Do we WANT to be able to predict an election result?
So what does it do to democracy if we can predict the results of an election with 100% accuracy? Nate Silver’s predictions at the NY Times’ Fivethirtyeight.com election poll blog correctly called the results of 50 out of 50 states in this year’s US elections. In 2008, Mashable writes, he only got 49 out of 50 states (Obama won Indiana by 0.1%). Here’s the side by side comparison Mashable showed us, in this tweet from interaction designer Michael Cosentino:
The ability to accurately predict the results of an election, even a relatively simple two party election as in the US, is quite new. As recently as this summer a blogger for The Economist by the name of “M.D.” wrote that forecasts in general are not very accurate, although “the 2008 election happened to be a good year for the forecast industry, with all 15 forecast models with which I am familiar, save one, predicting Barack Obama’s victory.”
Given Nate Silver’s results this year, I’m guessing that 2008 didn’t just “happen to be” a good year. What’s happening is that we’re getting very, very good at analysing big data. Also, more and more applicable data is available in a format that we can analyse – we’re using Twitter as well as traditional polls.
Interestingly a quick search on Google Scholar found plenty of articles discussing how to make more accurate election forecasts, but I didn’t find anything about whether perfectly accurate election forecasts are something we really want. Nate Silver’s prediction victory is reported in many news outlets (including Norwegian Dagbladet) but the only criticism of the model that I’ve seen is to question its accuracy – please tell me there are people considering what it means for democracy?
What is the point of voting, if we have 99.999% accurate predictions? Is voting an anachronism when we can simply analyse the population as a whole using astounding amounts of data? If predictions match election results perfectly, are they now unbiased? If we know that predictions are extremely accurate, does it change the way we vote, or the kinds of people who turn out to actually vote? Perhaps using prediction software that included the whole population could be more democratic than the current system of actually going to a physical place to vote, which has all kinds of built in exclusion of some kinds of voices.
But it is also easy to imagine a world where presidents are chosen by algorithms analysing the people’s sentiments and opinions. We’ve stopped using old-fashioned voting, because the software is so much more fair. But what happens when the algorithm is tweaked in favour of one of the candidates. So easy to do. Such profound consequences.