Yeah, they used the obviously bad model that each state's results are independent rolls of the dice. Which, flawed models can still be useful, but I seem to re

Yeah, they used the obviously bad model that each state's results are independent rolls of the dice. Which, flawed models can still be useful, but I seem to recall them being pretty arrogant about their prediction in 2016 too, before the fact.