BC #pollfail, in both senses, just as the Wildrose polling fiasco was beginning to fade from our memories. The Clark enviro-vandals are back in power, even if Christie herself is not, and the pollsters, once again, are left desperately searching for a clue in the cold entrails they’ve been consulting for weeks.

OK, to ask the obvious, what happened?

This sort of thing doesn’t just happen here. The Brits asked that eye-rubbing question in 1970, and again in 1992 when the polling results were historically out of whack.

Speculation abounded in 1970, when Labour Party leader Harold Wilson lost his sure thing. I recall commentary at the time as well that the surprise turnabout was an uprising against the polls themselves, a demonstration of electoral free will.

In any case I raised a few points after the Wildrose upset last year. They still seem pertinent.

At the very last moment in the Alberta race, a couple of antediluvian Wildrose candidates generated some unwelcome noise. Wildrose leader Danielle Smith blamed that, and strategic voting by Liberals, for her party’s defeat. But:

I would suggest that there is a third factor to consider—the polls themselves. The recursive nature of polling never seems to be taken very seriously, but it should be. Polls affect the very people being polled. As voters saw the “inevitable” victory of the WRP looming, various effects could be, well, predicted. One would be that some WRP voters, confident of the coming landslide, might stay home. Those opposed, of course, would be more moved to turn out, and, swallowing hard, to change their vote to a strategic one towards the end, prompted as much or more by the polls than by a couple of Albertosauri.

The sheer unanimity of the pollsters could not but have played a significant role in determining the paradoxical outcome of their public crystal-gazing. The more general question before us, then, is whether these sondages enhance or detract from the democratic process.

That very issue was well-canvassed in this 1997 article in the University of Toronto Faculty of Law Review. At this point I would agree with its conclusions: there should be a blackout on published polls for some reasonable period before E-day.

These polls deform the will of the electorate, placing voters in a real-life variant of the prisoner’s dilemma, in which the possible behaviour of one’s fellow-voters becomes a crucial component of many voters’ own decisions. Some simply opt out, thinking their vote isn’t necessary: in yesterday’s election, nearly half of the voters stayed home. Others turn out when they might not have before, trying to save what they perceive as their party’s lagging fortunes. Some (although this wouldn’t have been a factor in the two-way BC race) vote strategically.

The observers, in other words, are directly affecting what they are observing. This negatively affects the democratic process. The unanimous wrongness of pollsters’ findings in Alberta and BC paradoxically reinforces that conclusion.

Shut them down. Your views?