In Part II of this series, I left off with the issue of how much elections turn on candidates' policy platforms, and how much they turn on candidates' "valence" characteristics, like competence and honesty. Microeconomic theory does not prohibit placing any particular weight on either aspect of what candidates have to offer, so neither policy-based elections nor valence-based elections are intrinsically rational nor irrational. However, we can place a normative judgment on a process that places too much weight on one set of characteristics, and while policy positions are strategically selected, valence traits are not.
However, the position of those arguing that Clinton screwed up is not primarily based on her campaign platform in spatial terms. It is based more on things like "she didn't go to Wisconsin, so she didn't deserve to win." I'll save the empirical aspects for later, but from a normative and philosophical standpoint, this is a bizarre idea in that it denies any moral rights to voters.
If elections are intended to created representation and a bunch of touchy-feely-blah-blah-blah, then the result must, in some way, reflect either the preferences or the interests of the electorate. Note the "or." I am referencing an old distinction between what we call "trustee" and "delegate" representation. A trustee representative does what he believes is in the best interests of the public, and a delegate does whatever the public wants, regardless of what the representative believes. Regardless of the form of representation, though, elections are necessary to create some kind of accountability, in which case they cannot simply be games of Risk in which you lose because you gave up Australia, and Australia is the key to the whole game. That removes voters from the process in a moral sense.
Yet, if voters are so mindlessly mechanical that they can be removed from the process in that manner, 1) shouldn't we be really bothered by that, and 2) isn't that, itself, the real issue?
I have argued, and will continue to argue that the 2016 election turned primarily on two factors: the Abramowitz "Time for a Change" model, which gave an intrinsic advantage to Republicans on the basis of two previous Democratic victories, and Comey's late-stage intervention. The former effect is one in which voters behave relatively mechanically, but in a different form from "Trump came here, Clinton didn't, so who cares if he's an incompetent, lying sack of shit who grabs women's pussies?"
In a different way, the Risk electoral model removes voters from the process, morally. It treats them as boxes checked off, who have neither preferences nor interests, making either delegate or trustee representation impossible. Under the Risk model, there is no representation, only Zuul. If the campaign were a Risk game, that should trouble us, and for Trump apologists to make that case, and to do so as a defense and vindicating statement, is fundamentally an attack on the electorate, much like his "I could shoot somebody on 5th Avenue" statement was actually an attack on his own supporters. Trump's defenders are saying that voters have neither preferences nor interests. They are mechanical boxes to be checked off in a model in which representation has no place.
Is there a normatively compelling place for strategy in electoral campaigns? That's a harder question, but we need to address it before moving on to the empirical issue of whether or not this kind of stuff really can explain 2016, although I've made my opinions on that relatively clear many times anyway. I'll just go through it a bit more systematically.