Thursday, January 08, 2009

Stephenson's Anathem uses pulp fiction to explain free will

As I get 2/3 into Neal Stephenson's Anathem, I realize the central question it addresses (in a pulp fiction writing style) is consciousness and free will.

And I was reminded of a conversation I had back in college debating whether free will exists, and I was finding it hard to believe that a smart guy like the guy I was debating would disagree with my view that free will can't exist.

We had taken as a given not to consider quantum effects since they're crazy and mess up all intuition. Then in a classical world it seems obvious to me that free will is an illusion, since if there is a state vector that represents the universe, then the laws of physics will always determine definitively the next state for all eternity. Our choices are all predetermined by the state of the particles that make up our brain and its environment.

So the idea that free will is an illusion is disturbing and at the end of the night (literally as the sun was coming up) we had agreed that free will may be an illusion, but people (and any artificial intelligence we designed) must be designed to act as if free will is real.

I felt vindicated years later, listening to my Teaching Company course on philosophy where this Berkeley Prof basically said the current consensus view in philosophy is that free will is indeed an illusion because a brain is a physical system.

Years later, I partially resolved this problem by saying that maybe the future is pre-ordained, but complex (in the formal definition), in that not solvable in less than exponential time, and that it is probably provable that no machine could compute the future faster than the future could actually arrive.

Of course, underlying all this is this nagging idea that this argument depends on classical mechanics, and with quantum mechanics, these arguments fall apart.

So what Stephenson does (and I wonder if this is his own thinking, or he stole this) is reconciles these two ideas (quantum mechanics and complexity) in order to bring consciousness and free will back. Essentially, if i can summarize his 1000 page argument succinctly, is that indeed the future in unknowable because predicting the future is hard, he asserts it is essentially NP-hard. But we know that quantum computers can solve NP problems in polynomial time. And thus our brains are actually quantum computers, and we know this because since we can indeed predict the future. Moreover, given that the future is indeterminate until the quantum states collapse, our consciousness can even influence which states quantum distributions around us collapse to, altering our environment. In fact, this is the definition of consciousness and free will, because otherwise, you have to accept that free will is an illusion.

Anyway, I'm not entirely convinced, but fantastically neat ideas.

But it is especially cool because it is relatively easy reading as it is written in the style of Jurassic Park or Ender's Game or Harry Potter but develops ridiculously sophisticated ideas. So he'll stop the action for chapter long Socratic dialogues, but they're so compellingly written, like a duel, complete with formal rules like the wizard wand duels in Harry Potter, but battling all with ideas.


Anonymous said...

Very nicely written; I stumbled upon this jewel as well. The book is very good.

Ben Ho said...

if you haven't yet, his other books are all great. though the baroque cycle is a bit to get thru. i have reviews for most of them somewhere on my blog.