Pieter of Peaktalk is the latest person I’ve seen who notes an incredibly small “undecided” share of the electorate.

It seems to me that this flies in the face of everything political scientists believe about presidential elections; while the default reaction of most partisans, and independent leaners, is to vote for their party’s nominee (despite the caveat of reciprocal causation—party identification is influenced, in part, by the candidates fielded by the parties), it seems unusual for voters to declare themselves so firmly committed in the early stages of the fall campaign, and usually there is some shifting in commitments over time as the campaign continues. By contrast, the media analysis seems to reflect the degree of elite polarization, which—while high—is typical of presidential campaigns.