Wednesday, May 23, 2007
Civilization and the prisoner's dilemma
Last weekend I was reading Douglas Hofstadter's Metamagical Themas on the subject of the prisoner's dilemma. The prisoner's dilemma is a toy model of the central question of civilization: on what logical basis do I refrain from screwing the other guy? Or rather, since social structures are putty in our hands, what structures do we build so that friendly cooperation helps us both? Obviously we want to avoid exploiting every possible advantage, since we both lose if we pound each other flat. Likewise we want to avoid unthinking deference, since the first Viking to come a-viking with pound both of us flat.
In the book, Hofstadter talks about using computer simulations to find a good social structure. The researchers came up with a variety of rules for social interaction. Some of the rules were dead simple ("always be nice"), while others were complicated predictive algorithms. To start with, each rule had the same popularity factor. Each rule was pitted against all the others, and its success was judged by how well it did on average multiplied by its popularity factor. If a rule was successful, its popularity quotient was increased, and vice versa.
What happened was both boring and remarkable. The simple rules like "always turn the other cheek" promptly disappeared, as common sense would predict. Then the others fought it out for a long time, gradually converging on rules that were nice but willing to be vengeful. The remarkable bit is that one simple rule always did well, often being the best: Tit for Tat. If the other guy hurt you, hurt him back this time and then be nice next time. Swift vengeance followed by swift forgiveness.
That crystallized my thoughts about civilization and stability. There are people who want everybody to be nice all the time. All sunshine and butterflies, never a storm or wasp. They say that an-eye-for-an-eye leads to everybody being blind. They rarely articulate their logic clearly, a careless and dangerous habit, but it goes something like this: Nearly everybody is nice nearly all the time. It wouldn't take much for everybody to be perfectly nice all the time. Therefore we should make Nice our highest law. All the cow people should be roped into the corral of sweetness and light.
BULL. Their vague handwavy logic ignores a critical matter. The current situation in the First World—most people nice most of the time—is not a static equilibrium. We did not wake up one morning in the Valley of Milk and Honey, surrounded by mountains so steep we couldn't climb out if we wanted, trapped in eternal prosperity and happiness.
But this I will say to you: your quest stands upon the edge of a knife. Stray but a little and it will fail, to the ruin of all.
— Galadriel, The Lord of the Rings
The free world is a dynamic equilibrium. It is a broom carried upon the tip of its handle, or a rocket balanced on a pillar of fire. We climb a ridge into the heights, and if our legs buckle, our grandchildren might not live to see above the clouds. Look at the path behind us. It's ten parts daring leap, and nine parts slide and fall. So many cultures have reached material and spiritual prosperity, only to crumble and fall into darkness, sometimes for a generation, sometimes for a thousand years, and some terrible times forever.
We had all the momentum; we were riding the crest of a high and beautiful wave. So now, less than five years later, you can go up on a steep hill in Las Vegas and look West, and with the right kind of eyes you can almost see the high-water mark—the place where the wave finally broke and rolled back.
— Hunter S. Thompson
So I thought to myself "Self, I ought to write a Bill Whittle-style essay about how the prisoner's dilemma applies to industrial society. It would rock."
Oh, and as for an-eye-for-an-eye, what it leads to is a couple of idiots with no eyes, and the rest of the people helping each other get sand and stray eyelashses out of each others eyes.
Friday, May 18, 2007
Dark matter 2
A continuation of Dark matter: what do we know.
So dark matter is mysterious. How do we learn more?
Good question. The trouble is it doesn't appear to interact much with laboratory equipment. If it came in small, dense chunks, laboratory evidence would have turned up by now. We've done quite a few experiments with enough sensitivity to detect tonne-scale chunks—ultracold calorimeters and acoustic resonators, pendulums, optical interferometers, and so forth. The excellent stability of atomic clocks rules out higher-order quantum effects caused by dark matter.
We likewise see very little evidence of it in solar space. The planets and assorted rubble appear to move to ancient Newton's laws, with an ocassional tiny correction for that upstart Einstein. Deep space radars don't show spacecraft bobbing around in the wake of tonne+ masses. (Although maybe nobody has looked closely enough—I'll ask my personal space radar expert.) The Pioneer 10 and Pioneer 11 probes have subtle anomalies in their deceleration curves, but I expect that is just a chemical or mechanical effect.
Even within our galaxy, dark matter's effects have not been clearly observed. We are getting lots of good data from distant galaxies. However I doubt this will illuminate the quantum physics of dark matter, assuming it has any. We just don't have the resolving power to see fine enough details, and are not likely to get it soon.
So where do we look? I favor looking for dynamical friction near our galaxy's gigantic central black hole (Sagittarius A*). It has both immense gravity and close-orbiting stars. The stars orbit so fast that we could just about afford a telescope to watch them move from week to week. If dark matter comes in chunks less than a light-year wide, that's where we will see it.
Nearby globular clusters are also a good place to look for dynamical effects. If anything smaller than a galaxy is big enough to have a dark matter density gradient, they will. (Globular clusters can reach hundreds of light years in size.) Depending on how dark matter works, they might even have captured some of it. At the same time they are close and bright enough that we can distinguish individual stars, which lets us do detailed measurements of the cluster's gravitational structure.
I can't come up with any plausible laboratory experiments for dark matter. High-energy particles do not seem to have any interaction, or the effect would show up in particle accelerators. It would probably also eliminate the high-energy cosmic rays that we observe, which helps rule that out. If big stars capture dark matter, then it might affect the neutrino brightness curve of a nearby supernova. We could measure that with a terrestrial instrument, but nearby supernovae are once-in-a-lifetime events.