Did Facebook's Secret Experiments on Users Affect the Outcome of the 2012 Election?

Back in July, I wrote about the controversy over Facebook secretly manipulating the news feeds of some users, in what amounted to a huge psychological experiment.  The idea was to tinker with the priority of good or bad news items appearing in the subjects’ feeds, to see if getting hit with a lot of good news up front would improve their mood, or vice versa with those who were fed a diet of bad news.  Facebook insisted that the tinkering was fairly minor – there were only monkeying with the order in which news items appeared, not hiding anything from the users.  Many Facebook users strenuously objected to being manipulated, even subtly, without their knowledge.

This controversial incident wasn’t the first time Facebook and other social-media services have sought to manipulate users by surreptitiously changing their interfaces.  Over the past few years, there have been a number of stories about such experiments, often conducted with the benign express purpose of improving the service for all of its users.  There can be a rather fine line between harmless interface tweaks and user manipulation.

Mathew Ingram at Gigaom writes of another Facebook experiment that might have had a significant effect on the 2010 and 2012 elections… in the estimation of a liberal publication that studied the results, Mother Jones magazine.  It was a variation on the “good news/bad news” manipulation, but in this case, the algorithms of some users were tweaked to make them see more “hard news” items in their feeds, in the months leading up to the 2012 election, to see if it would make them more likely to vote.  

According to Facebook data scientists, it apparently did, or at least users surveyed afterwards claim it did.  “The number who voted (or at least those who said they voted) went from 64 percent to 67 percent,” Ingram writes.  That’s only based on preliminary results, mind you; the full report from Facebook data analysts isn’t due until 2015.  Meanwhile, they don’t seem terribly eager to let anyone know they’re playing these little games – of course, it wouldn’t be much of an experiment if the subjects were informed they were being experimented upon.  The word gets out when tech writers learn about the user manipulations months or years later at seminars, or by reading quietly-released white papers.

The 2010 election manipulation was also subtle, involving the size and placement of a button Facebook users could click to declare, “I’m voting!”  Basically, the larger and more obvious the button, the more likely the user was to vote.  Users who didn’t seen an “I’m voting!” button at all voted less often.

Facebook insists all of this is being done to improve the user experience and discover which features its account holders are most interested in seeing, and they have insisted that neither the 2010 or 2012 manipulations were designed to influence the outcome of specific contests.  It’s not hard to imagine how such user manipulation could be used for such nefarious purposes – how easy would it be to data-mine user information and ensure the most likely voters for certain candidates received an extra-large “I’m voting!” button, and plenty of hard news in their newsfeeds?  And of course, there’s the ongoing controversy among users of all political and apolitical stripes who don’t like the idea of being influenced without their knowledge or consent.

Ingram concludes, “The fact that Facebook can manipulate newsfeed design in ways that can influence voter-turnout rates is fascinating, and perhaps even encouraging — but at the same time, the implications of that are disturbing: what other kinds of behavior could it influence, or actually be influencing even now, without our knowledge?”  

It’s fascinating, all right, but I don’t see much reason to conclude it’s “encouraging.”  If tweaking someone’s Facebook page makes him significantly more likely to vote, then he wasn’t a very serious voter to begin with, and our political system is not well-served by tidal surges of unserious voters, contrary to what all those “get out and vote even if you don’t give a damn” media campaigns assert.  Turnout is not the only desirable indicator of a healthy political system.  Too much of our political debate is coming down to turnout anyway – all that matters is motivating reliable voters to trudge to the polls, rather than seeing to persuade them to vote for one idea or another.  Everything boils down to whipping your voters to the polls like cattle, while making the other candidate’s voters feel depressed enough to stay home.  Our political discourse should be better than that.  If it’s coming down to fooling around with peoples’ web-surfing experience to induce Pavlovian voting reactions, it’s hard to see how it could get much worse.