expectation? So he was trying to measure not just the accuracy of people's forecasts, but also the calibration of people's level of confidence to their accuracy. So he's got 18,000 observations over this long period across a very wide variety of complex domains. Questions like, will the inflation rate in France rise above 3% or not? Russia, will the USSR expand their border in this direction? So very specific questions were very specific. And anyway, so all this data, he summarized the results. So some interesting counterintuitive things came out of this. Number one, people are not very good forecasters in their own domain of expertise. In fact, when he asked questions to experts outside of their domain of expertise, they tended to actually be both more accurate and more well calibrated than within their domain of expertise. The experts most likely to be cited in academic journals or appear on TV or in newspapers, less accurate and less well calibrated than those who toil in greater (10/45)
You are viewing a single comment's thread from: