ted.com — Bruce Bueno de Mesquita, a consultant to the CIA and the Department of Defense, uses mathematical analysis to predict (very often correctly) such messy human events as war, political power shifts, Intifada ... After a crisp explanation of how he does it, he offers three predictions on the future of Iran.
Apr 7, 2009 View in Crawl 4
cresswgaApr 7, 2009
If you read the small print he said he is right 90% of the time when the experts were wrong. But it doesn't tell you how often his model contradicted the experts when they got it right and he didn't.Without knowing the accuracy of the experts he refers to it becomes a very misleading statistic.
mclewellApr 7, 2009
Why, Pythagoras dealt with geometry, not game theory?
darkyuubiApr 8, 2009
5! = 5*4*3*2*1 = 120Thats what he said. 5 factorial.
zmjone2992Apr 8, 2009
have you ever read a political science journal? obviously not.... there is a great deal of empirical analysis. it is a social science, and you are a moron.
kaelyiestaApr 8, 2009
The prediction outcome isn't necessarily going to contradict the experts agreement. Think of it as a two boolean variable problem space. You can have 4 outcomes,not 2, in this probability area: Both agree, both disagree and two conflicting predictions. His statement essentially is a conditional probability reducing the probability space to just when the experts are wrong, leaving two possibilities left. In that probability area, the model prediction disagrees with experts .9, and agrees(wrongly) with the experts .1.
kaelyiestaApr 8, 2009
If he were talking of lines, then you would be correct. But he is talking about information sent, which between two people is going to be two items, not one. Watch the video again, a pay closer attention. He states this explicitly in his arguments. His image is just simplified.
kaelyiestaApr 8, 2009
If each item of information is counted, and each person sends an item of information to each other, then you would have the first person send his information to all others (N-1), and the second person send his information to all others (N-1) and so on for all N people.If that's the case then the solution for 5 people would be 20. You can draw this yourself to be convinced, just make sure you draw lines to and from each two people.Since this isn't the number he gave(he said it was a factorial, not a combination), he must mean something a bit more dynamic where the information sent also considers the information that influenced the given person, so it's like an instant in the middle of this ongoing exchange and considering each influence into the past as part of the number of things to track.
doogiehowitzerApr 10, 2009
I know what 5! is, but it has nothing to do with his scenario