Early on Tuesday morning, Donald Trump lobbed the latest stink **** in a conservative campaign accusing Silicon Valley’s tech giants of anti-Republican bias.
In a series of tweets, the president said that Google search results were “RIGGED” against himself and other Republicans, asserting that 96% of the results for “Trump News” came from “National Left-Wing Media”. “Google & others are suppressing voices of Conservatives and hiding information and news that is good,” he wrote. “They are controlling what we can & cannot see.”
It’s an opportune time for Trump and others such as Senator Ted Cruz to target companies like Google, Facebook and Twitter. Non-stop controversy, from fake news to Cambridge Analytica, has made them convenient political punching bags. And these internet giants do have much to answer for, whether it’s their occasionally incorrect censorship decisions or their promotion of conspiracy theories.
Trump is not wrong when he says that Google is “controlling what we can & cannot see”. But he is wrong that the biases in Google’s search algorithms are partisan – or designed to harm his and other Republicans’ reputations. And the insistence on seeing algorithmic bias as stemming from partisan politics obscures the true nature of the various biases that shape our digital realities.
A search engine is fundamentally a giant bias machine that we use to sort and rank the unfiltered web. If Google returned an alphabetical list of every web page that includes a certain search term, no one would use it. Instead, we ask Google to make judgment calls about whether any particular webpage contains information that is pertinent to our needs. To figure this out, Google uses a massive number of signals to return a ranked list of pages that it calculates to be most useful to the searcher.
Although Google is not as transparent as many of us would like about how it makes these ranking decisions, it has published its guidelines, which show, for example, that it considers news sources to have high levels of “expertise, authoritativeness, and trustworthiness” if their articles “contain factually accurate content presented in a way that helps users achieve a better understanding of events” and “have published established editorial policies and robust review processes”.
These are not partisan attributes. Right-leaning news outlets are just as capable (and in most cases willing) to abide by standard journalistic practices as left-leaning outlets are.
The existence of a partisan correlation does not imply partisan causation. Or, put more simply, if a company that openly admits to downranking non-factual news outlets ends up downranking a number of non-factual outlets that often happen to be rightwing, the likely reason is that those outlets are non-factual, not that they are conservative.
This bias toward viewing the actions of tech companies as primarily partisan was also on display in the “shadow banning” debacle, another Trump firestorm that ensued after Vice News reported that Twitter was hiding the tweets and accounts of “prominent Republicans”.
The Vice article accurately observed that a handful of Twitter accounts belonging to Republican party officials were being algorithmically downranked, but it ascribed that downranking to one characteristic of the accounts (the users were Republicans) without any evidence. The much more likely explanation for the downranking was a previously announced algorithm change by which Twitter judges whether an account contributes to so-called conversational “health” based on how other users respond to a user’s tweets.
The desire to view a given phenomenon within the red/blue binary ended up blinding the reporter – and Trump – to what was really happening.
“It does muddy the waters because one of the concerns people like I have is that these platforms do prioritize certain voices,” said Safiya U Noble, a University of Southern California professor whose recent book, Algorithms of Oppression, documents how search engines like Google can reinforce structural inequalities and replicate societal biases, such as racism and sexism.
“The public relates to what’s happening on Google as if it’s a public information news portal, rather than an advertising platform that can optimize content based on the highest bidder,” Noble said.
Noble argues that the much more important bias on internet platforms is toward money and power; those with both can pull any number of levers to affect what is being returned as a top result on Google and what is being written about by the mainstream news media.
She also pointed out the absurdity of the president complaining about feeling powerless before a tech giant: “That’s not the same as people who don’t actually have political power.”
As an example: at the crack of dawn this morning, a certain billionaire was moved to send an intemperate and largely inaccurate tweet. Thanks to his decision, I am now writing this article. That’s power.