Google could influence Indian poll results, says study
13 May 2014
Altering Google search results can pose a real threat to democracy as it has a major impact on the voting preferences of undecided voters and could swing a close election, according to what could be a landmark study by US researchers focusing specifically on India's just-concluded general elections.
The study conducted in India in recent weeks suggests that Google has the power to fix elections ''without anyone being the wiser'' because of the power that search rankings have on people's opinions, the researchers said.
The studies show that the higher the 'rank' on Google, the more people trust the result - which is why companies are spending billions now to push their products higher.
Ahead of the results to be declared on Friday, perhaps some of the 'Modi wave' shown by exit polls on Monday can be attributed to these findings.
The study was conducted over recent weeks with more than 2,000 undecided voters across India participating.
The researchers have shown that votes can easily be pushed towards one candidate or another by about 12 per cent; and double of that in some demographic groups – enough to determine the outcomes of many close races.
In earlier research conducted last year in the US, it was found that search rankings biased in favour of a candidate could push the preferences of undecided voters towards that candidate by 15 per cent or more.
''This is a very serious matter – a real threat to democracy,'' said Dr Robert Epstein, lead researcher in the study and Senior Research Psychologist at the American Institute for Behavioural Research and Technology in California.
''If two candidates were both trying to push their rankings higher, they would be competing, and that's fine. But if Google, which has a monopoly on search in India, were to favour one candidate, it could easily put that candidate in office by manipulating search rankings, and no one could counter what they were doing.
''Even if without human intervention the company's search algorithm favoured one candidate, thousands of votes would still be driven to that candidate.''
Arivind Kejriwal, leader of the fledgling Aam Admi Party, was used as an unwitting guinea pig for the study. The participants were randomly assigned to groups in which simulated search rankings 'favoured' either Kejriwal, the Congress party's Rahul Gandhi, or the Bharatiya Janata Party's Narendra Modi.
Real search rankings and web pages were used, and people were asked to research all the candidates just as they would on Google. The only difference between the groups was the order in which the search results were displayed.
The results suggest that search rankings can be used to fix the outcome of races in India in which the winner is projected to win by a margin of up to 2.9 per cent. This can be done just by influencing undecided voters who use the internet – a small but important group of voters that is sure to grow in coming years.
Worldwide, the researchers said, upwards of 25 per cent of national elections are won by margins of under 3 per cent.
The study also shows that certain demographic groups are especially vulnerable. The voting preferences of 19 per cent of women over 35 were shifted in the study, as were the voting preferences of 18 per cent of voters who were unemployed.
''Of particular concern is the fact that 99 per cent of the people in our study seemed to be unaware that the search rankings they saw were biased. That means Google has the power to manipulate elections without anyone suspecting they're doing so,'' said Epstein.
''To prevent undue influence, election-related search rankings need to be regulated and monitored, as well as subjected to equal-time rules,'' Epstein said.