Recently it's become expected that if you want to be taken seriously as a forecaster that you should not only record your predictions in advance, but assign a confidence level. And that by following this methodology certain people, so called superforecasters, have been found who are significantly better at prediction than average. The problem with this approach is that while these individuals are great at predicting should things continue mostly as they have, they're actually worse at predicting extreme events, which are inevitably the most impactful.