Tuesday’s midterm elections in the US were unusual in a great many respects. Usually a snooze-fest for all but the political geeks, this time around the amount of vitriol, money, and just sheer energy spent were historic. The interesting question is how any of this actually changed the outcome of the election, which in the end was a noticeably milder version of the well-established historical pattern for midterm elections.
In fact, throughout American history, midterm elections have virtually had the same effect: flipping control of the House of Representatives to the opposing party, regardless of which party that happened to be. The fact that the outcome is so very predictable is one more reason that there is often little widespread public interest in midterm elections. But this time was different.
Record amounts were spent on the campaign overall, and on social media, which anyway went crazy about the election. There was no shortage of bots, nor of accusations of accounts being bots. Predictions and interpretations in every forum imaginable seemed to increasingly diverge based on party affiliation, to hallucinogenic levels. To put this in some concrete perspective, within living memory, $70M might be spent on a presidential campaign; in this midterm election, that amount was spent on a run for a senate seat in Texas – and that attempt was unsuccessful. As Ed Rogers noted in his column in the Washington Post, all of most media-visible and well-funded “blue wave” candidates were unsuccessful.
Given the increasing political polarization of the country, it seems logical that relatively few hearts or minds were changeable – or, if you prefer, manipulable. In the eagerness of the big Internet companies and their adherents to justify the value of their “likes” and “retweets” no attention seems to have been given to the question of objectively measuring influence, or the changing of opinions. Lots of people seeing lots of ads does not equate to lots of influence – or perhaps even any. If I don’t like the taste of Coke, no amount of ads for it will make me want one. I may think a particular ad campaign or brand effort is clever, but that doesn’t make me want to reach for a Coke. I suppose it is even possible that a Coke ad might make me thirsty – but not for a Coke.
Measuring influence is a meaningful way – the ability of a message to move people from one position to another in a non-ephemeral way – is quite difficult but technically possible. It requires longitudinal analysis, both in text analytics and social network analysis. Plus the ability to reliably identify “suspect” identities such as bots and trolls, lest they introduce a significant source of error. Such measurements would also bring with them accountability, which in many cases is doubtless unwelcome. In the end, perhaps the most interesting question lingering from the midterm elections is when measuring influence will become a matter of course.