I don't think this is as surprising as people think, and doesn't necessarily mean that "Florida is the linchpin for the election." Let's stipulate that Trump has a very hard time winning w/o Florida, so that 95% is, if anything, a bit bullish on Trump 1/
So let's take a view of swing states that includes AZ, NE-02, IA, MN, WI, MI, PA, NC, GA, FL, ME-02. That puts Biden at 220 EVs to Trump's 182. 2/
Right now, the models that are showing Trump at 25% or whatever to win have Florida as roughly 50-50, give-or-take. What we're doing is taking the *conditional* probability of Trump winning Florida. 3/
That is, we can take Trump's 50% chance of winning Florida, setting it to 100%, and then rescaling. In other words, we're looking at the *only* at the subset of maps where Trump wins Florida and rescaling. Remember, some of those maps probably have Trump winning handily! 4/
Which means we're probably only looking at maps where the popular vote is in the margin-of-4-percent-or-less. In that universe, Trump is almost certainly winning GA, NC and IA in most maps. So we're now at 220D to 248 R in the EC. 5/
Given that Trump wins FL, NC, GA and IA, and the popular vote is <4%, I completely buy that Trump wins a majority of those elections. Put differently, spot Trump 66 of the 136 EVs from swing states, and yes, his chances are very good. 6/6
