• ****
  • 8713 Posts
  • Check out my new blog!
    • View Profile
    • Capturing Christianity
DEBATE TOPIC: Does the evidence of Fine-Tuning confirm theism over atheism?

Opening Affirmative Statement: Bertuzzi - 2000 words maximum
Opening Negative Statement/Response: Hatsoff - within 7 days - 2000 words max
Bertuzzi's 1st Rebuttal: within 7 days - 1750 words max
Hatsoff's 1st Rebuttal: within 7 days - 1750 words max
Bertuzzi's 2nd Rebuttal: within 7 days - 1500 words max
Hatsoff's's 2nd Rebuttal: within 7 days - 1500 words max
Bertuzzi's Final Statement: within 7 days - 1250 words max
Hatsoff's Final Statement: within 7 days - 1250 words max

**Please do not post in this thread. Feel free to post in the comment thread made specifically for this debate.
« Last Edit: August 10, 2016, 08:24:04 am by Bertuzzi »
Husband. Father. Photographer. Blogger.

"No theodicy without eschatology." - Hick



  • ****
  • 8713 Posts
  • Check out my new blog!
    • View Profile
    • Capturing Christianity
Re: DEBATE: Does the Evidence of Fine-Tuning Confirm Theism Over Atheism?
« Reply #1 on: August 10, 2016, 12:14:07 am »
I’d first like to thank the RF forums for hosting this debate and secondly, hats off to hatsoff for agreeing to debate this truly fascinating topic. In my personal investigation into the subject, I have become increasingly convinced that fine-tuning constitutes very good evidence for theism over atheism. In one sense, fine-tuning confirms what Christians have known for millenia.

“The heavens declare the glory of God;
    the skies proclaim the work of his hands.
 Day after day they pour forth speech;
    night after night they reveal knowledge.” - Psalm 19:1-2

“For his invisible attributes, namely, his eternal power and divine nature, have been clearly perceived, ever since the creation of the world, in the things that have been made.” - Romans 1:20

From these passages alone it’s clear that Christians have long held that God is revealed in nature. Over the last 30 years or so, this biblical truth, perhaps not too surprisingly for Christians, has enjoyed pretty strong extra-biblical confirmation through recent discoveries in physics and cosmology.

Before moving on to the argument, I’ll first lay out some key definitions and abbreviations that will be used throughout the debate.


Fine-tuning; Fine-Tuning evidence (FT) - includes two claims. (i) the laws of nature and constants of physics and the initial conditions of any universe with the same laws as our universe must be arranged in a seemingly very precise way for the universe to support life [1]; and (ii) such a universe actually exists.

Life Permitting Universe (LPU) - A universe that can support embodied moral agents.

Embodied Moral Agents - An embodied conscious being that can make morally significant choices.

Theistic Hypothesis (T) - This hypothesis holds that there exists an eternal, perfectly free, perfectly efficacious, perfectly knowledgable, perfectly loving creator of the universe whose existence does not depend on anything outside itself.

Naturalistic Single-Universe Hypothesis (NSU) - The hypothesis that (i) there is only one universe with physics that do not significantly vary from one space-time region to another and (ii) there is no transcendent explanation of the universe (supernatural or otherwise).

P(A|B) - Represents the conditional epistemic probability of a proposition A on another proposition B.

k and k’ - k refers to our total background information, whereas k' refers to some appropriately chosen background information -- e.g., for the case of the fine-tuning of the constants, the total background information minus the fact that a particular constant C has a life-permitting value.

Other symbols: << will always mean “much, much less than” -- e.g., “P(A|B) << 1 will mean that P(A|B) is very close to zero, since P(A|B) cannot have a negative value. ~P(A|B) << 1 will mean that it is not the case that P(A|B) << 1.


The fine-tuning argument is fundamentally about modeling. Suppose we have two separate models, M1 and M2, and a set of data D the models seek to explain. If D is highly coincidental on M1 and not highly coincidental on M2, at least with respect to D we’ve got pretty good reason to suppose that M2 over M1 is true.

For example, let D be fingerprints on a murder weapon (and these fingerprints correspond in several points to the fingerprints of the defendant). Let M1 be the innocence hypothesis and M2 the guilt hypothesis. D is highly coincidental or unexpected if the defendant is innocent but not unexpected if the defendant is guilty. Now, of course M1 might still be true, we could find evidence that the fingerprints were faked or that the computer software that analyzed the evidence was buggy at the time, etc. We’d still need to look at all the evidence. But relative to D, M2 is confirmed over M1.

Another example from science would be atomic theory. “What finally convinced virtually all physical scientists by 1912 of the atomic hypothesis was the agreement of at least thirteen independent determinations of Avogadro’s number based on the assumption that atomic theory was correct.” [1] There are other scientific examples one could cite, but I will not do so here.

The principle underlying these judgements is known as the Likelihood Principle (LP) which can be stated as follows: an observation e counts as evidence in favor of hypothesis h1 over h2 if the observation is more probable under h1 than h2 [1].

Robin Collins’ argument, the argument I will be defending, is built on this principle. For the sake of both comprehension and brevity, I’ve simplified the argument into just two premises and a conclusion.


(1) P(LPU|NSU & k´) << 1.
(2) ~P(LPU|T & k´) << 1.
(3) Therefore, LPU strongly supports T over NSU.

In English:

(1’) The probability of the existence of a life-permitting universe given the naturalistic single-universe hypothesis and our appropriately chosen background information is very close to zero.
(2’) It is not the case that the probability of the existence of a life-permitting universe given the theistic hypothesis and our appropriately chosen background information is very close to zero.
(3’) Therefore, the existence of a life-permitting universe strongly supports the theistic hypothesis over the naturalistic single universe hypothesis.

Is this argument logically valid? It doesn’t appear immediately so, but (3) follows from (1) and (2) given LP. So faulting the argument for invalidity isn’t really an option.

Let’s now look at the evidence for FT.


The evidence for fine-tuning falls into 3 categories (this is important because detractors of the argument usually only level objections against (ii)):

(i) The Laws of Nature
(ii) The Constants of Nature
(iii) The Initial Conditions of the Universe


For any embodied moral agents to exist, there must exist certain laws of nature. If this set of laws did not exist, then complex, self-reproducing biological organisms would be impossible. If, for instance, a universal attractive force such as gravity did not exist, then there would be no stars. Gravity is what holds stars together. If there were no stars, then there would be no long-term energy sources, making the evolution of conscious moral agents virtually impossible. Moreover, without Gravity there would be no planets (or if there were, then agents couldn’t walk on their surface, they would simply float away).

Other examples of these sorts of laws are: the strong nuclear force, the electromagnetic force, Bohr’s quantization rule, and the Pauli-exclusion principle. If any one of these laws did not exist (and were not replaced with some kind of similar law) then embodied moral agents could not evolve anywhere in the cosmos. Thus, life requires a particular set of laws of nature.


Gravity is the weakest of the forces. If the strength of gravity were increased a billion fold (relative to the strength of the electromagnetic force), objects the size of humans would be crushed. Martin Rees has said that in such a world insects would need incredibly thick legs to support them and no larger animals could survive [2]. Now, a billion times increase seems like a lot, but the strong nuclear force is 10^40 times stronger than gravity. Relative to the density of matter in the early universe, a change in the strength of gravity by an estimated one part in 10^60 would have resulted in the universe exploding too quickly for stars and galaxies to form or collapsed back in on itself [1].

To put this into perspective, the likelihood of getting the strength of gravity right for life is nearly the same as being dealt 10 Royal Flushes in a row [3]. This level of fine-tuning defies credulity.

Another example of a fine-tuned constant is the cosmological constant (CC). This is probably the most widely discussed case of fine-tuning. Crudely, the cosmological constant measures the acceleration rate of the early universe. If the acceleration rate were larger or smaller by one part in 10^120 [1], space would have either expanded too quickly for stars and planets to form or collapsed back in on itself.

The put this figure into perspective, the likelihood of getting the CC right for life is the same as being dealt 20 royal flushes in a row.


By far, the most impressive case of fine-tuning is the low entropy condition of the early universe. Roger Penrose has calculated that the mass-energy distribution of the early universe must be finely-tuned to within one part in 10^10^123 [1]. This figure boggles the mind. Getting entropy right for life by chance is like being dealt 211 Royal Flushes in a row [3].


5.1 DEFENSE OF (1)

In the case of the laws of nature we could appeal to an epistemic probability judgement. Probability judgements in the cases of evolution, atomic theory, and so on, didn’t involve appeals to rigorously justified statistical probabilities. There were simply judgements made about what kind of world we’d expect under each competing hypothesis and then those judgements were trusted. When contemplating what we’d expect given NSU, it doesn’t appear that we should at all expect a particular set of laws that happen to perfectly allow for life.

In the case of the initial conditions of the universe, we could appeal to the standard measure of statistical mechanics. Under the standard measure, the particular state of entropy that describes our universe is as likely as the tiny portion of phase space it occupies. Roger Penrose has calculated this and arrived at the figure mentioned earlier (1/10^10^123).

In the case of the constants of physics, we’d simply compare the life permitting range (Wr) of that constant C to the total range of values (WR) it could take on [4]. Thus, Wr/WR << 1. And since NSU and k’ do not entail that the constant will fall into one part of the comparison range instead of any other part, it follows from the principle of indifference that the probability of LPU given NSU & k’ is less than or equal to Wr/WR.

Given the 3 kinds of fine-tuning mentioned, it seems abundantly clear that (1) is true.

5.2 DEFENSE OF (2)

A defense of (2) requires pinpointing some reason God would have for bringing LPU about. Since God is perfectly efficacious, to bring X about, He simply must will that X come about. In the same way I simply will my arms and hands to move as I type this. And as a personal explanation, God will have reasons for doing what He does. So if we have any sort of plausible reason to suppose it not unlikely that God would bring about LPU, then (2) is true.

Suppose we had a good theodicy for the problem of evil. In that case we would think God has good reason to bring about LPU (since physical pain and natural evils require a physical universe). But suppose we had only a defense instead of a theodicy. In that case we would have no conditional probability on LPU given T. But then (2) is still true.


If I had space and time I would detail some standard objections to the argument, but instead, I shall simply wait and let hatsoff present his objections. That, I think, will be a better use of our time.

In summary then we’ve seen the structure of the fine-tuning argument relies on the likelihood principle. A principle that is used all the time in science and everyday experience. We saw that the argument itself is rather simple and both premises seem true. We saw that the fine-tuning evidence itself is quite shocking in terms of how improbable LPU really is under NSU.

In essence we’ve seen the biblical truth noted at the outset confirmed through science.



[2] Rees, Martin. (2000). Just Six Numbers: The Deep Forces that Shape the Universe, New York, NY: Basic Books.

[3] To calculate this simply use the following formula: 1E60 = (649739)^n. 1E60 can be replaced with any fine-tuning variable (for instance, 1E120 for the cosmological constant). 1/649739 is the probability of being dealt a Royal Flush by chance from a properly shuffled deck.

[4] Collins, I think rightly, holds that the comparison range be equal to the “epistemically illuminated” range. The EI range, as he calls it, is the set of values for which we can know the values are life-permitting or not.
Husband. Father. Photographer. Blogger.

"No theodicy without eschatology." - Hick



  • ****
  • 6273 Posts
    • View Profile
Re: DEBATE: Does the Evidence of Fine-Tuning Confirm Theism Over Atheism?
« Reply #2 on: August 11, 2016, 08:49:19 am »
I.  Introduction.

Thank you to Bertuzzi for suggesting and participating in this debate.  I want to say up front that due to the adversarial nature of this format, I might argue a little more aggressively than I usually do in discussions about religion. To quote Kevin Scharp: "I don’t pull punches, but I also never attack character." Bertuzzi knows that I respect him and that I value his opinions when it comes to philosophy, but he also knows that I disagree vehemently with some of those opinions.  In what follows, we shall explore a few such points of friction.

Before addressing the individual points of the argument, let us make a few observations about its general character.  Folks on both sides of the issue have expressed sympathy for the project due to its alleged respect for the scientific method. And yet, this is precisely the opposite of what is actually going on. Fine-tuning is not predicted by theism, nor does it confirm any hypotheses motivated by belief in God. Instead, the apologetic project, here as elsewhere, results from an attempt to justify preexisting religious beliefs. It deserves, therefore, the unenviable label of "confirmation bias." The scientific veneer of the fine-tuning argument is just that---a deceptive illusion of respectability hiding the confirmation bias underneath.

That said, despite this overarching flaw in fine-tuning as an apologetic project, I plan to focus mostly on its problems as they occur with Bertuzzi's specific formulation.  In particular, I will argue that his defense of premise (1) relies on an implausible probabilistic principle, that his defense of (2) is incompatible with other parts of his argument, and that the rule of inference used to get his conclusion, (3), is inappropriate to the task. Finally, I will discuss how (3) is ostensibly irrelevant to the present debate topic.

II.  Category (i).

Bertuzzi divides the fine-tuning evidence into three categories (i)-(iii).  Category (i) involves the very existence of the laws of nature, but I will outline two reasons for striking (i) from consideration in his argument.

Recall that we need to fix appropriate background information k' in order to run the argument. To assess the probability of the physical constants falling in the life-permitting range, Bertuzzi suggests that we merely strip from our total background knowledge the values they actually have.  Collins himself is a bit more precise:

"k' can be thought of as including the initial conditions of the universe, the laws of physics, and the values of all the other constants except C" ([2, p343]).

This means the probability of potentially-suitable laws of nature existing, on k', is precisely 1. In order to salvage this part of the argument, then, we will need to use a different k' which excludes them. In that case, we no longer have natural variables in the other two categories to which we can apply a principle of indifference (hereafter, "POI"). So, it seems there is no obvious choice of k' which is compatible with all categories (i)-(iii).

The second and perhaps more serious difficulty with category (i) is that we have no obvious way of evaluating the epistemic probability of NSU even when given a better choice of k'. A POI won't do, as there are no natural variables in play. And while it might be true that we have no reason to expect suitable laws of nature under k', Bertuzzi needs to show much more:  that we should expect, with high confidence, for there not to exist such laws.

III.  Indifference.

A better but still deeply flawed case can be made for (1) by appealing to the other two categories, (ii) and (iii).  Here we actually have some natural variables and epistemically-illuminated ranges to which the corresponding ranges of life-permitting values may be compared.

However, Bertuzzi needs some kind of probabilistic principle to make use of these comparisons. It is well-known, after all, that having a specified range in which the constants must fall does not by itself yield any particular probability that they will do so.  To infer the latter, it would be very nice if we had some information about what kind of factors determine the values of the constants in question, given background information k'.  Unfortunately, such information remains, so far, unavailable.

Instead, to infer probabilities from the ranges, Bertuzzi employs a POI, presumably similar to that of Collins.  It turns out, though, that POI's have serious problems. Collins understands this, and, in an effort to avoid them, proposes a "restricted" POI (hereafter, "RPOI") applicable only to natural variables---that is, variables which stand out as uniquely most natural when considering possible outcomes.

Is Collins' restriction good enough to justify using a POI? I can find no obvious reason to think so.  Neither is it clear why it is more natural, when applying a POI, to consider the ranges of particular constants in a physical theory rather than its qualitative character.  For example, we could, in contrast to Collins' method, apply a POI to the two-valued question of whether the universe will admit complex physical systems, giving us a probability relatively close to 0.5 instead of the absurdly low values suggested by fine-tuning proponents.

Collins' RPOI also suffers from certain technical problems which commit him to extraordinarily implausible claims.  Namely, in order to defend the RPOI from the well-known normalizability objection of Lydia and Timothy McGrew, he argues that it is fully rational to believe that some given outcome is physically possible despite being absolutely sure that it will not occur.  Even worse, he admits that it should be rational for a person to be certain that any given one of an infinite set of outcomes will not occur, while simultaneously being cerain that at least one of them will (see [1, section 5.2]).  So, even if we are tempted to consider the RPOI, we must weigh whatever intuitive plausibility we find in it against the enormous implausibility of such seemingly outrageous scenarios.

IV.  Premise (2).

Bertuzzi suggests that it is enough to find a "plausible reason" to accept (2), but stops short of providing any such reason.  A good theodicy, he notes, will suffice, but no good theodicy is given, nor is there any obvious candidate available which is well-accepted by philosophers.

As an alternative, Bertuzzi suggests that there might be no epistemic probability at all for LPU given T&k', thereby making (2) trivially true.  However, that suggestion invites two very serious problems.  First, it is not clear why the nonexistence of that probability value is the only alternative to having a good theodicy.  Second, and considerably more seriously, denying that P(LPU|T&k') has any value at all prevents us from applying the law of likelihood (hereafter, "LL"), which requires us to compare the values of the probabilities in question.

V.  Likelihoodism.

In order to get (3), Bertuzzi needs more than the typical rules of inference like modus ponens or universal instantiation.  He claims to employ the likelihood principle (hereafter, "LP"), but that isn't right, either. Actually, the only obvious way to get what he needs is to press into service the highly-controversial LL. Roughly, LL states that evidence E supports hypothesis H over hypothesis J iff r:=P(E|H)/P(E|J)>1, where r measures the degree of support ([4, p32]).  (Note that LL is almost exactly what Bertuzzi mistakenly describes as LP, only with the addition about quantity r, required to achieve the "strongly" qualification in (3).)

Collins understands that he needs LL (although he calls it LP), and he furthermore appreciates its controversial nature, especially insofar as a myriad of counterexamples are given in the literature.  Due to these counterexamples, Collins proposes another "restriction" to make LL more palatable.  This time, he proposes that the hypothesis H being confirmed must not be ad hoc.  So, for instance, according to Collins, if H is widely advocated prior to observing the evidence then we may apply LL.

Yet counterexamples still abound, even for Collins' restricted version of LL (cf., e.g., [3]).  Here, absurd conclusions follow from hypothetical applications of LL, and they are not made less absurd by the hypotheses in question having been widely advocated prior to observing the evidence.

One can always impose further restrictions on LL to get around the given counterexamples, and then a likelihoodism skeptic could in turn try to produce additional counterexamples to circumvent the new restrictions.  Hopefully the present debate will not degenerate into such a seemingly-endless cycle.  Instead, I suggest we learn two lessons from the likelihoodism controversy:  First, our general principles should be subordinate to common sense unless we have outstanding reason to overturn the latter.  Second, we should not be overly confident that a controversial principle with limited application is appropriate in any given case.

In the case of the fine-tuning argument, common sense tells against using LL the way Collins needs.  His motivation for structuring the argument around k' (instead of using the information we actually had prior to discovering fine-tuning) seems to come from his desire to show that the evidence points the way he wants it to point.  A convincing test of the evidence, in contrast, should stand independent of preexisting bias.

VI.  Atheism or NSU?

Bertuzzi's conclusion, (3), is clearly insufficient to the present debate topic, where we discuss whether fine-tuning confirms theism over atheism. Instead, (3) only delivers support (not confirmation), and the support in question only applies when comparing theism to NSU. To quote Paul Draper, "Why on earth, then, would any atheist choose to defend single-universe naturalism instead of naturalism?"

Even when comparing theism to NSU, the significance of Bertuzzi's conclusion (3) will still largely depend on individual perspective, where the competing hypotheses can have wildly variable prior probabilities. Now, if you're already a God-believer, or if you're the kind of agnostic who finds the existence of God at least somewhat plausible, then you might be impressed by the strength of support for theism over NSU claimed in (3) (although here I must stress the word claimed).  But what if you go into the debate as a naturalist?

Again, how significant one finds (3) will vary from person to person, but I think it's fair to say that many if not most naturalists---whether rightly or wrongly---are deeply prejudiced against theism, so that it will take more than (3) to overturn their consideration of NSU.  To put things in perspective, let us consider Christian theism as an example.  Some of us, myself included, are so confident in the falsity of Christianity that we remain utterly unphased by the prospect of eternal torment in hellfire.  Is our confidence in the falsity of general theism much lower? Not in my case. For us, the prior probability of theism is not overcome even by such extreme values as alleged in (3).

VII.  Summary.

In sum, we have seen that, far from being grounded in the scientific method, the fine-tuning argument relies principally on confirmation bias, an anti-scientific strategy aimed at justifying, in this case, preexisting religious beliefs. Theism has not made any predictions which have been borne out by evidence, nor has science confirmed any novel hypotheses motivated by theistic belief.  We have seen furthermore that neither of the two premises (1) and (2) in Bertuzzi's formulation of the argument have obvious justification, and that the rule of inference LL used to get the desired conclusion (3) suffers from serious problems. Finally, we have noted that (3) itself does not actually do the work required of it, to confirm theism over atheism.  In light of all this, it seems we cannot take fine-tuning seriously as an argument for theism over atheism.


[1]  Collins, Robin.  "How to Rigorously Define Fine-Tuning" Philosophia Christi 7 (2005), pp382-407.

[2]  Collins, Robin.  "The Teleological Argument: An Exploration of the
Fine-Tuning of the Universe," The Blackwell Companion to Natural Theology (2009), Craig, Moreland, eds.

[3]  Fitelson, Branden. "Likelihoodism, Bayesianism, and Relational Confirmation," Synthese 156:3 (2007), pp473-489.

[4]  Sober, Elliott.  Evidence and Evolution: The Logic Behind the Science (2008), ISBN 978-0521692748.
« Last Edit: August 11, 2016, 09:08:15 am by hatsoff »



  • ****
  • 8713 Posts
  • Check out my new blog!
    • View Profile
    • Capturing Christianity
Re: DEBATE: Does the Evidence of Fine-Tuning Confirm Theism Over Atheism?
« Reply #3 on: August 13, 2016, 09:41:14 pm »
I would first like to thank hatsoff for a substantive first rebuttal. I especially appreciate his honesty when he says that although he respects my opinions, he “vehemently” disagrees with them. While I don’t “vehemently” disagree with his opinions, I do respect them (wrong as they are).

In his introduction he claims that the fine-tuning argument is guilty of “anti-scientific” confirmation bias because it attempts to justify pre-existing beliefs. It appears hatsoff isn’t off to a great start. Confirmation bias is a property of persons, not of arguments. In his introduction he’s made a category error. Should he want to correct this, several other problems arise [1].

Now, even if hatsoff is correct here, it doesn’t undermine any of the premises in the argument. So it’s completely irrelevant. I propose that we focus on the substance of the debate moving forward.

On “Category (i)”.

Hatsoff claims there is no obvious definition of k’ that is compatible with all 3 categories of fine-tuning. Unfortunately for hatsoff, there is an obvious definition that is compatible with (i)-(iii). Namely, all that we know minus what we know about LPU.

Note that getting rid of (i) doesn’t solve the problem since the initial conditions are also given in the quote from Collins. So it’s really an argument, if successful, to use only one category of fine-tuning. Not an argument expressly against using (i). For we might just use (i) and dispose of (ii) and (iii). But such measures aren’t necessary. And even if they were there are other options available, none of which lead to abandoning the fine-tuning argument [2].

As for how to defend (i), it’s unfortunate he didn’t respond to what I already said in defense of it. I didn’t appeal to a principle of indifference in defense of (i), I appealed to an epistemic probability judgement. The same sort made all the time in science (re: evolution, atomic theory, and so on). Here he might choose to express incredulity, but that won’t do as an actual objection.

So it doesn’t appear there is any good reason to abandon (i).

On “Indifference”.

In this section Hatsoff expresses his incredulity re: Collins’ restricted principle of indifference. But incredulity has never amounted to an actual argument against a position. He then says “we could apply a POI to the two-valued question of whether the universe will admit complex physical systems, giving us a probability relatively close to 0.5.” Not exactly. The principle of indifference is applied in situations where we have no reason to favor one option over another. To get to that point in this example we’d need to throw out our current scientific understanding of the cosmos. There’s simply no reason to do so.

Secondly hatsoff seems to think that the Vestrup/McGrew normalization problem is a problem for the restricted principle of indifference simpliciter. It’s not. The normalization problem is a problem for infinite (or unbounded) comparison ranges. Probabilities can still be evenly distributed over finite ranges. And there are several non-arbitrary proposals in the literature for such ranges [3]. Collins’ Epistemically Illuminated (EI) range approach is well-known (it’s also the approach he uses).

So it doesn’t appear there is any good reason presented to abandon the RPOI.

On “Premise (2)”.

Hatsoff seemed unsatisfied with my brief defense of (2). Not a problem. I will in this section defend the Soul Building Theodicy (SB).

SB is not, as some believe, about God compensating victims with goods that outweigh the badness of their suffering. According to sophisticated versions of SB, to defeat evil is to integrate it into a morally valuable whole that is incommensurably better than it could be without evil. “Evil is not a means to the good, but it is the cost of there being that good.[4]” It is in this way that evil is defeated.

The theodicy itself requires only one axiological assumption [5]. And that is that the best goods are the authentic display of love-manifesting virtues (IOW saintly virtues). Here it will help to consider some concrete examples. Consider the case of Saint Damien of Moloka’i. He spent 16 years in the Hawaiian islands serving lepers before succumbing to the disease himself. Secondly, consider Saint Maximilian Maria Kolbe. He was eventually taken to Auschwitz by the Germans for sheltering thousands of Jews. Then while in Auschwitz he volunteered to take the place of a man being lead to his death via starvation in a hole as an example. It is clear that these men’s lives added immensely to the value of the world.

As for how theism predicts suffering and hence LPU, see the argument below.

Let T be theism. Let K be a world with a significant number of opportunities for sainthood and not insignificant number of saints. Let S be the saint-seeking story according to which in creating humans God sought saints and all that entails. Let EH be the fact that since very near the beginning of human existence, there has been an abundance of intense human suffering. That is, wherever there has been human life, very significant levels of human suffering have been quite common.

a. T → K (From that God must create the best kind of world ensemble)
b. K → S (From the assumed value of sainthood)
c. S → EH (From that sainthood requires very significant trials)
d. EH → LPU (From that human suffering requires a physical world)
e. T → LPU (From a, b, c, and d)
f. If e, then P( LPU | T) = 1 (Theorem)
g. P( LPU | T) = 1 (From e and f)

On “Likelihoodism”.

Hatsoff begins by labeling LP as “highly controversial”. Indeed, LP is open to criticism, but to call it “highly controversial” seems a bit of a stretch.

Given that by evidence favoring one hypothesis over another, one means the ratio of h1 to h2 is higher after the evidence, i.e. P(h1 | e)/P(h2 | e) > P(h1)/P(h2), one can literally mathematically prove LP [6]. The traditional counterexamples are dealt with by other considerations, such as applying low priors to combat ad-hocness or by making it clear that LP applies only to incompatible hypotheses. Hatsoff is free to go down this route, but I do not think it wise for him.

It stands to reason that LP is a good probabilistic principle.

On “Atheism or NSU?”

I originally dealt with this objection in my opening but decided against it. I had hoped that if hatsoff denies NSU, he would at the very least mention what he does hold as an atheist. Instead he attacked the argument on other grounds. So it's not clear what he does hold.

Nevertheless, atheism is compatible with ~NSU. So what warrant does (3) provide for theism over atheism in the broader dialectical context?

The trouble is that “bare atheism” doesn't predict much of anything, indeed it doesn't even predict a universe at all [7]. What we do, regardless of the theory, is start with the assumption that the universe exists and is best described by our standard physical models. Over the last 30 years of so we’ve began noticing these models include various coincidences (i.e.: fine-tuning). A common option in dealing with these coincidences is extension. We can extend the model to account for the coincidences.

One such extension is theism. But what we’re after is an extension that is compatible with atheism. And remember, bare atheism lacks the explanatory resources to predict much of anything. NSU is the simplest and most natural extension for atheism. NSU gets us to a universe predicted by the standard model, but the problem is that NSU doesn't fare very well against it’s theistic alternative in explaining the coincidences.

If hatsoff has an alternative extension to NSU he’d like to throw out that (a) does a good job of explaining fine-tuning and (b) is compatible with atheism (and incompatible with theism), I’m all eyes.


In summary then, we’ve seen that hatsoff began his response by making a category mistake, which, if he chooses to correct it, opens himself to all sorts of criticisms. It would be wise, I think, if he dropped that kind of unproductive atheistic rhetoric altogether.

In his first objection we saw that he gave no good reason to dispose of (i). His challenge to define k’ wasn’t really a challenge at all. And then we saw that he hadn’t even interacted with the defense I gave of (i) in the opening statement.

Then we saw that he gave no good reason to reject the RPOI - his supposed counterexample was dealt with easily. From there he seemed to argue that normalization is a problem for the RPOI but we saw that it’s really only a problem for infinite comparison ranges.

Then, to bolster my defense of (2), I defended the soul-building theodicy. I argued that LPU is expected on T given only one plausible axiological assumption about the moral worth of love.

Then hatsoff said that LP is highly-controversial and so we should be weary of using it. There I argued LP can be mathematically proven and that the standard counterexamples are easily dealt with, so there’s no good reason to be weary of using it.

Lastly, hatsoff claimed (3) doesn’t establish that FT confirms theism over atheism. There I noted that atheism simpliciter lacks the explanatory resources to predict any universe at all. NSU is the simplest and most natural model that is both consistent with atheism and leads us to expect a universe described by our physical models. If hatsoff rejects NSU, he is free to offer an alternative.

Thus, we’ve seen that the argument I’ve presented remains unscathed. It is safe to say, so far, that the evidence of fine-tuning provides pretty strong evidence for theism over the explanatory atheistic alternative.


[1] First, not every person that attempts to justify pre-existing beliefs is guilty of confirmation bias (CB), for instance when evolutionary biologists look for more evidence of common ancestry. So he would need some reason to think that CB is present in this case. Secondly, as a Christian, my belief in God doesn’t depend on the success of the fine-tuning argument. Hatsoff on the other hand must find some way to avoid it’s conclusion. So it would seem there’s a better case for hatsoff suffering from CB. Thirdly, the burden of proof for hatsoff claim is so remarkably high, it’s virtually impossible he isn’t guilty of a hasty generalization. Something I imagine he is “vehemently” opposed in other contexts.

[2] For instance, one could run 3 separate arguments and produce a cumulative case against NSU. Or one could use a different kind of argument altogether, namely Collins’ method of probabilistic tension.

[3] Luke Barnes recently submitted a paper defending finite comparison ranges for many of the free parameters in physics.

[4] Dougherty, T. (2014-07-22). The Problem of Animal Pain: A Theodicy For All Creatures Great And Small (Palgrave Frontiers in Philosophy of Religion)

[5] This is likely the best area of attack. Note that this is perfectly in line with Christian moral axiology (Romans 8: 14, 17, Galatians 3: 25– 6, 4: 6– 7, Hebrews 2: 10– 11, John 3: 1). So running a problem of evil as an internal critique must assume it.

[6] I can provide said mathematical proof upon request.

[7] I'm defining “bare atheism” as the bare view (or proposition if you like) that “God does not exist”. From bare atheism it doesn't follow that a universe exists or probably exists. Additional propositions or assumptions would need to be added (ie: the proposition “the universe exists”).
« Last Edit: August 13, 2016, 09:49:21 pm by Bertuzzi »
Husband. Father. Photographer. Blogger.

"No theodicy without eschatology." - Hick



  • ****
  • 6273 Posts
    • View Profile
Re: DEBATE: Does the Evidence of Fine-Tuning Confirm Theism Over Atheism?
« Reply #4 on: August 20, 2016, 10:37:59 am »

Thank you to Bertuzzi for posting a timely and substantive first rebuttal, with minimal rhetoric.

Before I get to the meat and potatoes of my response, let me say something about my motivation in participating in this debate.  I have three chief goals in mind.

First, I want to use this opportunity as a springboard for learning more about the fine-tuning argument.  So, for example, one of the reasons it took me so long to reply was that I had to obtain and read a copy of Luke Barnes' paper which Bertuzzi referenced above.  This is an extremely interesting paper, although, as it turns out, it is almost entirely irrelevant to any of the points of dispute given below.

Second, I want to convince skeptic readers that the fine-tuning argument is a much worse one than they have been led to believe.  As explained in my opening statement, a lot of people---including skeptics---regard fine-tuning as one of the best and most promising lines of argument for theism.  However, in my opinion, it is actually one of the worst, as it is an exercise in justifying preexisting religious bias.  Even young-earth creationist arguments (e.g. the watchmaker analogy) are usually better than this. Bertuzzi complains that I am using the wrong label---"confirmation bias"---to refer to this problem, but that is no matter.  He is welcome to use whichever label he prefers.

My third goal is to convince believers that fine-tuning is not a good defense of their views.  I place this goal third on the list because it is the least realistic of the bunch.  Some readers, for instance, will be convinced of Bertuzzi's case before they have even read my response.  Others could get hung up on the tone in which I deliver my objections rather than their actual content.

Now, let's once more remind ourselves of the debate topic:  Does the evidence of fine-tuning confirm theism over atheism?  Remember that we are not discussing whether there is any good evidence for atheism.  That is a good topic to discuss, but it is a topic for another day.

That said, let us label and review my objections and Bertuzzi's responses, or lack thereof.

II-a. There is no choice of k' compatible with all three evidence categories (i)-(iii).

Bertuzzi's Response: He suggests we take for k' the data of "all that we know minus what we know about LPU." However, this effectively ignores the content of my criticism. Does k' contain information about the laws of nature or not---and, if it does, then how is the epistemic probability of the evidence in (i) obtaining anything but 1?  So, subtracting "what we know about LPU" seems far too vague to help us evaluate the epistemic probabilities.

II-b. Bertuzzi's probability judgment regarding category (i) is unsupported.

Bertuzzi's Response:  None relevant.  He has evidently misunderstood my criticism of his epistemic probability judgment about (i) as an argument from incredulity. He also seems to think I mistook him as wanting to use a POI with category (i).

Instead, I've pointed out that he hasn't provided any reason to accept the probability judgment he lays down.  As I explained previously, to use (i) in support of (1), he needs to show that we should expect, with high confidence, potentially-suitable laws of nature not to obtain given NSU&k'. He has not even attempted this.

III-a.  The RPOI is unsupported.

Bertuzzi's Response:  None relevant.  Here, too, Bertuzzi has misunderstood me, falsely accusing me of employing an argument from incredulity---this time, against RPOI.

Bertuzzi has a tough job, here:  He needs to show that RPOI (or some similar principle linking range ratios to probability values) is justified, at least in this case.  No attempt at this has been made, that I can tell.

III-b.  The choices of natural variables in the RPOI are unsupported.

Bertuzzi's Reponse:  None relevant.  Remember that I suggested that the two-valued variable of whether the laws of nature support complex structures (i.e., planets, stars, etc., and hence life) is just as natural, if not more so, than the variables representing unknown values of the constants left blank in k'.

"Not exactly," he answers, adding that this would require us "to throw out our current scientific understanding of the cosmos." But this reply seems to come out of left field, with no obvious connection to the issue I've raised about choosing natural variables.  He really needs to explain what he means, here.

III-c.  The RPOI has outrageously implausible consequences.

Bertuzzi's Response:  None---perhaps because he misunderstood the objection.  I agree with Bertuzzi that the normalization problem can be resolved by restricting to finite ranges.  However, this line of discussion, as seen in Collins' paper on the subject (reference [1] in my opening statement), reveals some deeper problems.  There, it becomes clear that in order to defend the RPOI, one must affirm some outrageously implausible consequences, as I described previously.  Namely, Collins affirms that it is rational to be certain that any given one of an infinite number of outcomes will not occur, while simultaneously being cerain that at least one of them will occur.  Bertuzzi has declined to respond at all to this very specific criticism.

IV-a.  Theodicy is implicated but not defended.

Bertuzzi's Response:  For the soul-building theodicy, we are furnished examples where men put themselves in harm's way for the sake of others. These selfless acts of sacrifice are indeed goods---but are they the best possible goods?  And even if they are the best possible goods, do they really outweigh the cost of introducing such terrible evils into the world, for instance the millions of acts of prolonged torture involved in the holocaust?  Bertuzzi says yes, but why does he think this?  We need some argument rather than just taking his controversial assumptions on faith.

IV-b. The claimed alternative to theodicy is undefended.

Bertuzzi's Response:  None.

IV-c.  The claimed alternative to theodicy is incompatible with LL.

Bertuzzi's Response: None.

V-a.  LL has well-known counterexamples.

Bertuzzi's Response:  Bertuzzi thinks it is unwise to get stuck in a cycle of counterexamples-and-restrictions for LL.  That's good because that's exactly what I argued in my opening statement!  However, that doesn't change the fact that many, many counterexamples are well-known, which even apply to the restricted version of LL that Bertuzzi wants to use.  He has not responded directly to any of these, as given in the reference [3] from my opening statement.

V-b.  Common sense tells against using LL in this case.

Bertuzzi's Response:  None.

[n]VI-a.[/b]  Support is not confirmation. 

Bertuzzi's Response:  None.

VI-b.  Bertuzzi's conclusion (3) about NSU does not, by itself, address the debate topic.

Bertuzzi's Response:  None relevant.  Instead, Bertuzzi demands an atheistic explanation for fine-tuning.  But why? Since the conclusion (3) of his deductive argument does not directly connect to the debate topic, does he now want to supplement it with an abductive one?

Here I think it is best to merely admit that we do not know why the universe appears to be fine-tuned.  I will leave it to the scientists to appeal to their expert experience in formulating the most promising hypotheses and testing them to see if they are true.  In the mean time, suffice it to say that I don't think that postulating an unembodied mind who created the universe using supernatural powers is a good explanation.  Let us instead wait and see what we can learn in the future, through the scientific method.


Bertuzzi has failed to respond adequately to any of the objections given in my opening statement.  In many cases, he didn't even attempt a response, and in other cases the response given is almost completely irrelevant and off-topic.  Only in (II-a), (IV-a), and (V-a) are Bertuzzi's responses actually on-topic, and in those cases they fall short for other reasons, as explained above.



  • ****
  • 8713 Posts
  • Check out my new blog!
    • View Profile
    • Capturing Christianity
Re: DEBATE: Does the Evidence of Fine-Tuning Confirm Theism Over Atheism?
« Reply #5 on: August 23, 2016, 09:37:09 pm »
Thanks to Hatsoff for the reply. I especially appreciate how he’s broken his objections down into sections and subsections. In effort to maintain clarity and flow, I’ll use the same headings for my responses.


In his introduction Hatsoff says that one of his goals is to convince skeptics the FTA is worse than they have been lead to believe. I doubt he will have much success here. The best objection to the FTA (the multiverse) hasn’t even been offered in this debate. Instead he’s focused on exploiting technicalities and calling into question widely held epistemic principles that are otherwise assumed in science. Hardly the sort of objections that will grab the attention of a science-revering skeptic.

Then he says he wants to convince theists that FT is not that great of an argument. Here again, I do not think he will have much luck. Not because theists are mindlessly uncritical or can’t look past his rhetoric as he arrogantly suggests, but because his objections will be seen as superficial, gratuitously skeptical, and weak. The best and most challenging objections to the argument have not even been presented in this debate.

II-a. There is no choice of k' compatible with all three evidence categories (i)-(iii).

As I’ve been thinking through this technicality, I’ve come to realize that Collins implicitly distinguishes between laws of nature and laws of physics (notice his switch of terms from section 2.2 to 2.3). A law of nature can be thought of as a general principle or a “causal power” which is irreducible to a mathematical formula. The laws of physics on the other hand are mathematical approximations. So, for instance, a universal attractive force such as gravity is a general principle or causal power that is mathematically approximated in Newton’s law of gravity F=G*m1*m2/r^2.

This distinction allows us to leave the laws of physics in k’, as part of our background information, and exclude the laws of nature. In response one might say that F=G*m1*m2/r^2 implies a law of nature. There's two problems with this response. First, G could equal 0 (a figure included in the comparison range), so that doesn't necessarily follow. Second, Newton’s law of gravity doesn’t imply the right set of laws of nature. If anything it implies only 1. But recall I mentioned at least 5 laws, each of which is required for life. Thus, Newton’s law of gravity doesn’t imply a single law of nature, nor the right set of laws.

And note that this objection is in fact a trivial technicality, since, as even hatsoff agrees, this issue can be easily avoided without abandoning (i). It is not a serious objection to the fine-tuning argument.

II-b. Bertuzzi's probability judgment regarding category (i) is unsupported.

Hatsoff’s demand for rigorously justified probability judgements would lead us to deny vast amounts of scientific knowledge. In the case of continental drift theory, for instance, there was never any requirement on scientists to demonstrate with “high confidence” the similarity of animal and plant life in Africa and South America would not obtain if continental drift were false. It was simply judged very unlikely that the similarity obtain if the theory were false and not if it were true. And this is precisely what we’ve got in the case of (i). It's an unlikely coincidence the right combination of laws of nature for life obtain if NSU is true.

If hatsoff wants to remain skeptical of vast amounts of scientific knowledge until his demands are met, no problem. He is welcome to join the ranks of science-deniers. But we are under no obligation to follow him on that.

III-a.  The RPOI is unsupported.

I will at this point do two things that will address concerns mentioned in both III-a and III-b. First, I will give an example that demonstrates the principle of indifference seems the only justification we have for assigning probability in everyday cases, and then secondly argue that Collins’ restriction to avoid Bertrand Paradoxes is used in scientific theory confirmation (and it would therefore be special pleading to allow it in the one case and not in FT).

First, suppose a factory has just produced (within the last minute or so) a perfectly weighted, perfectly symmetrical 50-sided die. Prior to this die being rolled we already know the probability of it landing on any given side. Namely, one in fifty. Since each side is symmetrically identical, we have no reason to believe it will land on any particular side. So we distribute the probabilities evenly over the possible outcomes. The principle of indifference seems to be the only justification we have for assigning probability here.

Secondly, according to Collins, quantum electrodynamics (QED) predicted the g-factor of the electron to within one part in a billion [2, section 3.3.3]. The important thing here is that QED’s prediction assumes Collins’ “natural variable” [3]. Prejudicing some other variable in the case of QED would mean that QED’s prediction of the g-factor doesn’t actually confirm QED over ~QED. Denying the merit of such an accurate prediction seems incredibly implausible. What this all means is that scientific theory confirmation has Collins’ natural variable assumption built in.

In the case of FT, there's no reason to prefer some other variable than what physicists and scientists already use in theory confirmation. It appears hatsoff’s radical skepticism has him once again pitted against science. This is what I meant when I said his variable would require denying much of what we scientifically know about the cosmos.

III-c.  The RPOI has outrageously implausible consequences.

To reiterate, the consequences Hatsoff cites are not consequences relevant to this debate - I have not appealed to infinite comparison ranges. Moreover, Hatsoff’s insistence that Collins defense is a “consequence” of the RPOI is patently absurd. His defense isn't the only defense out there. This objection is a complete and utter failure.

IV-a.  Theodicy is implicated but not defended.

Recall what is required to defend (2). All we need is to identify a plausible reason God might have for bringing about LPU. That is enough to show that P(LPU|T) isn't close to zero.

In the soul-building theodicy I presented, he questions whether the authentic display of love manifesting virtues is the best possible good. In my mind, the only candidate of a greater good on theism would be something like incarnation and atonement [4]. Which doesn't help Hatsoff since those require tremendous suffering as well. A question to the reader: can you identify a greater good on theism than love manifesting virtues?

He then questions whether soul-building outweighs the evil. But this misunderstands how evil is defeated in the model. The relevant question is whether the evil in the world is such that it plausibly isn't defeated. At best we can say that evil was plausibly not defeated in the lives of certain individuals at the time of their death. But we have no evidence whatever that evil will not be defeated in the afterlife. Note that sophisticated models of SB include defeated evils in the afterlife.

(2) therefore remains eminently plausible in light of SB.

IV-b. The claimed alternative to theodicy is undefended.

Due to limited space I'm willing to concede the point.

IV-c.  The claimed alternative to theodicy is incompatible with LL.

Same as above.

V-a.  LL has well-known counterexamples.

Unfortunately, hatsoff responded to precisely none of what I said here [5]. Also note that something like LP is used in my example above from continental drift. Here again, hatsoff is in the awkward position of denying science.

V-b.  Common sense tells against using LL in this case.

Hatsoff says common sense tells against using LL. And how does he justify this? By claiming that it’s just “common sense” that we should prejudice our state of ignorance prior to discovering FT. I’ll leave it to the reader to judge whether it’s just “common sense” to ignore new information (especially information revealed through science).

VI-a.  Support is not confirmation.

This is merely a difference in interpretation. I interpreted confirmation in the sense of LP.

VI-b.  Bertuzzi's conclusion (3) about NSU does not, by itself, address the debate topic.

It’s unfortunate hatsoff chose to handwave away what I said in response to this. Recall that what we are looking for is an appropriate extension of our standard physical models that is consistent with atheism. NSU is the simplest and most natural extension (and the one that is the most widely held) and hence why it was singled out. Unless and until hatsoff can present an alternate extension to NSU that (a) is compatible with atheism and (b) leads us to expect LPU to an equal or greater degree than T, it is safe to say that (3) obviously addresses the debate topic.


In summary then we’ve seen that hatsoff has offered no objections that do not also lead to radical scientific skepticism, are not the result of exploiting technicalities, or are not manifestly weak. (1) and (2) remain unscathed in this debate.

[1] The Bertrand Paradoxes are the well-known problem in probability Collins’ restriction avoids. Collins explains what they are in section 3.3.2 if you are interested.


[3] What Collins calls “natural variables” are the variables that occur in the simplest overall expression of the laws of physics.

[4] See Plantinga’s defense of Supralapsarianism.

[5] For reference: “Given that by evidence favoring one hypothesis over another, one means the ratio of h1 to h2 is higher after the evidence, i.e. P(h1 | e)/P(h2 | e) > P(h1)/P(h2), one can literally mathematically prove LP (in a footnote I offered to explicate said proof). The traditional counterexamples are dealt with by other considerations, such as applying low priors to combat ad-hocness or by making it clear that LP applies only to incompatible hypotheses.
« Last Edit: August 24, 2016, 10:27:31 am by Bertuzzi »
Husband. Father. Photographer. Blogger.

"No theodicy without eschatology." - Hick