This piece by Nate Silver, How I Acted Like A Pundit And Screwed Up On Donald Trump in FiveThirtyEight, is ostensibly about how he messed up in his predictions on the rise of Donald Trump. What I think is worth reading is how he goes about his work and what he learned from his mistakes. Specifically, it’s a great study on how important models are and how a good model works and what it can tell us.
Related, Paul Krugman talks about his model here: Economics and Self-Awareness in The New York Times. Like Silver, he uses models both to understand and predict. Obviously they are modelling different things, but in both cases good models are the basis of their thinking and the work they do.
It’s likely too much to ask now, but eventually anyone doing analysis and making predictions should have to disclose the models they are basing their decisions upon. The opinions of anyone not having such models are likely not worth much.