>>12929932Here's a rough summary of the first 25% (I'm still watching the video)
For a particular county in a particular state, he got a graph of population against age.
He compared it with a graph of voter registrations against age. Similar shaped graph, and in my opinion, this is as you should expect.
Then he gets a graph of number of voter ballots against age. Here he notes that there's a VERY similar shape between registrated voters per age, and ballots per age. I'm unsure if this is unnatural or not.
He says you can map the number of registrated voters to ballots (at almost 1:1) by using a ratio that varies with age. This ratio vs age can be described as a 6th order polynomial, which he calls a key. At this point I'm sceptical, because the fact that he had to go to a 6th order polynomial says to me that he was really struggling to make them fit.
But the interesting thing is, using this same key, you can map every single county's registrated voters to ballot number in the exact same way, and it fits perfectly.
As soon as you go to the next state, a new key is needed (a new 6th order polynomial), but then once again it fits all counties in that state.
