Fair, but I believe it’s equally fair to measure likely voters, which has proven to be relatively accurate within a tight band for elections. In looking at the math I was looking at measured likely outcomes, not outcomes we hope would happen (aka far better voter turnout).

As to third party candidates, it’s a challenge. However, I’d offer two counterpoints:

— More than a dozen GOP nominees, for better or worse, led to Donald Trump. Up to everyone to judge whether more choice led to a better choice or not.
 — There’s actually 5 candidates on more than 20 states, and dozens otherwise (http://www.politics1.com/p2016.htm), but the math treats even Johnson and Stein as rounding errors. Not always the case (’80, ’92 the two most recent cases) but the case here.