Wednesday, May 9, 2018

Fine-Tuning: Barnes vs Malpass

Fine-Tuning: Barnes vs Malpass
< Prev   |   1   |   2   |   3   |   Next >


Capturing Christianity hosted a fine-tuning debate between Luke Barnes and Alex Malpass in 2018-04-21.

For those who don’t already know, cosmic fine-tuning is the observation that given the physical laws of our universe, certain constants (such as the mass of the electron) and quantities (such as the entropy of the early universe) in our universe are “fine-tuned” in the sense that the life-permitting parameters of the universe are extremely narrow, such that if they were altered even slightly physical life would not have existed. The fine-tuning argument (FTA) argues that the fine-tuning of the universe, i.e. that the universe is life-permitting instead of life-prohibiting in these circumstances, is evidence for theism (sometimes using a relatively modest conclusion; e.g. claiming that design is the best explanation for the universe being life-permitting).

In my opinion the debate was excellent, and the debate was mostly over philosophy rather than the scientific claims. Perhaps that was to be expected, since the theist arguing on behalf of FTA (Luke Barnes) is a cosmologist and the person arguing against it is a philosopher. I’ll do an overview of the debate, but since the debate is nearly two hours long I’ll obviously have to skip over some details.

Debate Overview

Opening Statements

Luke Barnes gives his opening statement starting at around 2:58 explaining fine-tuning and describing some of the science involved, giving examples of fine-tuning such as the masses of electrons and quarks (3:15 to 4:55, where if the parameters fall outside the narrow boundary not only do you not get life you don’t even get chemistry) and the Higgs field (6:23 to 7:12). Barnes says there are good reasons for God to make a morally significant universe (and thus a universe with life), whereas on naturalism the sort of universe we’d expect is a dead one (23:09 to 23:24).

Alex Malpass’s opening statement starts at around 25:15 and he more or less concedes the physics Barnes presents, at least arguendo (24:48 to 26:13) while noting that some disagree with the fine-tuning claim. Best I can tell though, while the scientific opinion isn’t unanimous, the consensus does seem to be that cosmic fine-tuning is real. I find it unlikely that atheist physicists like Stephen Hawking and Leonard Mlodinow would have affirmed fine-tuning if it weren’t real.[1]

To his credit as a philosopher, Malpass presents the overall structure of the fine-tuning argument better than Barnes does at around 30:36 to 32:00.
  1. P(L | N) << 1 (the probability that the universe is Life-permitting given Natrualism is much less than 1)
  2. ~P(L | T) << 1 (it is not the case that the probability that the universe is Life-permitting given Theism is much less than 1)
  3. If E is more probable on H1 than H2, then E supports H1 over H2.
  4. Therefore, L is evidence for T over N.
Something called the odds form of Bayes theorem will be helpful here, the general structure of which is this where e.g. P(T|L) represents the probability of T (theism) given L (the universe is life permitting).


The theist wants the posterior odds to be greater than the prior odds. Of particular note in this debate will be the values of P(T) and P(N), i.e. the prior probabilities of T and N, respectively. The lower P(T) is, the worse the posterior odds will be. This brings us to what Malpass calls the “Goldilocks hypothesis problem.”

Goldilocks Hypothesis Problem

One thing that affects the prior probability of a hypothesis (i.e. its probability prior to examining some particular data) is its content, i.e. how much it “claims.” For example, the claim “An animal exists” has less content than “A flatworm with two heads exists,” and so “An animal exists” has greater prior probability; it’s broader and has less content than the claim of a specific type of animal existing. Malpass correctly notes that there’s a trade off between the content of T, it’s prior probability P(T), and P(L|T).

If T were “An omnipotent God desires L” then P(L|T) will be high but at the cost of P(T) being low. To see why, let R represent “It rained this morning” and let TR be “An omnipotent God made it rain this morning” in which case P(R|TR) is high, but P(TR) will have to be low since we can’t reasonably infer that an omnipotent deity caused it to rain every time it rains. The narrower and more specific T is (thus packing more content in it), the lower the prior probability of T. Conversely, making T broader (having less content), e.g. T meaning nothing more than “an omnipotent being exists” will make P(T) greater; but making T broader in this way will make P(L|T) lower.

The Goldilocks hypothesis problem is how precisely to define the hypothesis T in such a way to not make it ad hoc (e.g. “An omnipotent God desires L”) which would give it a lower prior probability, but also not make it so broad that P(L|T) isn’t terribly large. Put another way, T can’t have too much content, but it also can’t have too little content. If T is defined too broadly with too little content, Malpass seems to think that there’s a risk that P(L|T) is equal to P(L|N) (32:32 to 33:58). So how to define T in a good way so that P(T) and P(L|T) are both not so small?

Barnes gives a decent response to this (40:38 to 43:56). God is a free and morally good being who would choose the best actions, and a morally significant universe with free agents would be good. So if T is more or less “standard” theism where God is not only supremely powerful but also good, then T isn’t ad hoc and P(L|T) is not unreasonably low.

The Stalking-Horse Naturalism Hypothesis

Malpass explains what he calls the stalking horse naturalism hypothesis at around 49:33 to 57:14). If the theist can give God some sort of disposition to make P(L|T) relatively high, couldn’t the naturalist do the same thing? The naturaliast could give naturalism a “mysterious disposition” (56:24 to 1:00:48) to result in a life-permitting universe, call this form of naturalism ND, such that P(L|ND) is high.

The disposition for God to make P(L|T) not that low seems reasonable on theism (God being good), but physical reality having a disposition to result in a life-permitting universe seems rather ad hoc and potentially does little more than push the problem back a step. To illustrate, suppose naturalism’s disposition is some factor x that results in the universe falling into the extremely narrow parameters as specified in a certain mathematical equation. Then it seems that factor x would itself be fine-tuned so that it points to one set of narrow parameters rather than another set of parameters. In other words, naturalism’s disposition would itself have to be finely-tuned to be disposed to have one narrow set of parameters instead of some other set of parameters.

Barnes gives a somewhat similar objection (1:01:10 to 1:03:01). Barnes points out that we’d have to consider P(ND|N), i.e. the probability of naturalism having that particular disposition given N simpliciter. Naturalism could have had a disposition for a dead universe instead of a life-permitting universe, and when a set of parameters is chosen at random, there are far more dead universes than living universes. The impression I’m given is that Barnes is thinking that P(ND|N) is pretty comparable to P(L|N).

< Prev   |   1   |   2   |   3   |   Next >

[1] Hawking, Stephen; Mlodinow, Loendard. The Grand Design (New York: Random House, Inc., 2010), pp. 143-144, 157-162.


  1. I critiqued your arguments about God and time:

    1. Thanks! You misspelled my name (as of this writing), but that's OK. Makes me glad I wasn't the only one who made typos and mistakes in their blog posts. :-P

    2. Haha. I didn't realise that. I corrected it.