With the American election fast approaching, the campaigning is getting heavy. This has resulted in claims throughout the year that the campaign is rigged, and Google is helping.
So with that in mind, we wanted to find out, could Google actually rig an election? Or, at the very least, influence votes?
If you listen to Robert Epstein, that answer is yes. Epstein coined the term “search engine manipulation effect”, which is the change in consumer preferences as a result of manipulated search results by search engine providers. According to experiments Epstein undertook and published, search engine manipulation effect can shift voting preferences by 20% or more. In some demographics, this sway can be up to 80%. This is a huge swing, when you consider the small margins elections are often won by.
This seems impressive, until you actually question the conditions of the study. The results come from a lab experiment, rather than a field experiment. Participants were asked about unfamiliar political candidates and shown rigged search results. It’s easy to believe what information you’re presented with on Google when that’s all you know about the topic. In reality, participants would have far more information than what they see on Google, and be far more informed. Undecided voters generally avoid political advertising, and would also receive information from social networks. If undecided voters only got their information from Google, perhaps their views would end up skewed.
There are also others accusing Google of favouring a political candidate.
Earlier in the year, SourceFed accused Google of altering its algorithm to bury negative stories about Hillary Clinton. In June 2016, when users searched “Hillary Clinton cri”, Google wouldn’t search for “crime”. Instead, it searches for “crime reform” or “crisis”.
Note: as of October 2016 when you search the term, it still doesn’t come up with crime, but rather:
On Yahoo and Bing, it’s a whole different story:
Because of the difference in search suggestions, it’s easy to see why Google could be accused of trying to hide negative stories about Hillary Clinton.
There are some fairly rational reasons as to why Google isn’t quite siding with one candidate and rigging the election for them:
CNN weighed in on the debate, with a Google spokesperson highlighting that Bing and Yahoo don’t have as complex algorithms as Google, and interpret search queries more literally. Google’s algorithm filters out inaccurate information from the autocomplete. Google also does not show a query that is offensive or disparaging when displayed in conjunction with a person’s name. However, while Google seems to block suggestions of crime, you can still easily receive negative autocomplete suggestions for Hillary Clinton, including “email” and “Benghazi.”
Rhea Drysdale also noted the example is fairly selective, and can therefore seem misleading. SourceFed failed to mention similar autocomplete suggestions happening for Donald Trump. “Donald Trump lawsuit” does not appear as a suggestion when typing in “Donald Trump law”. Instead, Google suggests “laws” “law and order” and “lawyer interview” while Bing mentions lawsuit second, and Yahoo mentions “lawsuit” and “lawsuit history.” Similar suggestions happen when you search for celebrities that aren’t in the election. Suddenly, it doesn’t seem like Google is favouring a candidate. It’s important to ensure information you’re presented with is all in context.
It is hard to claim that Google is being completely neutral. After all, while Google won’t show offensive or disparaging results next to someone’s name, it is deciding what content is and isn’t offensive. As a public company, Google is also one of the biggest supporters of the Clinton campaign, spending millions in political lobbying. Google definitely has the motivation needed to rig an election. But we’re going to need a bit more contextual evidence than SourceFed and Epstein can offer to believe Google is currently rigging an election.