Obsessing over last week's or this week's polls is irresistible, but they are pure noise (see here). This essay in the NYRB makes for timely reading (I've been reading the Achen & Bartels book, on recommendation of Jason Brennan, and it is instructive on a number of points). 10% of the electorate might actually be undecided; everyone else will largely vote their tribalist loyalties. Events in the world, or what the candidates say and do, are not wholly irrelevant, but apparently fairly close to being so. Here's two things that might matter:
1. An economic collapse of 2008 proportions in the next couple of months--that would favor Trump, not because he'd have a rational response to it, but because the Democrats would be blamed (also irrationally).
2. A catastrophic terrorist attack (e.g., a 9/11-scale event in the US) or catastrophic military event (e.g., Israel bombs Iran, Russia invades Urkaine). This might favor Trump as well. (Assuming ISIS favors Trump, which they presumably do [assuming they are somewhat instrumentally rational], then that probably raises the risk of a domestic terrorist incident in the coming months.)
I realize all of these are pretty dramatic scenarios. But read the NYRB essay. Clinton's pick of Kaine is explained pretty easily by the thought that there's only 10%, if that, of the electorate up for grabs. And barring one of the preceding extreme scenarios, it's still a very good bet Clinton will win, and win by the margin she's been leading for most of the last eight months, namely about 4% or so. If 1 or 2 occur, or something else comparable, then all bets are off. Anyone have contrary evidence? Informed criticism, with links preferably, of Achen & Bartels would also be welcome.
