Re:Why Inequality Matters
by TeMPOraL on Thursday October 16, 2014 @01:27PM
(#8466107)
But there comes a class of problems which I guess need no fictional example, because we're dug up to the neck in them - coordination problems. For instance, it would be better for people to agree and stop polluting the environment, or making crappy food (I don't mean fast-food, I mean borderline-unsafe food made as cheaply as possible), or switching away from fossil fuels. But we can't do that, because anyone who tries will get quickly outcompeted in the market by anyone who defects and keep doing the wrong thing. Therefore, we're stuck in local optima instead of moving to better global ones.
There is an excellent piece by Yvain that goes into deep details of those problems[0], so I won't be repeating him here much, but the world is full of such problems, and apparent stupidity of institutions and markets are usually result of inability to coordinate.
I'm not saying to give up democracy. But we have to appreciate that it has very troubling failure modes and that they're not theoretical.
Real-life democracy tends to be extremely impotent. It has tremendous operating costs (think of all the resources wasted in a zero-sum game of political campaigning), and in the end people still can't agree on things like building infrastructure (NIMBYsm) or creating laws that benefit the population. Again, I'm not saying "ditch democracy". But there is a huge trade-off between security of decentralization and effectiveness of centralization.
There's also the case that central planning doesn't suck that much - it sucks when done for humans by humans, because central planners operate at human speed. But consider that instead of Paco setting up routes, there is a computer named PCO, tasked with collecting real-time data from all drivers and optimizing routes. PCO may alter some of instructions it sends to drivers to explore the solution space, and it will quickly figure out that doing only rights leads to less fuel used. And then it may stumble upon other optimizations, like shifting delivery times in different parts of the city depending on traffic patterns - things that are impossible to figure out by Pablo, because they require global view.
I think that central planning got a bad name because of Soviet implementation, but with modern computing, maybe we should re-evaluate its costs and benefits.
There is an excellent piece by Yvain that goes into deep details of those problems[0], so I won't be repeating him here much, but the world is full of such problems, and apparent stupidity of institutions and markets are usually result of inability to coordinate.
I'm not saying to give up democracy. But we have to appreciate that it has very troubling failure modes and that they're not theoretical.
Real-life democracy tends to be extremely impotent. It has tremendous operating costs (think of all the resources wasted in a zero-sum game of political campaigning), and in the end people still can't agree on things like building infrastructure (NIMBYsm) or creating laws that benefit the population. Again, I'm not saying "ditch democracy". But there is a huge trade-off between security of decentralization and effectiveness of centralization.
There's also the case that central planning doesn't suck that much - it sucks when done for humans by humans, because central planners operate at human speed. But consider that instead of Paco setting up routes, there is a computer named PCO, tasked with collecting real-time data from all drivers and optimizing routes. PCO may alter some of instructions it sends to drivers to explore the solution space, and it will quickly figure out that doing only rights leads to less fuel used. And then it may stumble upon other optimizations, like shifting delivery times in different parts of the city depending on traffic patterns - things that are impossible to figure out by Pablo, because they require global view.
I think that central planning got a bad name because of Soviet implementation, but with modern computing, maybe we should re-evaluate its costs and benefits.
[0] - http://slatestarcodex.com/2014/07/30/meditations-on-moloch/