. But we have never observed a non-fine-tuned universe emerging from whatever cosmic process brought about our universe. So, we cannot really know if other values for the constants and initial conditions are really possible. But notice that this line of reasoning threatens to unseat all sorts of scientific inferences. We also have never observed a universe where the general theory of relativity does not hold; does that mean we cannot say anything about the probability of the theory given our empirical evidence? Surely, we can. Or again, suppose that scientists detected various temperature fluctuations in the cosmic microwave background radiation, and that when they translated these fluctuations into binary code, they spelled out the first thousand digits of pi. We would all conclude that we had pretty powerful evidence for some sort of cosmic intelligence, despite the fact that we had never observed a universe where the background radiation had had a different arrangement of temperature fluctuations.

Let's dig deeper. I take it that Joe's fundamental point is that for all we know, the present values of the constants are physically necessary; that is, the laws of nature dictate that the universe must permit life. So, for instance, while it is logically possible for an object to travel faster than the speed of light, such is not physically possible because superluminal velocity contradicts the special theory of relativity.

Might we say the same thing about the values for the constants? Well, those values do not follow from any underlying theory of current physics. That's why the electron mass, the various quark masses, the strength of the gravitational force, and various initial conditions like the rate of the universe's expansion and the amount of disorder in the early universe are all called free parameters. Current physics does not determine the values for any of these constants or quantities.

But the proposal is probably that future physics will determine these constants and conditions. We know, after all, that current picture of physics is incomplete, and that we need a quantum theory of gravity to unify quantum mechanics and general relativity. Maybe that theory will show that all constants are physically necessary. Well, maybe! But given that we have no reason to think that a final theory will fix the constants, we ought to conclude that the likelihood of the constants turning out to be physically necessary is very low. Matthew Adelstein gives the following thought-experiment: imagine someone, whom I will call ‘Frank,’ receives five royal flushes in a row in a game of Poker. We would conclude that Frank had cheated. But suppose Frank responds by claiming that for all we know, a final theory of physics will show that this outcome - his getting five royal flushes at this day and time - is physically necessary. This would be a most unimpressive response, Frank.

All of this goes to show that in the context of the FTA, we are concerned with the epistemic probability of various outcomes, given naturalism and theism. That is, we are concerned with our degree of expectation given the truth of the relevant hypotheses - a degree of expectation that is often not rooted in considerations of statistical frequency. So, when one says that the general theory of relativity is highly likely given the available empirical evidence, one does not mean that in a majority of universes we have observed containing said-evidence, general relativity holds true; we mean, rather, that GR leads us to expect this evidence to high degree, much higher than any alternative. In the case of someone's receiving a royal flush, the epistemic probability that this outcome is the result of physical necessity is exceedingly low. I do not see any reason to treat fine-tuning differently. In the absence of evidence to the contrary, we ought to treat all values for the constants and initial conditions as equally likely from our epistemic vantage point - that is, we apply a principle of indifference - and this application will then yield the conclusion that the likelihood of fine-tuning given naturalism is very, very low. Now, two caveats. First, I myself have raised a few doubts about Adelstein's argument. These issues are ones that I am currently thinking through. Still, it does seem to me that in the absence of reasons to think that the constants are derivable from fundamental physics, we ought to place a low epistemic probability on the claim that they are derivable in that way. How we cash that out and quantify it, remains to be seen. Second, I need to think more about the principle of indifference, because it gets a lot of grief. So, that represents an area in need of further development in my reflections on the fine-tuning argument.

God wouldn’t fine-tune.

Okay third sub-objection: if God exists, why is it the case that only a tiny subset of constants can support life? This objection rests on a misunderstanding, one that also features in Hans Halvorson's critique of the FTA: when one says that the values of the constants are unlikely, one means that they are unlikely given chance, not that they are unlikely period. If I walk onto the beach and see ‘S.O.S.’ etched into the sand, this outcome is improbable given chance processes - but it is not improbable given that someone, probably stranded on the beach, dug into the sand to arrange a message therein. So, theists need not claim that God “made it highly improbable” that a life-permitting universe obtains or anything like that.

#3 Psychoanalyzing God

Joe’s third objection is that we have no reason to think that God would likely create a fine-tuned universe. Therefore, however low the likelihood of fine-tuning is given naturalism, we have no reason to think that it is any higher given theism. The likelihood in the latter case is just inscrutable.

Now, I have responded in my published work by claiming that if life is objectively valuable, then, because God is perfectly good, it is not terribly unlikely that He would create a universe where conscious agents could arise. Doing so would be a good thing to do.

Joe responds to me by name, for which I am flattered! His response is (1) that on many accounts about the relationship between God and goodness, God determines what is good and what is not, meaning that we will just ask all over again why God decided that life is good, and (2) this response seems to end up undermining theism, because it leads us to expect that God would create a lot more life in the universe than He has.

With respect to (1), I don't think we need to get distracted by conversations about how God relates to morality. On the only plausible account about how this might work, God's nature grounds moral value, and God does not decide what properties His nature exemplifies; rather, God has His nature essentially.

But what about point (2), that my argument proves too much and raises further questions about God's psychology, such as: why would God not create a lot more life than He has? Why is most of the universe arranged not to support life? Theism seemingly predicts a universe with a much higher proportion of life than what we observe. I don't think this argument works. First, there is no finite upper bound on how much life God could have created. So, even if the universe were teaming with life, it would still have been possible for God to have created more of it, and therefore there would still be room for the Joe-inspired skeptic to reply that surely, if God wanted life to exist within the universe, there would be even more of it. That point suggests that the objection has gone wrong somewhere if, no matter how much life there was, one could always have issued the objection in question.

Second, even if we grant that the rarity of life is evidence against theism, it is plausibly outweighed by fine-tuning. Let's agree with Luke Barnes that the likelihood of fine-tuning given naturalism is 1 part in 10^136 That's an outrageously small likelihood, and therefore fine-tuning is very strong evidence for theism over naturalism. Suppose we now want to compare that to the negative impact that the rarity of life has on theism. Well, what is the likelihood of the rarity of life given theism? Even if we say 1 in a billion, that evidential hit is nothing compared to the strength of the evidence for theism from fine-tuning. By analogy, suppose in a murder trial the defendant's fingerprints are on the gun and there is a video recording of him committing the murder. That's pretty powerful evidence that the defendant is guilty. But suppose that the defendant has an alibi - his boss testifies that he was at work during the time of the murder - that would be some evidence against the guilt hypothesis, but it is outweighed by the other evidence. I think the same thing happens here.

For more on this kind of objection to theism, I recommend Matthew Adelstein's helpful Substack article.

Conclusion

There you go, that's why Joe's three objections to the FTA do not succeed. I had fun thinking about these issues and consolidating my thoughts. I hope Joe will consider coming on my podcast for further discussion, for which I leave an open invitation.

1. I presume this is the paper Joe is referencing, but I cannot guarantee it. It’s the only one about fine-tuning from Colyvan I could find. Granted, I didn’t look that hard.

2. Or I am badly confused. I am open to correction.