Silver & FiveThirtyEight: Wrong

Unscientific methods will do that to you

Funny thing about pseudoscientific metrics: they will inevitably fail

Nate Silver, 2012: "no precedents..for a candidate losing with two or three-point lead"

FiveThirtyEight, 2023 "Races within 3 points in the polls are little better than tossups...been shouting it for years"

Which is it? Those two observations are incompatible.

Their approach, because it's not scientific, can't explain sh*t.

*Mine literally predicts this is what we should see*

Gray = L

Black = W

Notice anything?

👇

If your response is “lol holy shit” you'd be in good company.

Notice what happens when a candidate's poll average - regardless of their “lead” clears 48%. And 49%. And 50%.

If “spread” were a valid metric - if it was sufficient to predict election outcomes (which the worldwide consensus of experts say it is!) then this observation would not exist.

Richard Feynman once said: if it disagrees with observation, then it's wrong.

And this “spread” obsession is wrong. QED

The question isn't IF it is wrong, it's what do we do about it?

I outline those steps in my book, which (insert Bernie Sanders meme here) I am once again asking you to preorder, please.

The publisher has requested that I not share too much content directly from the book, so I'll leave you with these posts from 2020:

But you know.

Little better than toss-ups.