| Who created God? |
[Nov. 5th, 2011|03:08 pm]
Scott
|
There's a faction in the atheist blogosphere trying to get atheists to stop using the "yeah, well, if God created the Universe, then who created God?" argument. They say that it's easily defeated, philosophically naive, and brings disrepute upon atheists to be caught spouting such an obvious fallacy. I quote Ivory Tower Metaphysics:
"Since many skeptics are scientifically based in their knowledge, a few have made a career out of making oversimplified or outright bad philosophical arguments. This post will cover one of the worst: the "who designed the designer" objection to the existence of God."
From Common Sense Atheism:
"I want to kill this sacred cow of atheism... because I am not loyal to atheism per se, but to truth and reason, because I want atheists to stop giving arguments and objections that are so easily rebutted...[and] because I want atheists to focus on objections that really matter."
With all due respect to people who are probably smarter than I am, I have yet to be convinced that "Who created God?" is not a valid argument against theism, or at least an alley that can lead to valid arguments. I'm going to sketch out how I would make such an argument here and see if the atheist blogosphere points out something flawed about it.
The Pyramid on Mars
Suppose that the newest space probe on Mars discovers a strange pyramid-shaped structure on the planet's surface, one that looks like it might be artificial. We ask two friends how they think such a singular object came to exist on an uninhabited planet.
Dr. Science says "There may be some unknown geological or meteorological forces that promote the formation of pyramid-shaped structures on Mars, in the same way that geology promotes the formation of bizarre artificlal-looking formations like the Giant's Causeway on Earth.
Crazy McLoonbat says "The pyramid was built by Pyramobot 5000, a giant sapphire-blue pyramid-building robot with a thousand flexible titanium arms."
Most people would agree that Dr. Science's explanation is more likely than Crazy McLoonbat's. And a big part of why we reject Crazy McLoonbat is wanting to ask "But who created the Pyramobot 5000?"
If we can figure out why the objection "But who created the Pyramobot 5000?" is valid, we can see whether the same objection might apply to "But who created God?"
Minimum Message Length
Minimum message length is a sort of triple-bladed spring-loaded version of Occam's Razor which states that the best explanation is the one that requires the shortest message to describe all observed data, and which I will describe here only very informally.
Suppose you make a record of the weather that looks like this: "On Monday, it rained. I got wet. On Tuesday, it rained. I got wet. On Wednesday, it was sunny. I stayed dry. On Thursday, it was sunny. I stayed dry. On Friday, it rained, I got wet."
You can shorten this by writing "On Monday, it rained. On Tuesday, it rained. On Wednesday, it was sunny. On Thursday, it was sunny. On Friday, it rained. When it rained I got wet; when it was sunny, I stayed dry."
Okay, fine. That's not much shorter. But if I had an entire year's worth of weather up there, it would be. And if I were writing it in the formal mathematics that you're supposed to use for this sort of thing, it definitely would be. So my "theory" that I get wet when it rains and stay dry when it's sunny has shortened the amount of text I need to describe my observations.
Science is an attempt to do this upon the scale of the entire world. Instead of saying "Here is an earthquake. It seems to have happened on a tectonic boundary. Here is another earthquake. It also seems to have happened on a tectonic boundary" science makes a theory: "Earthquakes generally occur on tectonic boundaries" and all of a sudden we can predict the location of earthquakes much more accurately.
This brings us to the other important criterion for explanations: they should have predictive power. Once we know earthquakes occur on tectonic boundaries, then even without any records of seismic activity in California we can predict it's probably going to have a lot of earthquakes.
The Holy Grail of science is to have a single, very short message that can describe the entire universe. There's a saying that the Theory of Everything ought to be able to fit on a single blackboard. If this were true, then it would be essentially a blackboard-sized program for generating the universe from a certain starting condition (like a Big Bang). We could stick it into a (very powerful) computer, wait a few billion years, and the computer would have generated the entire universe, with all its planets and stars and rocks and iPods and restaurants in precisely the right places.
...but since at the moment we can't operate on that level, instead we try to make our unexplained entities as few and as simple as possible. We're not really sure exactly what quarks are, but we can describe them mathematically with a couple of short equations, so they're pretty simple. On the other hand, it's hard as heck to describe human beings and their thoughts, actions, beliefs, and desires. So it's a good thing we think we can reduce them to quarks. And although we can't describe, say, the results of the US midterm elections in terms of quarks (yet!) we try alternate routes to simplify them, where we describe them in terms of things like economic growth and presidential approval rating. It would take at least a book to describe this in a rigorous philosophical way, but hopefully you can at least see the picture I'm trying to paint here.
Mysteries and Explanations
We can define a "mystery" as something that, if taken at face value, would fly in the face of all this elegance: something that seems to require completely different laws than everything else, laws that would require a massive amount of extra message length to describe. So if we start off with a world that runs on scientific rules, and then we see a magician pull a rabbit out of a hat, this is a mystery: if taken at face value, it implies that all our previous good work coming up with general laws describing the universe like "matter cannot be created or destroyed" are false. We would have to expand our message from [a few simple laws that might one day fit on a chalkboard] all the way to [a few simple laws that might one day fit on a chalkboard, plus many extra terms and forces describing the ways that rabbits can materialize out of hats for no reason.]
And we can define an "explanation" as something that reduces a "mystery" down to something more congruent with our current shortest message. For example, if we learned that the magician's hat contained a simple compartment, then we'd be back to business as usual: our set of physical laws could describe people making hats with compartments in them.
But this means an explanation has to be less mysterious than the thing it explains. If the magician tells you he gained the power to materialize rabbits by selling his soul to Lapid'xor, the Rabbit Demon of the Third Inferno, this leaves you even more confused than the original magic trick. If there are rabbit demons and no fewer than three infernos, why haven't you heard about them before, and where does that leave the calm rational law-abiding world you thought you understood this morning?
Given everything you already believe, it requires a smaller addition to say "and there's a secret compartment in that hat" than it takes to say "and there are demons, and souls, and magic powers, and at least three different infernos". Although these two sentences have similar lengths in English, in the MML formulation we have to actually explain everything we put in. Clearly it would take at least an entire book to explain what was going on with these rabbit demons and infernos, whereas the secret compartment is a very short and innocuous addition to your world view.
Return to Mars
So now we can get back to the Pyramidobot 5000. A pyramid on Mars is a mystery: our current worldview predicts that there should be no artificial structures on a lifeless planet. Our description of the universe seems to include a few short rules about the origin of life, and then the extra fact of a pyramid on Mars which completely contradicts them.
When Dr. Science explains the pyramid by invoking weather and geology, she's using things that are already part of our worldview: the very fact that I could invoke the Giant's Causeway earlier means we already believe that artificial-looking structures can be produced by natural processes. Since weather and geology are already in our world-view, we don't take any extra penalty for mentioning them here as long as we can credibly explain how they lead to a pyramid. Thus, the explanation shortens our minimum-length description of the universe, as it should.
When Crazy McLoonbat brings up the Pyramidobot 5000, he explains an unwelcome addition to our minimum message length with something that requires an even greater addition to our minimum message length. As much trouble as we were having adjusting our worldview to contain a pyramid, it will be far worse adjusting it to contain a giant Martian robot. So this seems to also cry out for an explanation, for a theory that allows us to include it in our world-view without increasing our minimum message length too much.
Dr. Science's explanation reduces the pyramid to something simpler than itself. Crazy McLoonbat's explanation reduces the pyramid to something more complex than itself.
...so who created God, anyway?
Another interesting mystery is that the universe exists. This seems to defy a lot of theories like "Matter can't be created or destroyed" and "There has to be a reason for things."
If we can explain the existence of the universe by the means of a single homogenous featureless explosion that occurred 15 billion years ago, well, that still leaves some mystery (what caused this explosion?) but it's a heck of a lot simpler than individually specifying all of our planets and stars and rocks and iPods and restaurants. And it holds out that enticing possibility of one day ending up with a Theory of Everything that lets us write the entire message length of the Universe on a single blackboard.
But if we explain the existence of the universe by means of God, then we have the same problem as we got when we explained the existence of the Martian pyramid by means of Pyramobot 5000. The early universe was pretty much just a cloud of hydrogen broken only by the occasional atom of helium. It was really, really, simple. It took billions of years and a whole lot of energy pumping against the thermodynamic gradient before all this simple hydrogen produced anything as complex as jellyfish, let alone intelligent life.
So when we bring in an intelligent being to explain the early universe, we're explaining the simple by means of the complex. We're taking something super-simple except for the one surprising fact that it exists at all, and explaining it by means of the something that looks a heckuva lot like the most complicated known structure in the universe - the human brain.
This is exactly the same problem as explaining the mysterious pyramid on Mars with the even more mysterious Pyramobot 5000, and should be rejected for the same reasons.
The Creation Question
There is a theist response to this argument, but it isn't pretty. The theist says "Well, actually, God is infinitely simple." This has been around since at least the time of Thomas Aquinas, who it is safe to say probably didn't know very much about information theory.
But the Christian doctrine of divine simplicity doesn't use "simplicity" in the same way as Occam's Razor, Minimum Message Length, or any other modern theory of complexity. Divine simplicity supposedly means that "the being of God is identical to the attributes of God". To modern materialist philosophy, this is meaningless and absurd: attributes aren't things that float around and can be part of someone's "being", whatever that is, they're emergent properties of something's structure. Something without structure is also without attributes. I realize this isn't much of an attack on the concept; I can only plead that this is less because the concept is too strong to attack, and more because the concept is such a non sequitur that it can not even be argued against. If I asked you to prove that colorless green ideas do not sleep furiously, you wouldn't know where or how to begin.
Luckily, I think that some of the more modern Christian philosophers identify God's simplicity in a way that at least touches upon information theoretic simplicity, which means it's meaningful enough to debate. In this case, though, they're simply wrong. I'll give a few reasons why.
Is God a jealous God, or a non-jealous God? Well, according to the Bible, we know He's a jealous God. Is He good or evil? According to theology, we know He is good. Does He have a tendency to create universes, or would He rather sit along in self-contemplation? Judging by the universe around us, He prefers to create universes.
But if there are three yes-no questions about which God falls on one side or the other, that means it must take at least three bits to specify God.
In reality, it probably takes several trillion bits to specify God. Go ahead and try it experimentally - start programming God on your computer, and don't stop until you have a program that, when run, will behave exactly as God does (this may be a bit much work for one person, so feel free to join a pre-existing project). When you're finished, post in the comments how many bits it took.
Programming God would be at least as hard as programming an AI, since God would have to display greater than human level intelligence. But AI is notoriously complex: even simple visual recognition algorithms and speech processing algorithms take hundreds of megabytes. The human brain, which implements visual processing and speech recognition and much else besides, needs about a hundred trillion synapses. And although recognizing the letter "A" may seem easy to you, that's only because a hundred trillion synapses in your brain have auto-arranged themselves into an incredibly complex processor without your conscious direction. Even a computer algorithm capable of distinguishing capital A from lower case a would probably be more complex than the entire early universe. If each synapse contains 1000 bits of information (I made this number up) and God is a thousand times more intelligent than humans (I researched this number carefully), then it would take a hundred quintillion bits to specify God. As explanations go, positing God is moving from "description fits on a chalkboard" to "description would not fit on the entire surface of the earth, even if you cleared off all obstacles, drained the oceans, and wrote in tiny microscopic letters".
Finally, consider an intuition pump. Suppose supporters of Bigfoot abandoned their belief in the great ape as silly, at the same time as they proposed a similar creature, Simplefoot. Simplefoot is quite like Bigfoot in every way, only he is infinitely simple.
This would be silly. You can't just add "...and it's infinitely simple" to the end of a pre-existing description. Either something is infinitely simple or it isn't. Bigfoot has a bunch of characteristics -- ape-ness, hairyness, large shoe size -- and adding "oh, those attributes are all part of his being" or "yeah, but he's still infinitely simple" doesn't mean that all of a sudden they "don't count".
So Who Created God, Anyway?
So now let's return to the original question: if an atheist hears "God created the universe", is it reasonable for her to ask "Then what created God?"
The atheist blogosphere says no. Their argument is the "why regress". According to Common Sense Atheism:
"Let us ask ourselves what would happen if we required that a successful explanation must itself be explained.
This would lead immediately to an infinite regress of explanations. We would need to have an explanation of the explanation, and an explanation of the explanation of the explanation, and an explanation of the explanation of the explanation of the explanation… on into infinity. And thus, we would never be able to explain anything."
That is, suppose you ask a scientist "Why are there total eclipses?" and the scientist answers "Because the moon covers the sun." You can then ask "Yes, but why does the moon cover the sun?" and the scientist will answer "That's how its orbit works." And so on to "Why does its orbit work that way?" and then "But why is there gravity?" and then "But why does space time bend?" and then the scientist will be unable to answer. But that doesn't mean that "Because the moon covers the sun" isn't a great and valuable explanation for "Why are there total eclipses?"
To which the answer is "Yes, you can always have a why regress for anything. But in a virtuous why regress, at every stage, your information becomes more coherent and your minimum message length becomes shorter. In a vicious why regress, at every stage your information becomes less coherent and your minimum message length becomes longer."
We accept "the moon covers the sun" because we already know there's a moon, that it's in about the same place as the sun (the sky) and that it's about the right size and shape to cover it. In fact, if the moon never covered the sun, we might be surprised and ask why it didn't. So this requires little or no addition of information to our description of the universe, and it eliminates the "mystery" of solar eclipses.
But if the scientist said "The great floating leviathan Hrogmorph covers the sun as he traverses the sky in his quest to eat the Moon Squid," we would have to add floating leviathans and moon squid to our message length, which might be even longer than just biting the bullet and adding solar eclipses (although if you're an ancient culture that thinks in mythological terms because all you ever think about are people and animals, it makes perfect sense).
So when a theist says "You atheists can't explain what created the universe" then the exact proper atheist response is "Yes, but what created God?". It is a way of saying "Yeah, there are a few bits of information tacked on to my theory in an inelegant way lengthening my message. But there are a few hundred terabytes of information tacked on to yours. So there."
(and one of the reasons I like Tegmark's mathematical universe hypothesis is that it holds out the seemingly impossible hope of removing even that one blackboard and ending the why regress once and for all)
The writers of both of the blogs I've referenced are very smart, and I know Luke especially knows all this already, but maybe they think that unless an atheist is willing to go into the whole explanation of minimum message length, they should avoid entering this whole area because they might lose. I disagree; that would be like saying there are some really smart theists who can come up with reasons why enzyme biochemistry "disproves" evolution, so unless you're an expert in biochemistry, you should never say creationism is false.
Sure, not every atheist is going to know everything about complexity theory. But I hope most of the smarter atheists will be able to realize that an epistemology which argues that Pyramobot 5000 is just as good an explanation for a Martian pyramid as wind and weather has got to have something wrong with it.
Added 6:00 - a much simper defense would be that the theist, in getting to this point, has probably hit up against the why regress problem herself. "Why does the universe exist? Okay, but then why did the Big Bang happen? Okay, but then what created the previous universe whose collapse caused the Big Bang? Oh, you can't answer that? Then there must be a God." If the "who created God" question forces the theist to explicitly prohibit infinite why-regresses, then she is forced to undermine her own argument. |
|
|
