Thread by @gelliottmorris

I still can't engage with @NateSilver538 on this site but this tweet misses a v important nuance of the post, which is that a model with fatter tails actually makes our aggregate predictions *worse* for 2008-2016, even though it makes the state-level predictions more inclusive.

This might have something to do with the way we parameterized our a correlated error and how relying on the fundamentals helped us shrink toward 50-50 in 2016 -- Nate needed fat tails then to control for poll error but our model hedged by picking up pro-D bias in swing states.

Also, adding fat tails on the order of what Nate uses only pushes our model down to 93% — so it's not like a huge difference. The bigger reason there's a gap bt our models is that 538 includes some very R-biased partisan data that's pushing their avgs toward 50-50 in key states

Also, this is an entirely separate issue than Nate inflating his time x uncertainty interaction, which was probably the wrong call as the polls turned out to be stable and rather Biden-leaning throughout the cycle (as we theorized).

So while our overall trend could be a few points overconfident (again, at the state level — bc the way we've set it up produces reasonably *conservative* national calibration), I don't buy Nate's argument that his model with a whole lot of movement in the fall is right by default

Anyway, thanks for coming to my TED Talk — I wish Nate and I were on good enough terms to @ each other here because obviously we have more in common than we disagree on