let's take a step back.

my prediction in 2016 was a trump victory, but i based it on a media analysis, rather than a polling analysis. i'll say this again...this is what i said...i'm paraphrasing....the posts will be up sooner than later....

while the data seems to point to a clear clinton victory so long as the election is fair, it is increasingly clear from the media coverage that the election has already been decided for trump.

so, when i put my mathematician's hat on, i predicted a clinton victory. but, i knew better than to wear a mathematician's hat when analyzing american elections; my media analyst hat told me that it was obvious that trump would win.

i'm not going to look this up now and i don't remember if i've already done it, but the polls in pennsylvania & wisconsin were really not as close as some people have pretended they were. i understand what error is, and i'm quick to point out a bad analysis of error when i see it. but, you have to also look at what the polls actually say and the reality is that all of the polls had clinton ahead in all of the states that she lost, by surprise. people arguing that the data was there really aren't being honest. what was there was the possibility of an error in the data, which is what the analysis is for. but, it's not correct to argue that the result was predictable because you got a 25% chance of trump winning, when you crunched the numbers; we don't run 100,000 elections, we run one, and the entire concept of a probability gets blurry and weird when you start dropping it into theoretical scenarios like that.

when all of the polls point to the same result, it's not a leap of logic to make the obvious conclusion. the reality that they might be wrong is always there, but nobody looking at the data saw this coming (although there were other people studying media that picked out clear warning signs), and anybody pretending they did is being disingenuous - what they saw coming, correctly, was the possibility that the polls were wrong, which is always there. and, that is the truth of it, if you take it at face value - that the polling was wrong, and more wrong than you can assign to statistical randomness.

so, i'll make a clear statement regarding the polling i've seen before i back off.

the race is currently actually very tight, where it matters, which is different than in 2016. four years ago, the polling that we had in the three key blue states that flipped - michigan, wisconsin, pennsylvania - universally had clinton ahead, even if by small margins, sometimes. today, those numbers are quite a bit closer than they were, then, even if the national polling doesn't look much different.

it consequently seems as though the data is suggesting that trump is going to get completely destroyed by much larger margins in states like california and new york (which he was going to lose anyways.) but is actually outpolling himself relative to four years ago in the important states.

many, many things could change in the next few months and weeks, and if the election is mostly mail then we're dealing with a situation that....we don't know if telephone sampling is useful for predicting mail voting or not. it's unprecedented.

but, all things equal, it would seem to me, right now, that my polling analysis is that trump is actually set to pull off a repeat victory via more or less the same path, even if the states are switched up a little bit.

the simple statement is this: in order for the democrats to win the election, they will need to improve their standing in specific states. right now, it's hard to see which states they hope to win in 2020 that they didn't win in 2016, given that biden's polling is essentially the same as clinton's, but slightly weaker across the board.

i don't expect to post frequent updates, but my caveat is that my media analyst hat was more useful than my polling analysis hat was, four years ago, and it's way too early to even put that hat on, right now.