My hot takes on contemporary Twitter discussions:
(1) I think Trump is likely to lose. Maybe 1-in-5 chance he wins(?)
(2) I don't think the 1988 example is terribly apposite; that was in the midst of a convention bounce, Trump is an incumbent, polarization, etc. BUT 1/

If Trump got the 25-point net turnaround Bush ended up getting he'd win a Reagan-like landslide. He doesn't need to make up 17 points. Remember, if he loses the popular vote by four there's a good chance he wins in the EC. So he needs to make up more like six points. 2/

(3) The more you emphasize "well, those older elections are TOTALLY different because of polarization, etc" the harder it is to model. You can try polarization variables, though Abramowitz tried that in '16 and IIRC it made his model worse. More to the point 3/

this is sort of ad hoc. If you're pressing this to the point that older elections are just from a different election distribution altogether, then you probably shouldn't use the data at all. 4/

Which leaves us with like eight datapoints, depending on when you think polarization sets in.
(4) The case for Trump winning would go like this: Despite presiding over an annus horribilis, his job approval seems to have bottomed, for now, at ~42%. 5/ https://www.realclearpolitics.com/epolls/other/president_trump_job_approval-6179.html

That's not good! He almost certainly can't win with that. It might go down, but there's a reasonable chance that as masks become more widespread the bending of the second wave curve we're seeing becomes more pronounced. 6/

Maybe Trump doesn't get credit, although while you can make (a) the pandemic is THE issue in the election and (b) Trump won't get credit for improvements work, that probably isn't the Occam's Razor takeaway. 7/

I've long said he needs about a 45% job approval to be about break-even for the election, a three-point improvement that happens to match up with him being behind 6-7% (net) in most of the battleground states. 8/

There are obviously answers to all of this -- if there weren't I would not have led with him losing four of five times or so. The point is, though, that he's not THAT far off where he needs to be to be back in this. 9/9