Lynne Kiesling
… or he’s being fatuous; I prefer to think he’s trying too hard.
Bryan Caplan’s got a challenge to come up with an example of radical ignorance/Knightian uncertainty that’s better than the Trojan horse example. Arnold Kling’s got a response for him, as does Tyler Cowen. They are all focused on the important distinction between situations in which you can reasonably assume that there is a known probability distribution and situations in which there is not.
I think they are all thinking too hard, although perhaps Bryan, Arnold, Tyler (and Robin, in Tyler’s comments) will all say that my example is degenerate: technological change and consumer adoption of new technologies. Sure, we can say that there is some path dependence in the development and adoption of new technologies ex post. But ex ante, given that we live in a non-ergodic world (to use Doug North’s phrase), can we really say that which inventions get invented, and which inventions get adopted, can be draws from a known probability distribution?
For example, in 2000 we could reasonably have argued that there was a known probability distribution out of which we could draw a probability that Apple would develop a music player. But that’s too obvious a question. The real problem of radical ignorance comes in the form that their innovation would take, and whether or how consumers would adopt it and use it, and adapt it to their various heterogeneous purposes. Could we have drawn those outcomes out of a known probability distribution?
No. And the precise reason we couldn’t is because of the non-ergodic nature of life. We simply cannot know either the exact form of future technologies, or how they will interact with and shape consumer preferences.
For me, that’s the canonical example of radical ignorance.
UPDATE: I said “ergodic” when I meant “non-ergodic”; that’ll learn me to post when I’m rushing off somewhere! Thanks to Gabriel for the catch.