The Instigator
Thrasymachus
Pro (for)
Winning
3 Points
The Contender
InquireTruth
Con (against)
Losing
0 Points

The fine-tuning argument for God's existence is a failure

Do you like this debate?NoYes+10
Add this debate to Google Add this debate to Delicious Add this debate to FaceBook Add this debate to Digg  
Post Voting Period
The voting period for this debate has ended.
after 1 vote the winner is...
Thrasymachus
Voting Style: Open Point System: 7 Point
Started: 7/4/2011 Category: Philosophy
Updated: 5 years ago Status: Post Voting Period
Viewed: 8,544 times Debate No: 17393
Debate Rounds (4)
Comments (21)
Votes (1)

 

Thrasymachus

Pro

(First attempt at the website. Prior apologies for blunders and breaches of nettiquette)

I think the fine tuning argument - in whatever form - is a failure.

The "fine tuning argument" is any argument that deploys premises related to the particular initial conditions/constants/laws of the universe as evidence that God exists. I particularly have in mind Robin Collin's argument in the Blackwell companion to natural theology, but similar arguments by Leslie, Craig, etc. also count.

An argument is a failure iff, after considering the argument, it would not be reasonable to raise one's credence in the conclusion of that argument. In other words, the argument does nothing to make what it is arguing for more probable or plausible.

Either a) Use your first reply to accept, then watch me make a fool of myself as I go on the philosophical equivalent of a drive-by, or b) perhaps a better idea, present the fine-tuning argument your find most persuasive, as a springboard for arguing about whether any such argument can be persuasive (if you do b, please leave your last reply blank, as I am too much of a luddite to work out how to jiggle the round system to give equal rounds with the challenger 'going first').

Any references can be included in the comments and not in the body of the main post. If you want to challenge and want any other conduct issues cleared up, feel free to use comments or email. Ditto if I have made further mistakes.

Enjoy life,
Thrasymachus

P.S. Thanks to prior commentators for correcting me earlier.
InquireTruth

Con

I accept my opponent's conditions.

As a courtesy, however, I will present a basic representation of my argument from fine-tuning. My opponent's burden, then, will be to show how my argument does not make the existence of God at least slightly more probable than whatever the probability may be without it.

I will be arguing that the fine-tuning of the universe is significantly more probable on theism than on an atheistic single universe hypothesis. And, if necessary, I will be arguing that the Multiverse hypothesis fails the rule of parsimony inasmuch as it posits an infinite amount of necessarily unobserved entities. That is to say, it multiplies entities beyond necessity.

My basic, employed probability calculus on this issue, given in Bayesian terms, is as follows:

PR (FT/T) >> PR (FT/ASU)

Where "FT" represents fine-tuning and "T" represents theism, this calculus asserts that, given theism, the probability of "FT" is greater than when "ASU" (atheist single universe hypothesis) is given. Of course, this probability calculus presupposes that the universe is, in fact, finely tuned. I will not argue this point unless my opponent chooses to deny its existence. I will, however, clarify what it is I mean by fine-tuning.


What is Fine-Tuning?

When I say fine-tuning, I do not just mean the universe is ordered in such a complex way, delicately balanced upon a razor's edge, as to allow for the existence of intelligent life - thought it is, of course. Rather, I will argue that the universe and all its constants and conditions, under the principle of indifference, could have been different. That is to say, there are innumerable alternative and equally as probable constants and conditions that could have occurred during or after the formation processes of the universe starting from and during the Big Bang. There are, on the other hand, very few outcomes that would permit a sustainable universe where anything at all is operable. The very existence of our universe requires finely tuned constants and conditions. Given solely unguided and natural processes, however, the probability of an operable universe, like ours, is very low. However, the probability of an ordered, intricate and operable universe given theism, or at bare minimum, intelligence, is very high.


Does This Argument Fail?

Per my opponent's definition, this argument fails if, and only if, "the argument does nothing to make what it is arguing for more probable or plausible."

This seems quite a difficult thing to do. It certainly seems clear, especially considering the prime principle of confirmation, that observed complexity is more likely on "T" than on "ASU." Perhaps what my opponent meant by failed was that the argument does nothing IF it is false. But that is obvious and goes without saying. So instead, my opponent must show that my argument is necessarily a non sequitur, inasmuch as the probability of theism is not increased even if all my premises are true. We will have to wait and see the approach my opponent takes.

So far as I can see, observed and ordered complexity are more probable on theism than on an atheistic single universe hypothesis. For instance, if I walked in the woods and saw a fully assembled log cabin, I would not insist that the trees happened to fall in such a way as to construct the cabin – I would rightly posit a designer – even if he be unobserved. Given the principle of indifference, as stated above, however, each outcome, including the fully assembled log cabin, was equally as probable under "ASU." In order for somebody to explain the log cabin without positing a designer, they would have to posit numerous assumptions and unobserved phenomena. On "T", though, we have very good reasons for believing that the universe would exist in an ordered, habitable and operable fashion. Which is more parsimonious? Which, my friends, is more likely?

I hope this will give my opponent an understanding of my position so that he may build his own accordingly.

InquireTruth


Debate Round No. 1
Thrasymachus

Pro

I thank CON for his construction of the argument, which closely follows Robin Collins (see Collins 2009). I wish to slightly modify the argument, for the sake of charity and clarity. I invite CON to correct me if I have unwittingly straw-manned his argument.

CON writes, "[T]he universe is ordered in such a complex way, delicately balanced upon a razor's edge, as to allow for the existence of intelligent life…". I would like to make this explicit: Theism does not predict 'Fine Tuning' as such, but rather predicts a universe fertile for intelligent life (or embodied moral agents). Call this a "Life Permitting Universe" (LPU).{1}

I presume CON is using epistemic possibility to drive the probability inference. So, when CON writes "[I] will argue that the universe and all its constants and conditions, under the principle of indifference, could have been different." I take him to be saying that it is epistemically possible for the universe to have different values for its constants and conditions (and, given indifference, they should be weighed equally). This appears the most defensible way of cashing out these possibilities - if CON had a better idea in mind, he is welcome to offer it.

We now have:

P(LPU|T & k') >> P(LPU|ASU & k')

LPU: Life permitting universe
T: Theism
ASU: Atheistic single universe{2}
k': our background knowledge, subtracting (inter alia) knowledge of our existence or this universe existing. {3}

Con is surely right to say the universe is 'delicately balanced' for Life Permitting-ness: across the epistemic space of possible values for these constants and conditions could take, the life permitting range appears miniscule. On Atheism (given indifference), the likelihood of landing in this miniscule range is as remote as any other miniscule range in the probability space (nothing turns on just how small this is - call it "one in a zillion"). Theism, by contrast, predicts a life permitting universe, as embodied moral agents or similar are part of God's design plan.{4} So, P(LPU|T&k') >> P(LPU|ASU&k'), and that a life permitting universe obtained provides stonking good confirmation of Theism.

How can this argument be a failure?


Epistemic access and modal explosion

Best science tells us that mild deviations of the constants and conditions would result in a universe that is not life permitting. It is silent about massive deviations - if, for example, a universe with some of the constants and conditions shifted by a few million orders of magnitude would be life permitting. We cannot reliably extrapolate our physical understanding out that far. The space of possible worlds with wildly divergent values is bigger than the space of possible worlds with values close to our own.

There are further limits to our epistemic access. Consider these possibilities:

Alien Physics (friendly): There is a set of physical laws entirely dependent on four variables, A, B, C, D. A-D can range over the positive reals. For all combinations of A-D, the resulting universe is life permitting.

Alien Physics (unfriendly): There is a set of physical laws entirely dependant on a million variables, all of which can range over the positive reals. Only the instance where each of these million variables equals one is the resulting universe life permitting.

Both alien physics, friendly or unfriendly, strike me as epistemically possible (it strikes me as epistemically possible, albeit remotely, that this universe could be one or the other - our best physics might be mistaken). One can conceive of more and more alien physics, to varying degrees of life friendliness, ad infinitum.

From this, it is clear that the possibility space is exploded far beyond our means of epistemic access: we only have reliable insight into the locality around our universe, not those with wildly different constants or laws of physics. In this far larger region of the probability space, the incidence of LPUs is inscrutable.

For the fine tuning argument to work, we need to assign P(LPU|ASU&k') as low - that life permitting universes are rare across all epistemic possibilities of how the universe 'could have been different'. Yet the fact that LPUs are rare in the tiny scope of possibilities we can examine provide no warrant to suppose that they are rare across the entire probability space, the vast bulk of which we do not have access to examine. If so, then the correct assignment for P(LPU|ASU&k') is that it is inscrutable. Therefore, whether it is less than P(LPU|T&k') is also inscrutable. Therefore, the fine tuning argument provides no confirmation for Theism over Atheism. Consequently, the fine tuning argument is a failure.{5}


Conclusion

CON may accuse me of having a modally promiscuous imagination: that alien physics or wild constants are impossible. It is hard to see why this would be so: there seems no warrant for thinking the space of epistemically possible universes comprises primarily of those with laws of physics like our own, and values for the constants not too far from our own. The problem is the possibilities I leverage to show the fine tuning argument is a failure are no more remote than those CON relies upon to make the argument in the first place.

I have closely addressed CONs original presentation, at the expense of a more general overview of fine tuning arguments. Might it be the case that this objection applies only to this particular form of the fine tuning argument, but the argument succeeds in other guises? This is unlikely - this presentation is the clearest iteration of the fine tuning argument (Craig's, for example, is far more clumsy, see e.g. (Craig, 2007 pp. 78), and the objection - that we lack access to say life permitting universes are rare on Atheism - will be applicable no matter how we cash the argument out. If any fine tuning argument can survive this objection, then this fine tuning argument can survive this objection. But it cannot.

Thrasymachus



{1} Also, "fine tuning" is a bit vague, and risks confusing the argument I think my opponent is making with subtly different ones (for example, (Roberts, 2011))

{2} I will grant for the sake of argument that multiverses are not a serious possibility.

{3} This is included for little more than completeness's sake, although subsequent developments will make it relevant. See (Collins, 2005) for a programme of how this 'subtraction' may occur.

{4} It does not strike me that P(LPU|T) should be very high: it is not clear to me why God would want worlds of embodied moral agents rather than aphysical spirits, for example. Regardless of my worries, P(LPU|T) is not that low - certainly greater than one in a zillion to derive confirmation from the fine tuning argument. I note this now only because it may be relevant later.

{5} I reject CON's assertion that "[M]y opponent must show that … the probability of theism is not increased even if all my premises are true". What I originally said was "An argument is a failure iff, after considering the argument, it would not be reasonable to raise one's credence in the conclusion of that argument. In other words, the argument does nothing to make what it is arguing for more probable or plausible." CON misquotes me by omitting the part in italics. Yet even if that was a fair redaction, CON's exegesis is crazy: formal validity is insufficient to "make what it is arguing for more probable or plausible". Here's an example of a formally valid fine tuning argument that fails to make its conclusion more plausible:

If there is fine tuning of the universe, then the moon is made of cheese
If the moon is made of cheese, then God exists
There is fine tuning of the universe
//God exists


So I don't need to show CONs argument to be formally invalid - I can settle for showing key premises to be false or inscrutable. If my argument that we have no access to make the probability assignments CON requires prevails, then the fine tuning argument is a failure (on this, cf. (van Inwagen, 2006. Ch. 3)).

InquireTruth

Con

I would like to thank my opponent for an argument that, to my knowledge, has not hitherto been showcased to the tenebrous eyes of debate.org and the shrewd members therein. With brogdignagian exactitude, my opponent has let no jargon slip from his fingers as he presented his case for affirming a very strict skepticism. Meaning and substance, however, can often be missed when nomenclatures replace common vernacular, so I may need to reiterate some points for our less familiarized readers.

So the new probability calculus that will be the center of our focus is:

P(LPU|T & k') >> P(LPU|ASU & k')

I think my opponent has done a great job of not caricaturing this position. Looking only at what we know (k'), it seems clear that the probability (P) of a life permitting universe (LPU) given theism (T) is greater than a life permitting universe given an atheistic single universe hypothesis (ASU).

When we look at it from this light, my opponent agrees, that such a formulation is "stonking [sic] good confirmation of Theism."

But why look at it from this light? While my argument relies upon what we know, my opponent begins his case by considering what we do not know, or, as stated, what we do not have epistemic access to. This makes his use of "epistemic possibility" [1] relevant, because it is only useful when dealing with situations in which we are ignorant.

The basic premise of his argument is that, since the region of the probability field that we have epistemic access to is much smaller in comparison to the region of the probability field that we do not have epistemic access to, all probability is inscrutable. That it is to say, we are far too presumptuous in thinking that fine-tuning confirms theism given everything that we do not know. Does this seem to be a powerful criticism?

It fails for at least two reasons.


When Arguments Prove Too Much

My opponent's argument suffers from being too effective! What a bizarre criticism, right? How can an argument be too effective? First, his argument is not specific, insofar as there is nothing inherent in the argument that limits its application just to considerations of our universe and its probability. Basically, the probability region of anything we may seek to prove must consist of a larger area of known, quantifiable scenarios than of unknown, epistemologically inaccessible scenarios.

Let us consider the existence of the external world. It is epistemically possible that I am merely a brain in a vat and all my feelings, thoughts and actions in and of an external world are illusory. Or, it is epistemically possible that none of my cognitive faculties actually yield true belief, subsequently leading me to believe that there is such thing as an external world. Or, there could exist a Cartesian demon whose sole purpose is to delude me. These scenarios could go on ad infinitum, and, because they can neither be known true or false by virtue of being epistemically inaccessible, they remain distinct epistemic possibilities. So even if the external world did exist, we haven't any good reasons for believing that it exists in the way that we perceive it. Because, according to my opponent, the likelihood of the external world existing or existing in a certain way is inscrutable - inasmuch as mostly all of the mass sum of the probability space is not within epistemic access and thus the occurrence of a scenario in which the external world actually exists in the way we actually perceive it is inscrutable.

So it is not so much that my opponent's argument proves that fine-tuning arguments (FTA) fail, when, in actuality, it proves that all knowledge fails (save, I imagine, a single proposition).

So it seems the crucial point is whether or not it is reasonable to limit our considerations to the epistemic field with which we have reliable insight. In other words, is it reasonable to consider only that which exists within our epistemically illuminated field? It seems only reasonable that we would only consider what we have epistemic access to when considering likelihood or else the likelihood of anything is inscrutable given the infinite space of epistemically possible yet necessarily unknown, epistemically inaccessible information.


Epistemically Illuminated Range

Instead, we should limit our comparison range to our epistemically illuminated range (EI) [2]. It seems practically common sense that we should only consider what is possible to consider when dealing with any sort of study, scientific or otherwise. But Rob Collins offers a helpful analogy. Collins likens our epistemic range to a dart board. We know, for instance, that the dart has fallen within our illuminated region because we are capable of observing it and making sense of it. Further, if we see that the dart has hit the bull's eye and the bull's eye is enormously small in comparison to our illuminated region, it would be strong evidence that the dart was aimed, even though we have no way of measuring or knowing the density of areas that exist out of our illuminated range.

The hypothesis that the dart was aimed is clearly determined by the measured ratio of the Bull's eye in comparison to the ratio of our observed illuminated range. What we know is that no bull's eyes exist within our epistemically illuminated range, save the one. And, our illuminated range is by orders of magnitude larger than the observed bull's eye. Thus, we know that P(BI/A & k') >> P(BI/C & k'), where BI is the dart having hit the bull's eye, A stands for the aiming hypothesis and C stands for chance.

My opponent wishes that we not restrict ourselves to the illuminated range but also consider all of the unilluminated portion. If this were the case, it is true that we could not get any sort of confirmation hypothesis to work, as the inference would be impossible to make given a possibly infinite space of unknown region. But, as I stated above, this actually makes any confirmation hypothesis impossible. However, even if the unilluminated region was rife with bull's eyes, it still seems rather improbable that the dart would hit the bull's eye that exists in an area where the width of the enitre observable region includes only that one bull's eye. So though we can give no exact confirmation formula because we don't have epistemic access to the unilluminated region, we can still assume the worst and see that the aiming hypothesis is preferred.


What is k'?

k' stands for our background knowledge. What we do not know is necessarily not part of our background knowledge as it is, by definition, not knowledge. So we need not consider the epistemically unilluminated region when considering P(LPU|T & k') >> P(LPU|ASU & k'). Our epistemically illuminated region, on the other hand, is part of our background knowledge and thus the above calculus avoids my opponent's criticism from the onset. Since k' includes both our epistemically illuminated range (EI) and fine-tuning (FT), and necessarily precludes the unilluminated portion, the problem stated by my opponent is illusory.


Sources:

(1) http://en.wikipedia.org...

(2) Collins, R (2009) The teleological argument: an exploration of he fine-tuning of the universe. In Craig W. L., and Morland, J. P. (eds) The Blackwell Companion to Natural theology. Blackwell publishing; Oxford


Debate Round No. 2
Thrasymachus

Pro

The story so far

If the probability of getting a life permitting universe given atheism is inscrutable, the fine tuning argument fails. To support my claim of inscrutability, I have shown that our epistemically illuminated range is far smaller than the probability space, and so the frequency of life permitting universes in our sample is no guide for the frequency across this all-but-unknown space.

CON offers a smorgasbord of counter-principles, analogies, and reductios as to why this is mistaken. All, sadly, are products of misunderstanding. Here, I'll show these misunderstandings, salvage the arguments CON uses from his mistakes - and then show these arguments fail anyway.


k' is not background knowledge

CON writes, "k' stands for our background knowledge." I disagree ("k' = our background knowledge, subtracting knowledge of our existence or this universe existing.") So does Collins: "[W]e cannot simply take k′ to be our entire background information k, since k includes the fact that we exist, and hence entails LPU." (Collins, 2009, pp.241) To my knowledge, so does all the literature.

If we include all our background information, P(LPU|k) = 1 by the weak anthropic principle - so P(LPU|k&ASU) or P(LPU|k&T) will be 1 too. The is the old evidence problem. The solution Collins borrows (and I accept) is we need to subtract out the background information of our observation of an LPU, giving us k' (cf. {3} above). CON is just wrong here, and his later assertions ("Since k' includes both our epistemically illuminated range (EI) and fine-tuning (FT)...", "What we do not know is necessarily not part of our background knowledge as it is, by definition, not knowledge.") are, to the degree I decipher them, hopelessly confused.


We can use our normal background knowledge normally

CON attempts a reductio to general scepticism by pointing to the a vast field of epistemic possibilities where a sceptical hypothesis is true (brain-in-a-vat, Cartesian demon, etc). Given I accept the frequency of these worlds is inscrutable, doesn't my argument oblige me to say the truth of any sceptical hypotheses is inscrutable?

Firstly I can use k, not k' - I can keep in background knowledge about existing in this universe. Secondly, I need not apply indifference: I can deem the infinitely possible worlds where a sceptical hypothesis is true as very remote possibilities, instead of equiprobable with the actual world.{6} The fine-tuning argument needs both subtraction and indifference to work, so cannot escape the sceptical trap as I did. My sceptical concerns are selectively toxic to the fine tuning argument, and so this reductio falls flat.


No good reason for EIR restriction

The centrepiece of CON's case is to justify restricting ourselves to the epistemically illuminated range (EIR) with the fine-tuning argument.

CON writes, "It seems practically common sense that we should only consider what is possible to consider when dealing with any sort of study, scientific or otherwise." Three counter-examples:

Statistical significance: "My sample suggests X, but my sample is too small to reliably represent the population at large. These results are of no great significance"

Skeptical Theism: "There are many evils that appear gratuitous. However, we should not believe they really are: there are likely goods beyond our ken we cannot consider that justify these evils."

Argument from sceptical hypothesis: "It is impossible to know whether or not you are a brain in a vat. If you are a brain in a vat, this table does not really exist. Therefore, it is impossible to know whether or not this table really exists."

If CON was right, we could dismiss these out of hand. He is not: the sensible thing to do when one realizes the truth of a proposition hinges on things one is unable to consider is to deem it inscrutable. In the case where our EIR is much smaller than the entire range, P(LPU|ASU&k') is almost wholly dependant on what we cannot evaluate - the frequency of life permitting universes outside the EIR. We should therefore deem P(LPU|ASU&k') inscrutable.

CON then offers the partly illuminated dartboard from Collins (cf. Leslie, 1989). CON presents this in a manner reliant on his prior reductio and assertion that it is 'practically common sense' to restrict ourselves to the EIR. I have defeated these above. For extra brownie points, here's why the dartboard analogy fails to help CON's case.

CON is right to say that hitting a bullseye confirms a bullseye aiming hypothesis, even in the worst case scenario of 'polka dot' bullseyes outside the epistemic range. Bullseyes are rare regardless of the state of the rest of the unilluminated region, so P(BI|C&k') is low.

If Theism predicted a 'bullseye' LPU, the fine tuning argument could side-step my sceptical concern in the same way. But the fine tuning argument does not predict a bullseye LPU, but rather just an LPU. The appropriate dartboard analogy is not 'aiming for bullseye' but 'red selection'. Unlike aiming, our likelihood of picking red by chance is sensitive to how common redness is outside our illuminated range, and the sceptical argument runs as before.{7}

Could we change the fine tuning argument to use a 'bullseye LPU' instead? The probability of landing in a bullseye LPU given Atheism is low, regardless of the limits of the EIR. Unfortunately, such a move switches the inscrutability to the other side: I'm willing to say that an LPU given Theism is not too surprising, but a bullseye LPU on Theism is inscrutable. So this variant of the argument fails too.


EIR and probabilistic tension

Having shown all elements of CONs case for EIR restriction fail, I now show why EIR restriction should be rejected.

Collins talks elsewhere about probabilistic tension - one should avoiding holding conjunctions of beliefs where one is unlikely given the others (see (Collins, 2009. pp.209-210)) Collins deploys this to block the Atheist move of "okay, P(LPU|ASU&k') is low, but P(LPU|ASU&e&k') is not low, where e is some anthrophilic extension, and I believe ASU&e". If P(e|ASU&k') is low, then this is under probabilistic tension, and should be rejected for collections of beliefs under less tension (e.g. ASU&k', or T&k'). This is a probabilistic gloss for why ad hoc-ness is bad.

Call "our universe is in the EIR" Q. Is k'&Q under probabilistic tension? Yes: by the assumptions of the fine tuning argument (indifference, subtraction, etc.), P(Q|k') is just the area of the EIR over the total range of the possibility space. The EIR is a minute bullseye in this probability space. As this fraction is miniscule, P(Q|k') is miniscule, and thus the addition of Q to our background suffers gigantic probabilistic tension. It therefore should not be added to our background information.

This makes intuitive sense. To consider fine tuning, we need to subtract away all knowledge of this universe existing, and simply consider the bare facts of the possibility space of all the ways a universe could be. Yet, once we have subtracted knowledge of this universe existing, we have no reason to presume the universe must be within epistemic range of this one.


Conclusion

CON has failed to defeat the sceptical concerns I present. These concerns show P(LPU|ASU&k') is inscrutable. Therefore the fine-tuning argument is a failure.


{6} I may be mistaken to include the falsity of sceptical hypotheses in my background knowledge, or to deem possible worlds where sceptical hypotheses are true as remote. But neither of these conflict with the sceptical concerns raised against the fine tuning argument.

{7} Collins uses the dartboard analogy as evidence for including EIR restriction in background. Doing so, he avers, is the best likelihood reconstruction for the analogy. The red selection counter shows my likelihood reconstruction (not needing EIR in background) is better than his (which does need EIR in background). These analogies do not make a case for EIR restriction.
InquireTruth

Con


Am I confused about k’?


My opponent insists that I am fundamentally confused about k’ because I said it entails all that we know. Now, of course, this background knowledge would have to exclude the fact of our universe existing or, as my opponent points out, the probability will always = 1. I took this as assumed, as all probabilities about prior events must subtract the outcome from their background knowledge or probability is impossible. So what should have been said was this: K’ equals everything that we know, save the knowledge of the existence of our universe.

At this point, all that is necessary for my opponent’s argument to fail is for there to be a good reason to include EIR (our epistemically illuminated region) in k’ when dealing with P(LPU|T & K).


Where an argument proves too much.


My opponent attempts to avoid the problem of the effectiveness of his argument by using our “background knowledge normally.” Does this avoid my criticism? I do not think it does.

If we omit the knowledge of our existing in a world external to ourselves (EW), then P(EW|k’) is inscrutable. This means that extreme skepticism is our only reasonable outlet. But we can avoid this all together by not omitting the knowledge of EW from k’, giving us k (normal background knowledge). But how is this not special pleading? What are the prevailing, relatively different features that make the possibility space in the aforesaid scenario substantially different then the possibility space regarding our universe? Why is it that we can treat the possibility space regarding EW as remote possibilities and not equiprobable? There are no known differences between the epistemically unilluminated possibility space of EW and LPU, making my opponent’s distinction special pleading.

So if my opponent can take k when dealing with EW, then he must be committed to taking k in all scenarios that have no relative differences (or at least none that are known). So in order for my opponent to avoid his own criticism, he must not permit the use of k’ all together.

Let’s look at some similar circumstances that require both indifference and subtraction and are similarly defeated by my opponent’s argument. Say I roll a 20-sided die and land on 12. I could say that the probability of my rolling 12 is 1/20, but that would be wrong unless I subtract my rolling 12 from my background knowledge (or else the probably = 1). Moreover, the only way I can say it was 1/20 is if I apply the principle of indifference, insofar as all other possibilities are equiprobable. But now we’re in a tight spot, because the only way to get 1/20 is to also limit our epistemic considerations only to that which we know, namely all known possible rolls. Or else we would be committed to giving equiprobable weight to all of the possibility space, mostly all of which we have no epistemic access to. This renders the simple equation matter of 1/20 impossible to achieve per my opponent’s criticism. For all we know, there could be something like a Cartesian evil genius that makes all rolls hit 12 more often than any other number. Or it could be that the roll may have fell within a crevice and stopped between two numbers. The only way to avoid this problem of inscrutable prior probabilities is to include EIR in k’.

In fact, there are no instances that employ both subtraction and the principle of indifference that avoid my opponent’s criticism. All efforts to calculate prior probability in such scenarios are impossible unless we include our epistemically illuminated range within our background knowledge.


Good Reason for EIR restriction


The above is actually powerful reason for why it is necessary to consider only our EIR when dealing with prior probability that includes subtraction and indifference.

Let me briefly address Pro’s examples of where limiting our EIR is dangerous:

Statistical Significance: This is a false comparison. A sample group that is too small has failed to consider that which was possible to consider, namely a much larger group. A more relevant comparison would be where a statistical study examines everything that falls within their EIR but is determined to be insufficient for not considering all that was impossible for them to consider per it existing outside of their EIR. Statistical comparisons MUST consider only that which is within their EIR or else they could never do statistics.

Skeptical Theism: This is relevantly different inasmuch as it relies upon propositions that are definitionally inaccessible. Meaning that the whole discussion takes place outside of our EIR.

Argument from skeptical hypothesis: This hypothesis itself shows why we have good reasons for restricting our EIR. If we do not, then everything is necessarily inscrutable given this latter point. The reality of this debate is inscrutable given the fact that we are unable to consider whether or not we are a brain in a vat. But if we include, however, our EIR in k, then we need not believe that literally everything, save a single proposition, is inscrutable. This point fails because it insists upon a strict Cartesian epistemology.

So the only way for my opponent to be right is to admit that his being right is inscrutable, inasmuch as every single proposition, save one, hinges upon things one is unable to consider. This sort of skepticism is rife with self-defeating clauses.

So the proposition that our EIR is much smaller than the entire probability range is itself inscrutable. It is possible that our EIR is the entire range. It becomes incoherent when we expand our range to propositions that exist outside of our EIR. It seems clear that we need to include EIR in k’.


Dart Boards and Red Selection


You’ll remember that my dart board analogy illustrated how we use the EIR and the irrelevancy of the unilluminated region, insofar as the bulls eye is very small in comparison to our illuminated range. But what of PROS use of the analogy of “red selection”? How does this red selection analogy differ? The red selection analogy is dependent upon the commonness of red within the probability space – which we do not know (if ignore our EIR). But what we do know, to some chagrin, is that within our EIR only one red selection is available, giving evidence that non-red selection is more common than red – inasmuch as one would expect to have more red selections that fell within our EIR if such selections were at all common.

Even worse, however, is the fact the analogy of red selection ALWAYS fails when applied to prior probabilities, meaning my opponent has simply poisoned the well. If 10 red selections are put in a hat with 10 other blue selections, and I pull a red one, I could not correctly insist that my selection had a probability of ½ without subtracting my knowledge of selecting red and assuming indifference. If I do not limit my EIR, as my opponent suggests, then the probability of my pulling red is not ½, but inscrutable – given that the unknown possibility space is larger than the known. So we should only assume that my opponent’s analogy is greater or more applicable if assume from the onset that we should not limit our EIR (to beg the question).


EIR and probabilistic tension


If we insert the fact that all probabilities that require subtraction and indifference also require inserting EIR into our background knowledge, then there is no good reason to exclude EIR from k’ when dealing with P(LPU|T & k’). This makes probabilistic tension unsuccessful because EIR is knowledge not excluded by our subtraction of the knowledge of the universe existing, making it a necessary property of k’ and not a conjoined, ad-hoc belief.


Debate Round No. 3
Thrasymachus

Pro

The story so far

I have argued that the fine tuning argument is a failure, as we lack epistemic access to say whether life permitting universes really are unlikely on Atheism: the fine tuning data only speaks of a vanishingly small sample of ways the universe conceivably could have 'turned out different', and that life permitting universes are rare in this sample provides no reason it should be rare in general. I cashed this all in Bayes.

CON then deployed a variety of complaints against this reasoning, and I replied that his complaints were mistaken, CON has come back to say that his complaints still stand. Here, I'll show why he's wrong.

Reductio

The bayesian gloss is that P(LPU|ASU&k') is inscrutable, because we have no way of estimating the frequency of life permitting universes across all conceivable possibilities (such as alien physics, zillion-fold variation in all the constants, etc.) CON argues that this leads to external world scepticism: isn't P(EW|k') inscrutable too? If CON means the probability of getting an external world on the background information subtract an observation of a life permitting universe, he is likely right. But that is not what we should be interested in: we are not interested if our background subtract the observation of a life permitting universe can confirm an external world, but rather if an external world is confirmed on our background knowledge subtract our convictions there is indeed an external world.

Providing there is a good argument against general scepticism, then P(EW|k-{EW}) will be pretty high. Theres a big literature on providing such arguments. So long as that case works (and CONs later case for EIR etc. doesn't) then no special pleading is going on when I assert P(EW|k-{EW}) to be high yet P(LPU|k'&ASU) to be low. So, again, there ain't no reductio here.

CON goes on to say that, unless we include the EIR into background, all sorts of things become inscrutable. This is a mistaken understanding. So long as we have good reason to reject external world scepticism, we can use most of our background to 'winnow down' the space of possible worlds we need to consider when doing probability. So the possible worlds where the 20 sided dice comes up 1-20 dominate the possible world space, because wacky alternatives (landing 24, say) are unlikely given what we know about being in this universe and so on - these are remote possible worlds. But, as I noted at the start of the programme, subtract away the fact we exist, and all bets are off.{8}

EIR:

CON attempts to show that, regardless of how good our epistemic illumination is, we should only use possibilities illuminated by it when doing inference. Our examples perhaps show why this is misguided.

Statistical significance: CON writes, "A sample group that is too small has failed to consider that which was possible to consider, namely a much larger group. A more relevant comparison would be where a statistical study examines everything that falls within their EIR but is determined to be insufficient for not considering all that was impossible for them to consider per it existing outside of their EIR. Statistical comparisons MUST consider only that which is within their EIR or else they could never do statistics."

These remarks from CON suggest that if our data-collectors 'do their best' and collect all possible data, we could perform reasoning on this even if, in fact, our sample doesn't give us access to the larger group. This is wrong - if your sample isn't good enough, your sample isn't good enough. If you can't even possibly consider enough to get access to the larger group, then your larger group is not visible. That is rather the point of statistical significance.

Skeptical Theism: I honestly have no idea what CON is saying here. However, the type of inaccessibility should not matter - so if goods beyond our ken are indeed beyond our ken, they lie outside our EIR so (according to CON) should be discarded, as we can only consider what it is possible to consider.

Argument from sceptical hypothesis: Again, CON simply misunderstands how aSH works - it can't be 'waved away' by something about how it relies on Cartesianism. No one writing on it thinks you can 'answer aSH' by just pointing to the fact the epistemic possibilities are beyond easy access. That's the entire point of the argument!

Dart boards and Red selection

CON misunderstands the analogy, so I'll point him to my previous round rather than making further confusion myself.

Conclusion

CON is write that if the principles of subtraction, indifference, and 'EIR' are somehow fundamental to statistics, then my objections do not work. But applying EIR restriction amounts to refusing to consider the limits of our knowledge. Alas, this (and other heterodoxies from CON) are mistaken.

Best wishes,

Thrasymac

{8} CONs love of the principle of indifference is bizarre. Because often we shouldn't be indifferent across known possibilities: consider a dice we know to be loaded, or a bag of balls with unequal numbers of white and black balls, etc. Indifference may be our initial presumption before any evidence changes our mind, but it is weird to say it is some fundamental tenet of probability theory to be used in all inference.
InquireTruth

Con

Much thanks goes out to Thrasymachus, for an always interesting topic choice and a very fun and critical analysis of it. In wrapping up, I give the story of this debate as I see it.

This is what I hear him saying:


Example 1

1. k' is background knowledge minus our knowledge of the existence of our universe.

2. Once we subtract the knowledge of our universe, we need the principle of indifference in order to assign prior probabilities to our range of possible outcomes.

3. The principle of indifference ascribes equal probability to all possible outcomes.

4. The amount of unknown, possible outcomes is far greater than known, possible outcomes, therefore the probability of LPU is inscrutable.

Then I illustrated, since there are no inherent attributes in the former logic that limit its application to universes, that such a criticism renders many things unnecessarily inscrutable.


Example 2

1. K* is our background knowledge minus our knowledge that 12 was rolled using a 20-sided die.

2. Once we subtract the knowledge of the rolls outcome, we need the principle of indifference in order to assign prior probabilities to our range of possible outcomes.

3. The principle of indifference ascribes equal probability to all possible outcomes.

4. The amount of unknown, possible outcomes is far greater than known, possible outcomes, therefore the probability of my having rolled 12 is inscrutable.


Examples digested:

Both examples above employ the same exact reasoning and the only way for the conclusion of example 2 to equal the obvious (1/20) is to consider only our epistemically illuminated range (EIR). In fact, this is exactly what PRO does:

"So long as we have good reason to reject external world scepticism, we can use most of our background to 'winnow down' the space of possible worlds we need to consider when doing probability. So the possible worlds where the 20 sided dice comes up 1-20 dominate the possible world space, because wacky alternatives (landing 24, say) are unlikely given what we know about being in this universe and so on - these are remote possible worlds."

The arguments in the literature against general skepticism (to which we are referred to a large body of uncited literature) are simply ones that require lowering the bar of what it takes to achieve knowledge, where certainty is not a factor, merely a firm enough justification is necessary (with no set and measurable standard). This weaker form of knowledge (against traditional Cartesian strands), actually operates entirely within our EIR and, quite decidedly, not outside it. This weaker form of knowledge is helpful, because it allows us to have knowledge of more than just the single proposition of our existing.

This means that his wacky alternative worlds where our physics are different and/or the requirements of life are changed have as much warrant as the alternative, epistemically possible worlds where dice fall between two numbers or are controlled by spiritual forces.

This can mean only one thing. He is subtracting more from our background knowledge when dealing with P(LPU|T&k') then he would from any other formula that requires the same steps.


Subtracting Too Much From K'

As you'll remember, k' stands for our background knowledge, that is, everything that we know, minus our knowledge of the universe existing. This is a rather big issue in that it involves the fundamental question that my opponent and I have been butting heads over: should k' include our EIR? You'll also remember that EIR stands for our epistemically illuminated region which is, in essence, the region that we have intellectual access to and can achieve knowledge within. There are some things that may well exist, but, by virtue of falling outside of our EIR, we will never know for sure. There are many reasons I believe that EIR should be included in k', not least of which is because it is always required of all prior probabilities that utilize subtraction and the principle of indifference.

I think I have finally put my finger on why PRO and I have been talking at somewhat of an impasse. It has become clear that he is subtracting too much from K'.

When we consider the probability of a life permitting universe it is often helpful to consider possible worlds. For instance, if conditions for life were x,y and z, we can consider the general likelihood of a universe within this possible world arriving by chance by looking at other possible outcomes. PRO completely eliminates the conditions for life and our understanding of physics from k' and then entreats us to consider all other epistemically possible scenarios created by this omission with equaprobable weight. This is bizarre, as we are equally as ignorant as to whether or not our minds yield true or verisimilitudinous perceptions of reality, rendering our Example 2 inscrutable -- why is it that we do not treat this epistemic possibility as equaprobable?

If we only consider what is within our EIR then alien physics and alternative life permissions are as much irrelevent to our probablity calculations as brain-in-a-vat type hypotheses are to everything else. So this makes probability very possible. Since there are only a finite amount of possible and epistemically known universes that support life, we can make very easy probability calculations. This is because the requirements of life and of physics are not things subtracted from k' as they are not conjoined to our knowledge of our life-permitting universe.


What are we left with?

If we treat k' then, as it ought to be treated, we can know certainly that in order for a life-permitting universe to exist, it must not, for instance, collapse upon itself immediately after forming (strength of gravity). So any one of our universes' parameters is fine-tuned for life relative to a set of possible worlds iff the life-permitting values for those parameters are rare in relation to all of the values they could have taken in the those possible worlds. While all possible worlds are not known, the finite life-permitting values are, allowing us to "winnow out" all possible worlds, whether unknown or known, that are incompatible with our known life-permitting values. So it does not much matter that we cannot epistemically fathom what a universe would look like with a nuclear strong force a thousand times different than ours, as we know it does not fall within our finite set of life-permitting values.

The only way to avoid this is to subtract our knowledge of the requirements of life and/or physics, which is disingenuous (especially since I explicitly stated that fine-tuning was included in k' and in fact, needs to be for the probability to even make sense in round 2).

This clarity makes his analogies irrelevent because they would only be comparable if the requirements of life was oustide of our EIR, and a required subtraction from k, which it is not.

Conclusion:

Again, I would like to thank my opponent for this debate and would look forward to discussing this issue further with him, as I feel like we have only just begun understanding one another.


InquireTruth

Debate Round No. 4
21 comments have been posted on this debate. Showing 1 through 10 records.
Posted by InquireTruth 5 years ago
InquireTruth
"I don't know why people can't use English..."

Are you referring to the Bayes formula?
Posted by Cerebral_Narcissist 5 years ago
Cerebral_Narcissist
I don't know why people can't use English...
Posted by modivarch 5 years ago
modivarch
"sesquipedalianism"

Ha! Learned a new word today - I have a new goal of using that at least 5 times before the end of the day.
Posted by joze14rock 5 years ago
joze14rock
Good debate!!!
I actually wrote a paper against Collin's "Fine-Tuning" Argument. I argued that his argument is ill-founded on a theistic prejudice (He is a Christian Apologist, if you didn't know).
I instead showed that Spinoza's argument of necessity works better... but that means accepting a God of imminent transcendence....
haha, well now I'm just going on a tangent!
Fun reading!
Posted by Switzerland 5 years ago
Switzerland
You shouldn't debate god's existence
Posted by Thrasymachus 5 years ago
Thrasymachus
Hello all,

Thank you for the debate. My apologies that my last round was more hurried that my previous rounds - medical work sprung on me, etc. Hopefully, if I am right in that CON is generally misunderstanding stuff, than my previous rounds will carry me through. If I'm wrong, and I'm generally misunderstanding stuff, then I'm stuffed either way.

Anyway, thanks to InquireTruth, and look forward to seeing what he (or anyone else) have to say.

Best,

Thrasymachus
Posted by angelie_pintor 5 years ago
angelie_pintor
why would you ever consider for a fact that the existence of god is a "failure" ? so you think all those "lucky" moments in your lives are just "coincidences" ?? I think not..
Posted by KristophKP 5 years ago
KristophKP
Guilty of many counts of sesquipedalianism.
Posted by Kinesis 5 years ago
Kinesis
This is excellent subsidiary reading to Collin's chapter in the BCtNT. I found it tough going the first time.
Posted by Thrasymachus 5 years ago
Thrasymachus
R3 References:

Collins, R (2009) The teleological argument: an exploration of the fine-tuning of the universe. In Craig W. L., and Morland, J. P. (eds) The Blackwell Companion to Natural theology. Blackwell publishing; Oxford

Leslie, J (1989) Universes. Routledge; New York.
1 votes has been placed for this debate.
Vote Placed by CD-Host 5 years ago
CD-Host
ThrasymachusInquireTruthTied
Agreed with before the debate:--Vote Checkmark0 points
Agreed with after the debate:Vote Checkmark--0 points
Who had better conduct:--Vote Checkmark1 point
Had better spelling and grammar:--Vote Checkmark1 point
Made more convincing arguments:Vote Checkmark--3 points
Used the most reliable sources:--Vote Checkmark2 points
Total points awarded:30 
Reasons for voting decision: I can't believe no one bothered to judge this. I see this as a fundamentally easy debate to judge. Pro's main contention was that there was no evidence the constants needed to be in a narrow range and Con made no serious effort to refute. Neither side used sources.