Thursday, August 27, 2015

Can Google Rig Democratic Elections? Have They Already?

What influences voters is a central question in Political Science.  It is widely accepted that, to some extent, people's votes are influenced by the media, family, friends, income, education, and much more.  But last week, research psychologist Robert Epstein wrote a controversial piece in Politico detailing how Google trumps them all and could outright "rig" the 2016 election.  He boldly declares, "the Search Engine Manipulation Effect (SEME) turns out to be one of the largest behavioral effects ever discovered" and that it is "a serious threat to the democratic system of government."

Based on data collected in a research study, he asserts that Google's search algorithm - the way Google decides in what order to rank search results for a given term - can easily shift the voting preferences of undecided voters by 20% percent or more, and even up to 80% in some demographic groups - "with virtually no one knowing they are being manipulated".

His logic is based largely on the fact that 50% of the time Google users only click on the first two results, and that 90% of the time they never click beyond the first page of results.  Therefore, if someone searched for the term, "Chris Christie", for example, whether the algorithm listed on the first page negative stories about the Bridgegate scandal or positive stories about New Jersey's improved budget during Christie's tenure as governor, this would influence undecided voters by 20% or more.  And surely, 20% would be enough to swing an election in a candidate's favor.

Epstein even suggests that if campaigns stopped flooding the airwaves with media blitzes that cost a fortune in the weeks before an election, and instead focused simply on finding "the right person at Google" who would tweak the algorithm their way, that would have far more of an effect in turning swing voters.

Afraid yet?  That seems to be the point.  But there are a host of reasons why not to take this conspiracy theory for face value.

First, the point that Google's search algorithm is influential is virtually indisputable, but so what?  Google has for a long time now acted as a gatekeeper, or filter, for what information people ultimately access.  It is not a censor of content, but its rankings basically function as one.  However, it is a huge leap to conclude that "the right person at Google" could decide "which candidate is best for us" and "fiddle with search rankings accordingly".  Outlandish as this may seem, Epstein himself states that this is a "credible scenario under which Google could easily be flipping elections worldwide as you read this".

It doesn't get much more conspiracy-minded than that.

For their part, Google senior vice president Amit Singhal responded directly to Epstein's allegations, stating:
There is absolutely no truth to Epstein’s hypothesis that Google could work secretly to influence election outcomes. Google has never ever re-ranked search results on any topic (including elections) to manipulate user sentiment. Moreover, we do not make any ranking tweaks that are specific to elections or political candidates.

Second, there is the not-so-small matter of causality.  Epstein suggests that undecided voters click on the first few links about a candidate and make their decision who to vote for based on what they see.  However, what he overlooks is that the opposite is also true.  As people surf the web, and blog, and tweet, and link to stories about candidates, they are the ones determining which links will come up first in the search results.  In other words, people are influencing the algorithm as much as, if not more than, the algorithm is influencing them.

In this way, many political activists across the ideological spectrum have long sought to game Google's algorithm to their preferred candidate's advantage.  But vocal activists trying to influence people's votes is hardly a new phenomenon and, when they succeed, it counters the notion of top-down algorithmic control.

Third, Epstein's study recorded undecided voters' preferences after they were exposed to stories listed in search rankings.  But how much staying power did those preferences have?  It would be instructive to know how those undecided voters actually voted on Election Day, and not just who they said they intended to vote for when Election Day eventually rolled around.

Fourth, there is the completely unfounded argument that "Google’s search algorithm, propelled by user activity, has been determining the outcomes of close elections worldwide for years".  That is flatly absurd, unless it is qualified with a statement that so has television, radio, newspapers, and virtually every other form of modern media.

Following that thread, use your own judgment to evaluate Epstein's claim that "it's possible that Google decided the winner of the [2014] Indian election. Google’s own daily data on election-related search activity... showed that Narendra Modi, the ultimate winner, outscored his rivals in search activity by more than 25 percent for sixty-one consecutive days before the final votes were cast. That high volume of search activity could easily have been generated by higher search rankings for Modi."

Is anyone convinced by that?  Isn't it at least possible that higher search activity actually caused the higher search rankings?  (Hint - that's the way the search algorithm actually works.)

Again, no one disputes the influence that Google search rankings have on a whole host of topics.  This is why an entire industry called SEO has come into existence - to game the algorithm for marketing purposes. And, yes, political campaigns have tried, and will continue to try, to game the algorithm to their preferred candidate's benefit.  However, charging Google executives with "rigging" elections (Epstein's wording, not mine) is grossly irresponsible.



Post a Comment

<< Home