I apologize for the delay in posts for those 3 or 4 of you devoted readers there are. This is a post I've had stewing in my head for a while and am finally getting around to writing down a few preliminary words about.
I've been spending a lot of time thinking about, or more precisely being vexed by, what I perceive to be the shifting and kind of unsteady landscape of knowledge in our modern world. I'm thinking about things kind of across the smorgasbord, from facebook, twitter, and the blogosphere overloading us on connectivity, to cheaper and faster genome sequencing and biotechnology, to the impossibly complex abyss of the financial system. As I immerse myself more and more in this world (for better or for worse), I have a deeply ambiguous feeling about its effects on me, and by extension our modern society. On the one hand, it seems incredibly cool and also incredibly important that knowledge is being democratized. On the flipside, though, I can't help but feel that something, some element of depth of understanding, is being lost in the face of all of this complexity, and this is what I'm trying to wrap my mind around. As a beginning to what I intend on making a series of posts, I just want to bring up a few issues I've been thinking about as of late.
1. Is wisdom a meaningful idea anymore? It seems to me that one of the classical distinctions between wisdom and simple knowledge was that wisdom entailed some sort of intuitive understanding, where knowledge implied simply learning or retaining a piece of information in your brain. Does this have any meaning for us when the things that we are trying to "intuit" are multi-dimensional, microscopic, or beyond the scope of our senses? Is evolutionary computation, which I've talked about a bit before, an example of how we can incorporate the "wisdom" of our decision making into much smarter thinking machines than our own minds?
2. What should be the role of emotion in our decision making? Without going in too deep into details of emotional decision-making theory, it seems pretty clear to me that developing emotions is really important in our ability to cleave through the complexity of the world and ignore some decisions that don't really need to be made. If nothing else, it's a useful filter. In the face of so much information, how is the role of emotions going to change, since in my experience, emotions (at least those involved in decision making) develop slowly, something which seems to be at odds with the current pace of information processing and gathering.
So, just stew on those for a while, and I'll be back with more thoughts soon!
Showing posts with label evolutionary computation. Show all posts
Showing posts with label evolutionary computation. Show all posts
Sunday, March 15, 2009
Sunday, January 25, 2009
Competition, Farms, and the Democratization of Innovation
Undeniably, we are now living in a world that is exceptionally complex. Although the world may not be any more complex and dynamic than it was a few centuries or millenia ago from a physical perspective, it seems clear that from a social perspective the world has skyrocketed in its complexity, especially since the industrial evolution and the process of "globalization." It also seems clear that this trend will almost certainly continue exponentially in every sector of life. From the personal and social connections we make to financial markets and international governance, the world is getting bigger and more interconnected.
In general, I find this to be a pretty hopeful and beautiful thing: knowledge is being democratized (by the internet, as well as big-minded projects like PLOS and One Laptop Per Child) and I think for the first time we are really able to actually imagine the solutions to problems that are fundamentally international in nature (e.g. the drug trade, hunger, poverty, and especially global environmental issues).
This new kind of world, though, is also going to bring about a whole new set of problems. This was highlighted for me recently during the beginning of the economic shitstorm when I was listening to an NPR interview, which was actually about artificial intelligence. What stuck with me about the interview was the discussion of the computer programs that determine what happens in the financial markets. Obviously, the financial markets are too complex for anyone to predict in their specifics. And so, lots and lots of shareholders (which i think include private, small-time investors, although I could be wrong about this, I know embarrasingly little about the world of money) put sort of preset "sell" mechanisms on their holding. So, if the values of certain stocks fall below these setpoints, it can start a sort of wave of selling, which then can sort of reveberate out through the market. What was amazing, and sort of shocking to me, was to hear that not only is this very complicated (and apparently important) system being run at least in part by a bunch of computers, but that the effects were so complicated that nobody could figure out what happened, so that they had to write other computer programs to figure out what the other computer programs were doing!
So, this is scary because (a) the machines are taking over (huge vindication for all the sci-fi nerds), but also because (b) it represents how we've created mechanisms in the world that can themselves gain complexity and then sort of go beyond our ability to easily understand and control.
So, long introduction, but this got me thinking about how things like innovation, regulation, and control are going to have to change in this new way-too-complex future. In particular, I think we're going to have to come up with "smart" systems for understanding and controlling these very complex global systems.
One model for this that I find very intriguing is called "evolutionary computation," from computer science. From what I understand, evolutonary computation is a name for lots of different styles of computation that involve creating iterative programs that will progress slowly towards a "fitter" solution through many generations of calculation.
One example is from architecture. Imagine there is some leeway in how you can arrange some structural elements (maybe "struts") spatially in a building, but you want to find the best arrangement. And you know that you want to maximize some aspect of the building, let's say "toughness." People are creating programs that will randomly generate a whole variety of different arrangements of struts (the parallel of mutations in a natural population), and then they test all these different arrangements for "toughness." The specific arrangements of struts that perform best get to have "offspring," or new arrangments that are roughly like them but with some more mutations, and then the whole process is repeated. The result is basically that the program roughly imitates the smart processes of evolution and designs something all on its own.
Although this is a rough example, I think it still points to a useful direction in how we can sort of "decentralize" the analytical thought process. I think the same kind of thing can apply to societies. If we can figure out ways to bring innovation and analysis away from central "brains," such as centralized policies or governments, and out into the rest of the world, this opens up tons of new possibilities. And, it may even be the only way to move forward as the mechanisms of the world (such as financial markets and the internet) become too gigantic for a centralized brain to handle easily.
So, as a final endpoint, I want to bring this around to what I've been getting really interested in recently: local farms. I've recently started volunteering at a local organic farm, and I've been incredibly impressed with the innovative and dedicated work that is being done with basically no resources and very little overhead support from either a government or an academic institution. They're working on projects to build a solar-powered kitchen, expanding the visibility of good food by bringing it to impoverished communities for very cheap, having a roving biodiesel delivery system for their crops, and tons of other stuff. My point is that this is what we need more of: small-scale, decentralized innovation and problem-solving. In the spirit of Thomas Friedman, I think the best way for us to stay competitive in the modern world is to decentralize the system of innovation and encourage people to make change on the grassroots level (something that would seem especially appropriate in the age of Obama). There's lots of smart people out there with ideas that could make swift and effective change. Let them do their thing.
In general, I find this to be a pretty hopeful and beautiful thing: knowledge is being democratized (by the internet, as well as big-minded projects like PLOS and One Laptop Per Child) and I think for the first time we are really able to actually imagine the solutions to problems that are fundamentally international in nature (e.g. the drug trade, hunger, poverty, and especially global environmental issues).
This new kind of world, though, is also going to bring about a whole new set of problems. This was highlighted for me recently during the beginning of the economic shitstorm when I was listening to an NPR interview, which was actually about artificial intelligence. What stuck with me about the interview was the discussion of the computer programs that determine what happens in the financial markets. Obviously, the financial markets are too complex for anyone to predict in their specifics. And so, lots and lots of shareholders (which i think include private, small-time investors, although I could be wrong about this, I know embarrasingly little about the world of money) put sort of preset "sell" mechanisms on their holding. So, if the values of certain stocks fall below these setpoints, it can start a sort of wave of selling, which then can sort of reveberate out through the market. What was amazing, and sort of shocking to me, was to hear that not only is this very complicated (and apparently important) system being run at least in part by a bunch of computers, but that the effects were so complicated that nobody could figure out what happened, so that they had to write other computer programs to figure out what the other computer programs were doing!
So, this is scary because (a) the machines are taking over (huge vindication for all the sci-fi nerds), but also because (b) it represents how we've created mechanisms in the world that can themselves gain complexity and then sort of go beyond our ability to easily understand and control.
So, long introduction, but this got me thinking about how things like innovation, regulation, and control are going to have to change in this new way-too-complex future. In particular, I think we're going to have to come up with "smart" systems for understanding and controlling these very complex global systems.
One model for this that I find very intriguing is called "evolutionary computation," from computer science. From what I understand, evolutonary computation is a name for lots of different styles of computation that involve creating iterative programs that will progress slowly towards a "fitter" solution through many generations of calculation.
One example is from architecture. Imagine there is some leeway in how you can arrange some structural elements (maybe "struts") spatially in a building, but you want to find the best arrangement. And you know that you want to maximize some aspect of the building, let's say "toughness." People are creating programs that will randomly generate a whole variety of different arrangements of struts (the parallel of mutations in a natural population), and then they test all these different arrangements for "toughness." The specific arrangements of struts that perform best get to have "offspring," or new arrangments that are roughly like them but with some more mutations, and then the whole process is repeated. The result is basically that the program roughly imitates the smart processes of evolution and designs something all on its own.
Although this is a rough example, I think it still points to a useful direction in how we can sort of "decentralize" the analytical thought process. I think the same kind of thing can apply to societies. If we can figure out ways to bring innovation and analysis away from central "brains," such as centralized policies or governments, and out into the rest of the world, this opens up tons of new possibilities. And, it may even be the only way to move forward as the mechanisms of the world (such as financial markets and the internet) become too gigantic for a centralized brain to handle easily.
So, as a final endpoint, I want to bring this around to what I've been getting really interested in recently: local farms. I've recently started volunteering at a local organic farm, and I've been incredibly impressed with the innovative and dedicated work that is being done with basically no resources and very little overhead support from either a government or an academic institution. They're working on projects to build a solar-powered kitchen, expanding the visibility of good food by bringing it to impoverished communities for very cheap, having a roving biodiesel delivery system for their crops, and tons of other stuff. My point is that this is what we need more of: small-scale, decentralized innovation and problem-solving. In the spirit of Thomas Friedman, I think the best way for us to stay competitive in the modern world is to decentralize the system of innovation and encourage people to make change on the grassroots level (something that would seem especially appropriate in the age of Obama). There's lots of smart people out there with ideas that could make swift and effective change. Let them do their thing.
Labels:
competition,
evolution,
evolutionary computation,
farming,
politics
Subscribe to:
Posts (Atom)