imdef @imdef@social.cleverlibre.org

A mathematical (not political) post on forecasting, uncertainty, and martingales.
A few years back, Nassim Nicholas Taleb and Nate Silver had a fierce debate about how to forecast uncertain events like elections. There are some interesting general principles here worth thinking about. The thrust of NNT's point was that Silver's 538 forecast was too volatile through time — it wiggled too much to be a "good" forecast. Why does this matter?