Google's Secret Search Algorithm Could Manipulate the Results of an Election
It seems like Google knows us better than we know ourselves. It is the hub of our preferences and a self-referential guide created by our interests and for our interests.
Looking into its fibrous nexus is like looking into a mirror, one which is customized with suggestions and tips crafted to better suit our lives.
The algorithm works so well that it can be manipulated, despite Google’s motto plainly stating: “Don’t be evil.”
Toying with the search engine in order to manipulate user preferences for pithy things like choosing political candidates would seem to violate this mission statement.
Psychologist Robert Epstein has recently did just that. He created a series of experiments with a fabricated search engine called “Kadoodle” which used manipulated search engine rankings. When participants searched for a particular political candidate, the search engine would give an advantage to one of them by moving favorable links to the top of the search and unfavorable ones to the bottom.
According to the study, 75% of subjects showed no awareness of the search engine manipulation and preferences for political candidates shifted monumentally towards the favored candidate. Epstein’s research suggests that the outcomes of real elections, particularly those that begin as heated contests, can be determined or at least significantly influenced by this targeted manipulation of search engine result rankings. The scarier secondary conclusion is that this calculated manipulation can happen unbeknownst to the people searching and the voting public at large.
Internet search rankings have a huge impact on consumer choices because people click on links that are higher up on the result list. A person would be more likely to purchase a used copy of the DVD Airplane! when it shows up on Amazon.com, as opposed to a West Virginian’s Craigslist post.
Corporations have noticed this obvious trend and have injected an estimated $20 billion annually to ensure that their products rank amongst the top of search results. The problem with using this same philosophy in the context of political choice is that it poses a potential threat to the proper functioning of the democratic voting process.
But any results from this actually happening outside of the “Kadoodle” experiment would be correlative and not causal.
General exposure is one of the key ways in which candidates win political office. This equates to television advertisements, publicity campaigns and most importantly personalized emails and advertisements for users who support a given candidate.
A member of a political candidate’s marketing team is not going to have the facility to tacitly hack into the Google algorithm and ensure that no nasty articles appear about the politician in question. But if they could, more power to them.
The general voting public is not going to make their ultimate predilections about candidates based on whether they Google them and an article about a secret child in Malaysia pops up. If anything, it would seem remarkably suspicious if every search result sung the praises of a person running for office. There’s not only dirt about politicians in the nooks and crannies but also on the front porch of the internet.
The “Kadoodle” conundrum elicits a question about government regulation and that perhaps it should have some domain over search engines to ensure that results do not get manipulated. It’s remarkably impractical to suggest that someone would comb over search results for political candidates in order to ascertain whether they were favorable or not. How many favorable articles in a row is too many? What constitutes as favorable? Would they take into consideration the individual composing these pieces of fanfare?
It would also be difficult to take user preferences into account. If you like Cory Booker on Facebook, you’re going to get Cory Booker ads. Regulation of personal internet usage is the absolute last thing America is looking for, especially after PRISM gate.
Epstein’s study is a fascinating glimpse into the influences that favorability and ease of access have on choice. But until and unless search engines fall under the control of anyone besides users, the campaign process will remain as is: predicated on money and influence and blissfully defunct.