Are internet platforms suppressing speech and hurting civic culture? Google’s critics could build a case for this claim on a censorship-friendly search app for Chinese citizens, YouTube’s confusing demonetization of news-heavy YouTube channels, or Google’s work on military AI. Critics of Facebook could bring up the massive Cambridge Analytica scandal or its failure to spot foreign misinformation campaigns. Both groups could mention Google and Facebook’s connections to covert surveillance programs like PRISM.
These are all great examples of things you won’t find in The Creepy Line, a new documentary claiming to expose dark secrets behind two of the world’s biggest tech empires.
The Creepy Line is written and directed by the filmmakers behind conservative documentary Clinton Cash. It takes its title from Google executive Eric Schmidt, who said in 2010 that Google’s job was to “get right up to the creepy line and not cross it.” Schmidt has gotten no end of flack for that quote, and it’s easy to point out that Google is, in fact, often very creepy. But The Creepy Line makes a more specific, partisan argument. It claims that Google (and Facebook, which the film refers to almost interchangeably) deliberately manipulates its service to suppress conservative users and ideas, and — more ambitiously — that Google tweaked its search algorithm to swing the 2016 election in Hillary Clinton’s favor.
In other words, the film epitomizes a popular claim that Silicon Valley is censoring American conservatives on web platforms. This claim plays on bipartisan fears about the power of internet gatekeepers, and it’s sparked several lawsuits, multiple congressional hearings, and a few tweetstorms from President Donald Trump. But the evidence has been mostly speculative or anecdotal, and The Creepy Line is supposed to provide a rigorous, scientific analysis to back it up. Instead, it’s a blinkered and misleading guide to how internet platforms work.
The Creepy Line’s core reasoning is that since Google has been caught favoring its own commercial services by antitrust regulators, and its leadership is liberal — Eric Schmidt worked with the Clinton campaign in 2016 — Google is probably playing favorites in politics, too. (In Facebook’s case, the filmmakers point to a well-known 2016 report of anti-conservative bias in Facebook’s human-run “trending” topics — although, again, the film treats Google and Facebook as a nearly single entity.) “If you have the power to change an election, and you view the other person as terrible, would you use it? I mean, who wouldn’t?” said director M.A. Taylor in an interview with The Verge.
This theory has been fueled by news about political discussions within companies — like Google’s internal controversy over employee James Damore, who was fired after circulating a memo criticizing the company’s gender diversity efforts. Breitbart recently published a leaked video from a Google meeting just after the 2016 election of Donald Trump, which includes co-founder Sergey Brin calling Trump’s presidency “deeply offensive.” But that’s not direct evidence that the services are engineered to promote their creators’ politics.
Instead, The Creepy Line hinges mostly on research from psychologist Robert Epstein, who published a paper last year alleging that Google placed pages with a pro-Clinton bias in the top 10 results for a “wide range of election-related search terms.” While Google dismissed the study as a “poorly constructed conspiracy theory,” Epstein’s work has received positive mainstream attention, and he’s been profiled by outlets including The Washington Post and The Outline.
Epstein’s documentary sections bafflingly obfuscate how Google search works, though. He asserts that a query like “What’s the best dog food?” makes Google “look up all the dog foods it has in its database and all the websites associated with those dog foods,” and then “put them into an order” — with, say, “Purina first, and some other company second.”
You can generously interpret this as a convoluted description of identifying third-party pages that mention keywords related to dog food, including the kind of “Best dog foods of 2018” lists that Google actually turns up. But it strongly implies that Google will deliver an independently ranked list of dog food brands, so when Epstein extends this analogy to presidential candidates, it sounds as though Google is ranking politicians for users who look up ballot information, with Hillary Clinton in place of Purina.
Epstein defended his description in an interview with The Verge, saying search engines are inherently built around ranking results based on preferences. But The Creepy Line conflates search algorithms’ well-established bias regarding things like page formatting styles or keyword placements with an independent and coherent political bias toward specific causes or candidates. Eventually, the argument becomes almost paradoxical, with subjects castigating Google (and Facebook, by vague extension) for not un-biasing something that they also argue is literally defined by bias.
That doesn’t mean Epstein’s research is invalid, and other researchers have also pinpointed ways that Google’s rankings can deliver politically biased results. But those biases are widely variable and seem to reflect how well users’ search keywords match rhetoric that’s more popular with one political side than another. In searches for a scandal involving government funds, for instance, adding the dollar amount brought up more conservative-leaning stories, which were especially inclined to emphasize just how much taxpayer money had been wasted. In other words, Google’s political biases look a lot more complicated than a straightforward election-rigging attempt would be.
At least Epstein’s arguments are delivered in detail by a first-hand source. Otherwise, The Creepy Line is stitched from cursory summaries of news items illustrated by repetitive strings of scary headline snippets, at least a couple of which don’t refer to the incident being discussed. Most of the substantive expert commentary is cribbed from public speeches by technology critics like Jaron Lanier and Tristan Harris. The Creepy Line’s filmmakers found a high-profile star in psychology professor and self-help guru Jordan Peterson, who talks about having his Google account briefly disabled for unclear reasons last year. His other commentary, however, is mostly meandering observations like “it’s an interesting word, creepy, because it’s a word that connotes horror.”
The film devotes time to legitimately creepy instances of Google and Facebook behaving badly, like a Google smart speaker bug that made it randomly record conversations or a controversial Facebook mood study that stacked some users’ News Feeds with positive or negative posts. It embellishes the truth here, claiming without evidence that “it appears” some young people may have “done harm to themselves” based on Facebook’s experiment — possibly based on an academic blog post about the case, which warned about hypothetical self-harm to make the point that nobody knew the study’s effects. Even so, it’s talking about good-faith concerns raised by real antitrust and privacy experts.
But The Creepy Line uses these cases only to establish a narrow precedent for anti-conservative censorship. It conspicuously avoids creepy line-crossings that affect other social or political groups, even when those cases would back up its argument that Google and Facebook are dangerous.
The Cambridge Analytica scandal could help demonstrate the problems with Facebook’s personal data collection, but it involves a company linked to Trump’s presidential campaign. Facial recognition provides one of the clearest real-world examples of algorithmic bias, and both Google and Facebook are invested in the field — but the bias is against dark skin, not political sentiment. The film gives Jordan Peterson space to relitigate a legal dispute with the Canadian government over gender identity, but it doesn’t mention that many LGBT YouTube vloggers have also claimed they’re being censored by Google.
The film’s political bent sometimes sends it in bizarre directions, like a section that criticizes social networks for fighting “fake news” disparaging Hillary Clinton, illustrated for no clear reason by stock footage of liberal personalities Rachel Maddow and Kathy Griffin making funny faces. Its conclusion is that fake news doesn’t matter because it only makes readers believe terrible false rumors about people they already dislike, which has been, to be clear, an explicit goal of successful propaganda operations throughout history.
Taylor and the film’s writer Peter Schweizer say that they’re aware of issues affecting non-conservatives and simply wanted to focus on a single thread of the conversation. But that focus completely changes the argument. If conservatives are disproportionately suffering on Google and Facebook, an incident like Peterson getting locked out of Gmail looks like political warfare. If people across the social and political spectra are getting accounts suspended, YouTube videos demonetized, or stories de-ranked, it looks like bad service from a platform with a legitimately disconcerting amount of power.
This distinction is probably obvious to many people. But The Creepy Line suggests that any seeming technical glitch or bad decision is a coordinated step in the master plan of Silicon Valley’s digital “kingmakers,” even when that’s far from the most obvious conclusion. It does things like insinuate that Google’s lack of a customer support hotline is inexplicable and suspicious, or claim that “very few people in the world have ever seen” an error page for google.com, so spotting one indicates that Google has cut off search access to punish a user for criticizing it. It glosses over the fact that these kingmakers didn’t even get their favored candidate into the White House, although Taylor argues that Google only put “roughly 10 percent effort” into its manipulation efforts because it was confident Clinton was already winning. “That won’t happen again,” he says.
The kingmaker theory makes policing Google and Facebook’s algorithms sound easy because it suggests that everything the companies do is deliberate. Epstein, for instance, suggests search engines should implement an “equal time” policy based on the rules for TV and radio stations — except that where stations simply have to offer political candidates equal access to advertising slots, Google would change its search results to make sure it presented equally positive and negative results for related search terms.
I asked Epstein whether this could make it harder to get an accurate impression of a situation. In a political race where one candidate was credibly revealed to be a pedophile, for instance, should Google pad the top search results with positive stories about them? If users search for information about a hurricane, and most results suggest it’s dangerous, should Google inject other stories telling readers not to worry? “If you’re starting to think about truth, then you have to ask a completely different question — which is, do we really want a private company telling us what’s true?” he countered.
Elsewhere, The Creepy Line offers only vague guidelines for how regulation might work, though it raises the possibility of breaking up tech companies and repeats a false claim that Section 230 of the Communications Decency Act only provides legal protections for “neutral platforms.” Schweizer argued in an interview that biased services should be legally defined as media companies because “there are serious regulatory burdens that media companies face,” apparently citing rules that apply to licensed TV and radio networks.
“Google doesn’t have any of those [burdens],” he said. “They don’t have any restrictions on market dominance, they can engage in anti-competitive behavior against Yelp, against others for their own products. Those sorts of things are very different in the media space.” (Tech companies and media companies are both barred from anti-competitive behavior, and numerous American lawmakers have discussed potential antitrust complaints against Facebook and Google.)
The Creepy Line can exist because there’s a lot we don’t know about Facebook and Google’s inner operations and because the companies’ services are incredibly useful and important. The film’s most compelling moment belongs to Jordan Peterson, solemnly describing the effect of losing access to Gmail: “You come to rely on these things, and when the plug is pulled suddenly, that puts a big hole in your life.”
But the evidence we have suggests something arguably scarier than a political rampage: that Google and Facebook (among other companies) have created unpredictable sorting systems that are too sweeping and complicated to fully control, designed surveillance and data-mining tools without considering their dangers, and achieved a level of ubiquity that makes escaping their orbit nearly impossible. Despite its dark name, The Creepy Line appeals to the comforting logic of conspiracy: when something goes wrong in life, it’s because an all-powerful entity thinks you’re important enough to attack.
The Creepy Line premiered in New York City this week; full release details have not been announced.