Why the funny name?
First things first. What's with the ridiculous name? I'm glad you asked, observant reader, because that's probably the best way to introduce what this blog is about. In Greek mythology, Prometheus and his brother Epimetheus were Titans who were given the responsibility of creating humans and animals to populate the earth. While Prometheus was teaching the humans things like how to do math and write poetry, Epimetheus was giving all the cool gifts, like fangs and claws and wings and venom, to the animals. When Prometheus finally was done with our lessons, there were no good physical traits left to give us. Not wanting to see his beloved humans devoured by lions and tigers and bears, he decided to steal Fire from the gods and give it to us. I'd love to go more in depth, but if I start talking about mythology here I'll probably never stop, so we'll leave it at that.
Econometheus will be a blog about artificial intelligence and economics; specifically, how to use economic models to design methods useful for artificial intelligence. Similar to Prometheus teaching humans about reason and culture, it's about my efforts to give computers the gift of thinking according to utility-maximizing behavior.
Economics and Simulations
Economists typically shy away from simulations because they're more concerned with studying the reasoning and intuition behind their theories. They want to boil human behavior down to bare essentials, and explain complex decision-making processes in as simple and elegant a model as possible. To an economist, a model is just an abstraction that provides a framework for discussing complex ideas. The problem with simulations is that they really only prove that a model does what it says it does. But I'm not interested in using simulations to understand economics better. I'm interested in using economics to design more interesting simulations.
Now, simulations do have plenty of valuable uses. They are good for gaining an understanding of how complex systems work when you have a good grasp on how to specify the rules that govern the system. In this blog, building simulations is my end, and using economic models will be my means of designing those simulations. My Holy Grail would be to design a system to run something like a Minecraft server full of bots that build their own society, with division of labor, currency, markets, the whole shebang. I'll start small and try to layer in more economic models to get more sophisticated behavior as the system grows. It's also going to be a programming exercise, because this system will get more and more complex. I'll start with Python, but sooner or later I'll probably have to move to something that runs a bit faster. It'll take me a while to get there, but I think it's a worthwhile goal. Any feedback I can get from you, reader, would be greatly appreciated.
About the Author
Now, some details about my own background will probably help set some expectations for what kind of knowledge I'll be able to share on this blog. I studied economics in grad school, but I've really only studied A.I. as a hobby, so I only have some degree of expertise in one of the two subjects. I work on lots of hobby projects (like programming simple games, and toying around with machine learning algorithms), so I'll share what I learn as I go. At any rate, even if I don't attract many readers, it will give me a chance to organize the projects I undertake, so that over time I'll be able to build something cool, instead of just having done lots of little things with no common purpose.
Okay, so much for introductions; I do have more interesting things to write about, and it's time to move on to those.
No comments:
Post a Comment