I’m not being entirely facetious when I say that I blame the media.
Rust belt industries aren’t the only enterprises that have been hollowed out by free trade; so has North American journalism. The respective causes are only marginally different.
While our industrial strength has been transferred overseas, media organizations have adopted a similar model by doubling down on digital. The result has been that the number of media workers (aka journalists) in North America has declined by almost 50% since the turn of the century.
The French government did a study a few years ago which claimed (then) that the amount of content on the Internet increases by 10% annually. But the number of eyeballs (aka consumers) has remained relatively unchanged. This has forced news organizations that still think their main role is to deliver readers to advertisers, to look for new sources of revenue. But nothing seems to be working. Eyeballs are still going online.
Increasingly for cat videos.
In desperation, news organizations are resorting to gimmicks including clickbait, more bloviating commentary and ultimately, less real reporting. The CBC is now encouraging some senior journalists to engage only in opinion commentary - mostly on the CBC website, as a potential source of audience growth. This seriously undervalues real reporting by giving the impression that the public broadcaster in Canada prefers opinion over fact-based reporting.
New organizations are also contracting out their intellectual and journalistic property to pollsters who increasingly resemble modern day snake oil salespersons. It is in fact cheaper to pay a polling company than to double down on shoe-leather reporting. So editorial control of polling has been left in the hands of non-journalistic commercial interests.
This is one of the better explanations about how we were so deluded by the pollsters.
To be fair, pollsters are under similar economic pressures: they are also cutting corners by pandering to self-selecting, internet polling and avoiding the time-consuming aspects of serious polling, such as "return to sample" - calling back later if no one is at home when the pollster calls.
There is also the still unexamined aspect of "response bias" - a large percentage ofTrump supporters may have been less likely to admit their preference to a stranger cold calling.
Believing their own flawed results, a lot of polling simply stopped several days before the election, assuming the race was over (see above Dewey Defeats Truman*).
At NPR, I once suggested in a story meeting, that we stop reporting polls for the week before a mid-term election.
“But what will we have to talk about?” said one anxious host.