Recently, a study on online rankings’ effect on voters’ habits was conducted with a sample group representative of the U.S. voting population in terms of age, race, political affiliation and other traits, to decide who Californians would have voted for in a 2010 election. To maintain social objectivity, the election chosen was for the prime minister of Australia.
ONLINE REPRESENTATION WARPS PUBLIC PERCEPTION
“What we’re talking about here is a means of mind control on a massive scale that there is no precedent for in human history,” claims Robert Epstein, a psychologist at the American Institute for Behavioral Research in Vista, California. He argues further that the higher a politician ranks on a page of internet search results, the more likely a voter is to give them their full support and vote.
“I have a lot of faith in the methods they’ve used, and I think it’s a very rigorously conducted study,” adds Nicholas Diakopoulos, a computer scientist at the University of Maryland, College Park, an uninvolved yet interested party. “I don’t think that they’ve overstated their claims.”
To test these claims, Epstein and colleagues built a pseudo-search-engine named Kadoodle, which returned a list of 30 websites for finalist candidates; 15 for Tony Abbott and 15 for Julia Gillard. The majority of Californians knew very little about either candidate before the test began (as was intended by selecting Australian politics), but the volunteering voters were not told that the search engine was already rigged to display the voting results in an order biased toward one or another candidate, e.g., a volunteer subject would see 15 webpages with info about Gillard’s platform and administration programs, followed by 15 likewise results of Abbott’s.
The results showed precisely what the scientists predicted; namely, that subjects spend far more time reading web pages closer to the top of the list than they do any other. More interesting than this are the effects the difference in artificial rankings made. The search engine listing one candidate over all others swayed the number of undecided voters into voting for the hierarchically favored candidate by 48%. This was contrasted with another group of voters who chose nominally across an unbiased search engine. Strangely, the few who did realize they were being manipulated were actually more likely to vote in agreement with the biased listings. “We expect the search engine to be making wise choices,” Epstein says, “[w]hat they’re saying is, ‘Well yes, I see the bias and that’s telling me…the search engine is doing its job.’”
INCREASING SAMPLE SIZE
The second experiment used the same process on 2,100 participants brought into the test through Amazon’s labor crowdsourcing site Mechanical Turk. Subjects were again chosen according to racial, religious, ideological positions representative of the U.S. voting population. This much larger data pool allowed the researchers to be deduce which demographics were the most susceptible to search engine manipulation. Divorcees, republicans, and subjects with a low understanding or familiarity with the candidates were the most malleable, and the better informed, married or those with a $40,000 to $50,000 annual income were more difficult to sway or corrupt.
The most malleable and corruptible group were moderate republicans. May the reader’s complete lack of surprise match the writer’s.
Altered search results increased the quantity of undecided voters espousing to choose the favored candidate by 80%.
“In a two-person race, a candidate can only count on getting half of the uncommitted votes, which is worthless. With the help of biased search rankings, a candidate might be able to get 90% of the uncommitted votes [from specified demographics],” explained Epstein.