If one pollster had failed to accurately predict one result, you could conduct a fairly simple investigation: Had they chosen their samples incorrectly? Had they asked their questions misleadingly? When every poll gets it wrong, with increasing and alarming frequency, the problem is no longer methodological but metaphysical. There are, of course, some perfectly reasonable explanations. For starters, there are anything-but-surveyable patterns of voter suppression and voter lethargy, together with steady influxes of new, never-before-surveyed voters in the electorate. There’s also the huge methodological difficulty that most polling is traditionally carried out via phone, and large sectors of the electorate are no longer happy to answer the phone when unknown numbers appear on the screen.
But these second-order obstacles aren’t enough to explain the current collapse of poll-driven political certainty. They’re just excuses, even if they’re not untrue. Something about the whole general scheme of polling—the idea that you can predict what millions of undecided voters will do by selecting a small group and then just simply asking them—is out of whack. We need to think seriously about what the strange game of election-watching actually is, in terms of our relation to the future, our power to choose our own outcomes, the large-scale structure of the universe, and the mysteries of fate. And these questions are urgent. Because predictions of the future don’t simply exist in the future, but change the way we act in the present. Because in our future something monstrous is rampaging: it paces hungrily toward us, and we need to know if we’ll be able to spot it in time.
more here.