How You’ll Know When the “Facebook Election” Happens.

TL;DR: When sentiment predicts votes as well as polls.

The weekend after any election, the postgame analysis starts flying. And, in digital politics the one thing more certain than death or taxes is that somebody, somewhere, will be making overbroad claims about what social media or digital can or can’t do.

And, honestly, people have been talking about the internet and elections for decades, and some folks would flag Jesse Ventura’s win in the 1998 Minnesota gubernatorial race as the first real win for the internet.

This year, on the Sunday after a chaotic midterm, Ben Smith posted, “The Facebook Election” to discuss their partnership with Facebook, to report on sentiment analysis in relation to the 2016 election. And, honestly, I look forward to seeing what kind of reporting that produces. I am pretty sure it’s going to be better stuff than when CNN cuts to their random millennial correspondent to discuss what’s happening online in relation to whatever story they’re following at the moment.

But the bigger question that political professionals is how to integrate this online chatter into our polling. And surely that time is nearly here. But the emphasis is on the word, “nearly.”

The big, gaping hole in “The Facebook Election” is that they immediately start reporting on the breakdown of positive and negative chatter about 2016 presidential campaigns, but say absolutely nothing about how this methodology would compare to ACTUAL 2014 voting results.

And I’d have to imagine this idea had already occurred to them — but I’d suspect the answer is, “it doesn’t match up very well at all.” And I say that based on my previous research here at Trilogy Interactive with my colleague Will Bunnett. We’ve not found a lot of predictive value in either online followers and vote totals, nor Twitter sentiment analysis and the results of the Iowa Caucuses.

That was a long time ago, and Facebook is getting almost universal adoption among the voting age population, but that doesn’t automatically mean that analyzing posts and comments are any more predictive than a simple poll. And until that’s proven, we need to slow our rhetorical roll.

The internet is a hugely important part of campaigning today, and I wouldn’t stay in this profession if I wasn’t 100% convinced that the money my clients spend online isn’t the best possible use of their limited resources.

However, that doesn’t necessarily mean that it’s a good idea to violate the first rule of the internet, namely “NEVER READ THE COMMENTS”, and try to mine that data to predict anything other than that people can be complete jerks.

I’d love to you hear your thoughts. You can find me tweeting about online politics, whisky, and cephalopods at @SteveOlson — and if you liked this post, I’d appreciate you clicking the “recommend” button below. Thanks!