Comment Re:Psychohistory was terrible science (Score 1) 293
My PhD supervisor (or dissertation advisor, as Americans would call it) often uses the example of psychohistory when explaining the role of statistics and prediction in social science. I personally like the analogy of those physics slit-experiments (can't remember the correct name of it, but the one where you pass individual photos through a slit and they form an interference pattern on the other side, thus suggesting the wave-particle duality of photons). You can't predict what an individual photon can do, but when you have enough of them, you get a distinct pattern.
One thing my supervisor and myself disagree on is the role of agency (or free will). He says that, as social scientists, we aren't interested in free will. When you're analysing the behaviour of a group of people, the overall trends are more important than why any one individual made a certain decision. I don't like to write off agency quite so quickly, but I'm willing to concede that individuals can easily get lost in a crowd.
A few people in this thread have pointed out that considering the trends of billions of individuals, and having those individuals ignorant to your prediction, are important components of psychohistory. Although I suspect that the point about ignorance isn't that central, I'd be prepared to believe that, if you had good information about the beliefs and attitudes of billions of people, some good algorithms, and a bloody big computer, that you'd probably do a reasonable job of predicting a couple of years into the future. I think it's probably less difficult than that other staple of science fiction: faster than light travel.
One thing my supervisor and myself disagree on is the role of agency (or free will). He says that, as social scientists, we aren't interested in free will. When you're analysing the behaviour of a group of people, the overall trends are more important than why any one individual made a certain decision. I don't like to write off agency quite so quickly, but I'm willing to concede that individuals can easily get lost in a crowd.
A few people in this thread have pointed out that considering the trends of billions of individuals, and having those individuals ignorant to your prediction, are important components of psychohistory. Although I suspect that the point about ignorance isn't that central, I'd be prepared to believe that, if you had good information about the beliefs and attitudes of billions of people, some good algorithms, and a bloody big computer, that you'd probably do a reasonable job of predicting a couple of years into the future. I think it's probably less difficult than that other staple of science fiction: faster than light travel.
