Tuesday, April 9, 2013

Review : Richard Swinburne's Existence of God

Often times, it is easy to criticize the religious for not being fully abreast of the knowledge on a subject that they so easily reject.  Pretty much all creationists who reject evolution reject some caricature of it.  When those who are more knowledgeable about a subject try to explain these sorts of things to the ultra-dense monuments of ineffable stupidity like Ray Comfort, they will simply have no clue what you're talking about, and take their ignorance as the correct attitude...  because apparently, ignorance is a good thing in their world.

Still, there are always times when the apologist tries to lead you down a path of looking up various sources in favor of their position.  Now, most of the time, it's someone I'm actually already familiar with like Lee Strobel, Dinesh D'Souza, Ravi Zacharias, or William Lane Craig.  When someone points me to a source I haven't actually read, though, it would be hypocritical not to actually follow through while simultaneously demanding the same of others.  And if there's a halfway decent argument (which there typically is not), or at the very least, if it takes some effort to find the flaws, then I'll at least look for it.

So I was suggested to read one of the works of Richard Swinburne which I actually hadn't read before -- The Existence of God (2010).  At least insofar as Swinburne's antics in the public eye, I've not been impressed, but supposedly he applies more rigor when he's writing.  Fine, here's my review.

Nothing special.

Oh, it tries to show some element of rigor very early on.  For a good while, Swinburne appears to be fairly strict in defining all his terms.  It all tends to fall apart the moment he tries to take this over to his actual argument.  When initially told about it being a probabilistic argument, I expected something along the lines of Dembski's works, which always had the fundamental flaw of dealing in unconditional probabilities, and ignoring the fact that background knowledge does change the assessment of probabilities.  This indeed was the case with Swinburne's work, but he at least, is not trying to go at it from the tack that Dembski is taking where he was trying to attack the position of his opposition.  Swinburne, instead tries to posit a particular god hypothesis and try to push for its likelihood.

Swinburne's definition of "God" (his hypothesis h) is defined as --
a ) God is a nonembodied person (a spirit) that has always existed and always will exist, and is "omnipresent."
b ) God is the source of being and power behind all substances.
c ) God is perfectly free in the sense that nothing other than his own will can causally influences his choices.
d ) God is "omnipotent"—able to do whatever is logically possible.
e ) God is "omniscient"—he knows at any time whatever it is logically possible to know at that time.
f ) The existence of God is an inscrutable "brute fact" that simply does not have an explanation.
g ) God has the qualities of a "person" insofar as having intention, beliefs, personality, etc.

Already, he has started to define a god that is a little bit different from what most any religion has.  Specific gods like Yahweh, Krishna, Ahura Mazda, Zeus, Allah, etc. all have many more properties than that.  So right here, he's being fairly generic, but he's also putting some restrictions on God that probably don't exist in other religions either.  For instance, his definition of omniscience does not leave room for absolute precognition.  It leaves room for the deity's expansive knowledge of the current state to make a limited future prediction likely to be accurate (in the same sense that a military commander can predict the actions of his adversary, while not requiring divine clairvoyance to do so), but he can only loosely predict, but not "know" ahead of time, the actions of free-willed agents.  You might also notice that Swinburne hasn't added any aspect of good or evil to the defining characteristics.  That is because the moment you add that, you have effectively made it impossible for there to be a high probability for h, irrespective of anything.  Instead, Swinburne makes an argument later on after establishing his basic definition based on the assumption of "objective moral values" that he claims makes it a perfectly reasonable addition among the attributes of the god of h.  He's basically sneaking in the "goodness" without saying it from the outset.  So much for rigor.

Swinburne, in the very beginning chapter accepts, as well he should, that all deductive arguments for the existence of deities fail because too many of their premises and assumptions are open to debate.  He mainly states that using uncontroversial premises is essential to forming a good C-inductive or P-inductive argument.  At that point, I'd say he'd have a book that I'd not really disagree with.  But in spite of that, he absolutely needs to have substance dualism as part of his argument in order for there to be such a thing as a "disembodied person."  He's clearly not so naive as to think that's not a concept open to debate.  Also, his argument for attributing "perfect goodness" to his god is fundamentally underpinned by the concept of "objective moral truths."  This is hugely contested proposition.  Swinburne, of course, realizes this, and his best retort is to Godwin his way out of it.  Specifically, he brings up the example of whether or not there was anything wrong with Hitler slaughtering millions of Jews, and that anyone who says it wasn't is saying something which is ostensibly wrong.  Problem is that he leaves it at that, as if to imply the fact that the repulsion that we all commonly feel to the idea is objective in nature rather than merely being a subjective feeling that happens to be more or less universal across humanity.  Does he really think that an appeal to emotion like that would settle an issue that has been debated for centuries?

But even if I grant his proposition of moral realism, he still has a problem in how he attributes "goodness" to his god.  He argues that a free agent unfettered by irrational influences will necessarily carry out acts in whatever way he feels to be "good."  The problem here is that not all "good" is created equal.  Should we instead presume that doing something "good" that is "good" in the sense of gratification or self-preservation is an example of irrational influence?  Swinburne never really makes that clear.  What if god is a sociopath?  Sociopathy can definitively be shown as orthogonal to matters of rationality, but it is certainly not orthogonal to the moral dimension.  And since he insists that his h does not necessitate goodness in order to avoid a probabilistic counterargument against it, he has to make this equivocation on "good" in order to make the implication, lest he be left with a morally empty deistic god.  After all, his allegiance to the god of Christianity is something he doesn't really hide.

A big part of his probabilistic argument is to apply an old corollary of Occam's Razor, and in doing so, argue that God's existence is more likely by extension.  Swinburne repeatedly dismisses competing hypotheses on the grounds that they are more apparently complex, and therefore less likely.  Simplicity, however, is a pretty subjective thing, and that's why in general, this rule is applied very loosely.  Swinburne, for instance, argues that infinites are simpler than finites because they don't have any need for definition.  His wording is that finites "demand an explanation for their limit [to a specific quantity.]"  He seems to ignore the flipside of it that not having any such definition (as infinity is an undefined by nature) means you have an intrinsic risk of apparent self-contradictions and paradoxes which are technically not at all paradoxical when you look deeper.  I don't think any mathematician would ever tell you that the infinite is simple.  Even disregarding this folly, there can be valid reasons for apparent complexity simply because the underlying determined facts related to a given hypothesis are inherently complex.  A scientific hypothesis (say, h'), in a vacuum may seem needlessly complex, but when given the volume of observations (o) it attempts to explain and the background knowledge (k) in the field of study, h' might actually be the simplest possible explanation.  i.e. Even though P(h') should be low by Swinburne's gut-check metric, P(h'|(o & k)) would actually be very high for the same reasons.  This is really where Swinburne's primary thesis fails utterly.  He's actually not alone in this, since every probabilistic argument for theism fails in this way.  Properly taking into account the sheer volume of information that is contained within (o & k), is something that is not easy to do.  The dishonest bit is of course that all theists conveniently disregard, ignore, or feign ignorance of knowledge which weighs heavily on a subject simply because it's not in their own best interests to take it into account.  Swinburne not only does this, but he also goes further and exalts intuition, proclaiming that it is unreasonable to dismiss it so easily.

Throughout Swinburne's piecemeal statements for arguing why his god hypothesis h should have any significant probability, he simply fails to address the vast volume of relevant observations and background knowledge.  Indeed, I would question his capacity to even identify whether certain information is relevant at all.  Throughout the book, he never really provides solid reasons why he rejects anything, nor does he provide very rigorous reasoning why he thinks certain probabilities should be low or high.  Instead, it's peppered with "maybe", "perhaps", "on an intuitive level", "it seems to me...", "I prefer to [take the position that]...", "I am inclined to suggest...".  How any of these really fall in as putting merit into his argument is never validated by any means that steps outside the boundaries of personal feelings and intuitions.  It only seems an extension from his earlier valuation of personal feeling and an overall disinterest in the very concept of criteria by which to establish validity for the valuations.  He puts on a facade of having such rigor when laying out his Bayesian formulations, but it seems to end there, as he offers no strictness or even full adherence to his own professed strictures when he gets around to actually offering raw numbers.  It's all intuition, feeling, and a mess of examples that really don't take a big picture view of anything.  He always arrives at the position that traditional vague arguments make good c-inductive arguments for the existence of god, and the traditional refutations thereof don't make good c-inductive arguments against the existence of god...  but the only reason the latter might even appear to be the case is simply because he has defined his hypothesis h to intrinsically lack for attributes such as goodness and detectability.  That's fine for the arguments against, but it really gives him nothing more than wishful thinking and appeals to emotion and "common sense" on the arguments for.  But according to him, that's enough.

For instance, in Chapter 6, he goes into the ratio P(o | h(x) & k) / P(o | k), for a given hypothesis h(x), and background knowledge k, and observation(s) o.  He calls this the explanatory power of the given hypothesis.  The first flaw here is that it's really not that straightforward to find P(o | k).  The second is that the numerator is something that can easily be fudged.  This is the sort of problem we have to face when dealing with a god hypothesis.  If we use Swinburne's definition, then its total lack of specificity means every observation o, which is logically possible has a P(o | h) of 1.0, regardless of any k.  Now a scientific hypothesis h', on the other hand, tends to operate in much more specificity, such that P(o | h') may be quite low, but P(o | h' & k) can be 1.0.  Furthermore, many of the arguments that bring to light scientific theories, eliminate the necessity of a god, making any argument that relies facts explained by these theories vacuous.  You can't get anywhere further by adding on a magical being that can't be verified even in principle when it is provably unnecessary.

One can't really say that Swinburne is totally ignorant of it all.  In fact, throughout the book, he definitely demonstrates a relatively decent degree of scientific literacy.  He is also sensible enough not to argue that his hypothetical god created everything directly, but "used PZed's organic model" a la Mr. Deity.  But despite that, he still collects really old and really intellectually vapid arguments for god and interprets them as C-inductive arguments and thereby accumulates a net probability that is greater than 50-50...  but how he gets there, and how he gets to ascribing the effect of each argument, and how he disregards competing viewpoints is where it gets really fuzzy and prone to flaw.  In particular is his position that personal religious experiences are not to be disregarded.  Well, a big part of that is that we know such experiences to be unreliable.  They're extremely prone to self-deception or outright lying, and this is explicitly verified.  Swinburne contends this is simply not the case because they're just too ubiquitous.  Indeed, religion itself is ubiquitous, so religious Even if they were reliable, he seems to completely ignore why they're not usually considered in the first place -- because they're not consistent!  If we could actually reliably get something out of them that was truly universal, then you might have something worth exploring.  It's a problem because a Buddhist in central Thailand will have completely different experiences from a Muslim in Kandahar.  It's a problem because a Hindu in Chhattisgarh will have a completely different experience from another Hindu in neighboring Andhra Pradesh.  A Pentecostal in Pennsylvania will have a different experience from a Methodist in Misssissippi.  How exactly are you going to pin down what experiences are significant if you can't even get a consistent result?  Is there one that points specifically to Swinburne's h?

I think the biggest problem of course with Swinburne's argument, all said, lies with how he uses his particular hypothesis, h.  We can look at all the various god hypotheses like Yahweh, Allah, Krishna, Ahura Mazda, Zeus, Toutatis, etc.  None of these really fall exactly in line with Swinburne's h.  Although he doesn't really tout that he has any argument for any of these other gods, he doesn't deny the intention of favoring his preferred god.  At a number of points, particularly in chapter 6, he makes the mistake of defining ~h as the strong atheist position (i.e. explicitly stating that there is no god).  To be accurate, ~h actually includes all the god hypotheses that do not fit Swinburne's specific h.  In spite of the fact that he is technically only arguing for his own h, he seems to drift away from the fact that his own h is not really the same as just any old god hypothesis despite its inherent vagueness.  In truth, I actually doubt that Swinburne is so absent-minded as to honestly miss this.  I'm inclined to think it's a deliberate confusion tactic in order to try and allow room to piggyback the presumed merits he applies to his own god hypothesis onto others and vice versa.  In order to try and attach h to his own preferred god, he interjects late in the game that h "may" offer a happy afterlife of indeterminate length in a heaven-like realm of unspecified location...  and that h could "perhaps share the suffering" of his creations, potentially by taking a physical incarnation on Earth.  How very wide and grand the potential possibilities of this nondescript god!  And therein lies the weakness.  All of these things are additional baggage that would make h too complex to be probable by his own rule, but he ultimately has to go there in order to make the proposition in any sense meaningful.

P(d), where d is the proposition that he is being willfully deceptive seems low on the surface...  but for b is this book and At is the fact that he's a Christian apologist, P(d | b & At) = 1.0.