Thursday, June 24, 2010

Pawns and Processors, Bytes and Bishops

In a game famous for it's prodigies, Magnus Carlsen of Norway stands out as something special. He was the third youngest chess grandmaster ever (age 13). Youngest #1 player ever (age 19). At 19 he also holds the third highest ELO score ever, putting him with the top chess players of all time (although wikipedia informs me that score is subject to inflation and thus debatable). Apparently you don't usually hit peak chess ability until 25 at the earliest, so being the best in the world at 19 promises for a long, dominating career.

Why do I bring him up? Because his childhood spent mastering chess included lots of time playing against computers. Carlsen is obviously one of those rare individuals that excel at some task far beyond the abilities of others. But could screen time have contributed to his almost unique skill? It's an answer we'll have to wait and watch for as new generations of chess players enter the world stage. Carlsen could be a fluke, or a precursor of what's to come.

The argument for software being a major cause of Carlsen goes like this: natural ability is not enough to be the best in the world. It's a requirement, but insufficient. Chess greats throughout history have put countless hours into studying chess: memorizing moves, playing games. This is true of any skill: you need to paint lots of pictures, write lots of words, hit lots of fastballs to make it big. Natural aptitude is a clay that gets formed into a specialized machine with repetitions and practice.

But not all practice is equal. Joe Mauer, 3 time A.L. batting champ, perfected his swing against a contraption his dad invented that would drop a ball at his eye level. With so little time between seeing the ball and the ball passing his waist, Joe had to develop an exceptionally quick swing. More generally, how often do strong high school sports programs exist for decades at a time? Partially elite performing schools attract transfers, but partially the talent in a program encourages everyone to perform at a greater level.

As a child, I played chess against my mother. For a long time she'd beat me every game. Eventually I surpassed her, winning most (but not all) of our matches. And then I stopped improving. It wasn't until years later that I started playing chess on Yahoo that I noticed myself improving again. I had found a challenge.

If you practice a competitive skill against players below your level, it's much harder to improve. Failures teach us. If you only see 50 mph fastballs, how do you learn to hit a 95 mph one? You're not developing the reaction time, the muscle memory, or the poise to do so. So it is with a chess prodigy. Eventually you surpass your parents, local children, then local adults. You can travel the world playing tournaments against the current generation of greats, but there's only so many such games you can get in. Often the chess prodigies will find a skilled player to mentor them, but the truly great will surpass even them.

Enter the computer, capable of calculating dozens of moves in advance. The computer can now play on par or better than any human in the world. In the coming years, that advantage will only grow. Any child with a computer now has access both to talented players across the world via the internet, and against their own, personal AI chess master. There is no need to arrange games against other greats, traveling far to play: from the comfort of your room you can pit yourself against the best.

As computers soar in their processing power, defeating all comers, they threaten to trivialize chess. But simultaneously they offer us the opportunity to reach previously unattainable levels of skill. And chess could just be the start. From math to literature, sports to debate, engineering to leadership computers may soon give us the opportunity to prove ourselves in challenges previously unknown. Carlsen is a wonderfully brilliant young man. He may also be a pioneer.

Wednesday, June 16, 2010

The other solution to the Prisoner's Dilemma

Sometimes I'll sit down and blog the thoughts that come to mind. Other times I'm thinking interesting ideas and happen to be near the computer. Then there are the topics that bounce around my head for months, where I intend to write something, but it always feels so...daunting. Eventually you'll get my theory of negative mass, but...not yet.

Of course when you wait long enough somebody else might present your idea first. Thanks "Saturday Morning Breakfast Cereal." Besides making me laugh, you've now gotten me to write a blog post. (I'm sure this idea has been presented before that, but I'm not about to go 'research' the history of it. Independently arrived at by at least me and SMBC. All I need to know. Hurrahs for not being in academia).

So read the comic if you haven't already.

I've seriously been meaning to blog that for months. A slightly different take, naturally, but the idea of Jesus as fundamentally presenting an alternative solution to the prisoners dilemma. In our pro-capitalist society we've come to accept the turn-coat solution to the prisoner's dilemma as the correct one. It's the reason it's a famous thought experiment. Game theory says you should rat out your compatriot. But I find it interesting how many ways Jesus comes out against that solution.

'Turn the other cheek' is one of the fundamental lessons Jesus taught, and I feel like Christianity has lost sight of that. The impulse to fight evil is strong. It's the natural reaction. Half of the solutions to the prisoner's dilemma involve the party trying to do good losing out to the party being mean. So the solution where you're both mean, and thus can't take advantage of each other is tempting. But ultimately, isn't the cooperative solution best?

"It's easier for a camel to go through the eye of a needle than for a rich man to enter the Kingdom of God." I know this religion can be a bit critical of science, but you don't need very advanced physics to understand that won't happen. Even the poor in America are proportionally fairly rich. I have to wonder if it bothers Americans that Jesus has essentially told us we aren't getting in to heaven. The devout must assume they'll push the camel through. Good luck.

But again, from the Prisoner's Dilemma standpoint, it makes sense. You don't get rich in the cooperative column. You get taken advantage of. You get fleeced, your money taken. You build up wealth in the turncoat column. That's the basis of game theory, of objectivism. The idea that things will go well for everyone if you always look out just for yourself. But Jesus' point, and it's a valuable one, is that looking at your own self gain is not the point. Look at the global good. That's what you should optimize.

If someone hungry steals your bread, isn't that a net positive in the world? Now two people can eat. The concern is always that this line of thinking leads to the immoral getting an advantage. People so often oppose social safety nets because some percentage might take advantage of it and not work. So? Just because someone's bad, why are they less deserving of comforts and rewards?

It's a weird way to think, but when enough people think that way things are better for everyone. If you can accept good things happening to bad people, and bad things happening to good people, you can see to it that better things happen to people in general. I find it interesting to think of heaven and hell not as brimstone and harps waiting on the other side of the graveyard, but as the future. If we're good, our children can live in peace and sustenance. If we're bad, they'll have to eke a living from a scourged shell of a planet. And how to we get to the heaven-as-a-good-future? By ignoring our personal desires, ignoring the overwhelming urge to meter justice, and to just do best by the world. Two thousand years later I think we're still misunderstanding Jesus. Do unto your neighbor as you would have him do unto you.

Sunday, June 13, 2010

Putting the fun in fungible

Money is a classic example of a fungible resource. That means you don't particularly care which twenty dollar bill is yours, or even if you've got two tens instead. Houses are not fungible: you can't just arbitrarily trade two houses and expect both participants to be happy. If I lend someone five dollars, I don't expect them to retrieve that particular bill when they pay me back. If I lend them "Catcher in the Rye", though, I don't expect to be returned "Alice in Wonderland" or even another, heavily highlighted copy.

This is all nice, and helps the economy run smoothly. But the fungible nature of money can make thinking about money deceptive. Five dollars is five dollars, but five dollars spent in one place is not equivalent to five dollars spent elsewhere. The money is fungible, the act of spending money is not.

Think about a decision to send a spacecraft to the moon, at a cost of perhaps a billion dollars. At that price, we could probably buy every human in the world a serving of rice for a week. But does that mean the spacecraft took 50 billion meals of rice away from the world? No, of course not. Rice production and spacecraft production are such disparate worlds that money spent in one shouldn't really affect the other very significantly.

What did spending the billion dollars on the spacecraft cause? Well, oil, steel and electronic elements were probably the biggest physical expenditures. Thus a little less steel might exist for sale on the market. That in turn may have inched its price up just a fraction, and perhaps a few people in the long run ended up holding off buying a new car. Electricity was certainly used up heating and lighting factories and offices, powering computers where the design occurred. Because electricity is also fungible it's hard to point to one particular thing the spacecraft effort impacted. But perhaps higher energy costs ultimately caused a handful of families to keep the thermostat a little lower that winter. Resources like electricity and steel have limited availability, so in one sense choosing to use them for one task removes their availability for another task. But you can't convert rice into steel to meet one need with another.

Which is all a simplification, of course. Oil use could increase demand for alternative fuel sources like ethanol which could use up corn, which could increase demand for rice. Or the electricity used by the offices might just come out of excess capacity they had just in case, increasing efficiency in the system and thus being 'free'.

When you've got a billion dollar project, the priciest part is usually the humans. Thus the investment in a spacecraft is less significant in terms of allocation of steel then it is in allocation of minds. The money goes to scientists to design and build this craft. Perhaps it'll improve employment rates a tiny bit, but it seems likely most of the people involved in building a spacecraft could find other work. So the projects biggest impact might be taking minds away from other tasks.

One line of thinking holds that even though none of the scientists would have been rice farmers, they might have been, say, a manager at an industrial factory. So now a line worker becomes the manager, opening up a position a farmer's son fills instead of working the farm. But even if there is some truth to this, every subsequent link lessens the impact.

People are the ultimate in non-fungible resources. We're all different. Perhaps one of the scientists involved in the spacecraft effort would have worked designing cars instead, pushing fuel efficiency up a tiny notch. Or maybe another would have been a manager, and done a poor job at it, causing rippling effects in stock prices and resource usage. The existence of a billion dollars worth of work for aerospace engineers might influence the next generation in college, producing more in that field and less in, say, astronomy. That might influence very gently our understanding of the cosmos. It might be an inefficient allocation if fewer spaceships are build in the next generation.

And we could do the same thing for buying a fifty billion servings worth of rice. How does land usage change? Nitrogen usage? Transportation of the rice? Who do you need to farm and distribute the rice, and what would they have done otherwise? Although each project costs $1 billion dollars, they aren't fungible in how they impact the world.

It's easy to just visualize the economy as a ledger. A dollar is spent here, fifty dollars of oil are burnt there. Because dollars are such an easy abstraction to think about, it becomes tempting to equate things with the same cost. A second car or a vacation can be equivalent decisions in terms of cost, but they can ripple through the economy in very different ways.

You'll never predict all the ways an individual economic action will ripple through time. But it is important to keep in mind when thinking about money. People often complain about spending money on NASA when people are hungry in the world. But they're aren't trivially connected. Cutting NASA doesn't end world hunger. Similarly, the recent financial crisis has revealed that a company's bottom line isn't necessarily proportional to it's positive impact on the world. Money is an invaluable abstraction, but it's good to sometimes take a step back and think about the millions of moving parts involved when you buy a pack of gum from the store.