If you accessed the Google search engine on Election Day in 2018, you were probably amused and delighted to see the words “Go Vote” where the word “Google” usually is. Perhaps you even said to yourself, “What a great public service Google is performing!”
I, on the other hand, was deeply concerned, because over the six years I’ve been studying Google, I’ve learned that virtually nothing the company shows you is what it seems. The “free” services it provides aren’t free at all. They’re just gussied-up surveillance tools, making you feel good while you give up massive amounts of information the company uses to generate nearly $100 billion a year in advertising revenue.
The information you supply is also shared with Google’s business partners and government agencies, both in the United States and other countries. Worse still, the more information you give up, the less freedom you actually have. More and more, Google is using the information they have about you to influence your thinking, decision-making, behavior, beliefs, attitudes, and purchases.
As any con artist will tell you, and as Google’s leaders know full well, the more you know about people, the easier it is to manipulate them, and Google is currently tracking more than 2 billion people worldwide over more than 200 surveillance platforms.
Worse still, it is using manipulation techniques that have never existed before in human history and that are largely invisible to people. It’s those techniques that I’ve been discovering, studying, and quantifying in my scientific work, and the more I’ve learned, the more concerned I’ve become—especially about the possible impact of these manipulations on our children (I’m the father of five).
I’m going to introduce you to a subtle, yet powerful, vote manipulation that Google used on Election Day (Nov. 6) in 2018. I call this technique the differential demographics effect, or DDE.
That’s a mouthful, I know, but don’t worry; I’m not going to get too technical. I’m just going to give you a few basics—enough, I hope, to help you understand how brilliantly deceptive Google is. In other words, to help you get a feel for what we’re up against.
Strength, Resources, and Reach
I have sometimes imagined that Google’s top executives keep a copy of Machiavelli’s classic “The Prince” on their nightstands. Written in the 1500s, Machiavelli’s timeless guidebook for people obsessed with power advises the ambitious leader to appear, always, to be benevolent in the eyes of the public while always being ruthless behind the scenes.
That’s exactly how Google’s surveillance tools work—the search engine, Gmail, Chrome, YouTube, Android, Google Home, and others—and that’s how DDE works too.
On the surface, the “Go Vote” reminder looks like a public service, but if you’ve followed any of the recent leaks of emails and videos from Google, you know how staunchly liberal Google’s leaders and employees are. More than 90 percent of political donations from Google employees have gone to Democrats since 2004, and Google’s leaders were distraught following Trump’s win.
At a company-wide meeting on Nov. 11, 2016, Ruth Porat, Google’s chief financial officer, was unequivocal in her response. After acknowledging her strong support for Hillary Clinton (whom I also supported, by the way), she added: “Our values are strong. We will fight to protect them, and we will use the great strength and resources and reach we have to continue to advance really important values.”
Would Google display a “Go Vote” reminder to its U.S. users on Election Day—a reminder that would be seen by Americans more than 500 million times that day—if there was the slightest chance that doing so would give more votes to Republicans than to Democrats? It seems unlikely, but how could an apparently benign reminder like this help Democrats?
To understand how this works, you need to look at demographics. As you’ll see, no matter how you cut this cake, Google’s “Go Vote” reminder generates more votes for Democrats than for Republicans. What’s more, the precise number of votes can be calculated in advance.
The Net Gain for Democrats
As of Oct. 28, 2018, 35 percent of voters in the U.S. identified themselves as Democrats, 29 percent as Republicans, and 36 percent as independents. Among independents, 53 percent leaned Democrat and 42 percent leaned Republican. Given this breakdown, what impact would Google’s “Go Vote” reminder have had on the election?
When Facebook sent “Go Vote” reminders to 60.1 million of its users during the 2010 election, that caused an additional 340,000 people to vote that day—0.57 percent of the number of people to whom the reminder was sent.
Because Google is a highly trusted information source, a similar reminder from Google is likely to have a much larger effect than a reminder from Facebook, but let’s be conservative here (so to speak) and assume that Google’s prompt was no more effective than Facebook’s.
A total of 118.5 million people voted in the midterm elections, but Google couldn’t reach them all. Taking age and other factors into account, I estimate that Google would have reached 87.7 million of those voters. Of that number, it’s likely that just under 87.2 million people would have voted without Google’s prompt.
In other words, Google’s “Go Vote” message probably sent about 497,000 people to the polls (0.57 percent of 87.2 million) who otherwise would have stayed home.
Of these, about 174,000 would have voted for Democratic candidates (35 percent of 497,000), and about 144,000 would have voted Republican (29 percent of 497,000). The “demographics differential” here—the net gain for Democrats—was just over 29,800. The net gain for Democrats among independents gives us an additional 19,700 people, so the total net gain for Democrats was 49,500.
That might not sound like much, but remember that Trump’s Electoral College win was made possible by a total margin of only 77,744 votes in three states. Anyway, I’m just getting started.
On average, ballots contain about 17 races. That number is from a 2010 study, but I checked it by sampling recent ballots in 13 states; the average number of races I found per ballot was 17.9—fairly close to the 2010 number.
If our extra 49,500 Democrat-leaning voters voted a straight or nearly straight Democratic ticket, that would have given Democratic candidates upwards of 841,000 additional votes (17 times 49,500). That’s upwards of 841,000 more votes than Republican candidates would have received.
Google Users Lean Left
To save space—and to keep from boring you to death—I’ve left out a couple of details here, the main one being that my calculations reflect the fact that Google users tend, on average, to lean slightly left. In other words, the “Go Vote” prompt didn’t actually reach a cross-section of American voters; it reached Google users, and that boosted the net gain for Democrats.
To see how this works, imagine you are dropping “Go Vote” pamphlets from a helicopter over random locations in a large city in which half the people are Democrats and half are Republicans. Your pamphlets will probably prompt an equal number of people from each party to vote, right? (So why are you even bothering to drop them? Think about that when you’re looking at Google’s “Go Vote” reminder.)
But what if you drop more pamphlets into neighborhoods where people are more likely to favor Democrats—into neighborhoods, say, where more young people live? The more frequently your pamphlets reach demographic groups in which Democrats dominate, the higher the proportion of Democrats you will cause to vote.
That’s likely what happened with Google’s “Go Vote” prompt.
According to a 2013 study of how Google usage varies by state, usage is significantly higher in states often considered to be “blue” (66.5 percent, according to my calculations) than in states often considered to be “red” (60.1 percent), so Google’s prompt may have reached more people in blue states than in red states.
Of greater concern, exit polls conducted on Election Day in 2018 suggested that young voters (18–29) favored Democratic candidates by a whopping 35 percent (67 to 32 percent) and that somewhat older voters (30–44) favored Democratic candidates by 19 percent (58 to 39 percent). Does Google reach more younger people than older people? Probably so. According to a 2015 study, as well as data I’ve collected in my own studies over the years, nearly everyone under 45 uses Google, whereas older users lean a bit more toward Bing and Yahoo.
Messing With Republicans?
Everything I’ve told you so far assumes that Google is both cautious and honest—that it never, ever distorts its content to advance its own interests. The European Union, which has levied massive fines against Google for, among other things, biasing its search results to favor its own products and services, would beg to differ, and so would our own Federal Trade Commission. Leaked emails from the company also show that Google employees talk to each other about burying conservative content; why would they talk about doing so unless they actually could?
In the present instance, a cautious and honest Google would have displayed the “Go Vote” reminder to all U.S. Google users on every search they conducted. But what if the company showed the prompt mainly to Democrats, or what if it showed the prompt more frequently to Democrats—or both?
If Google hid the prompt from a few Republicans, the net gain to Democrats would have been larger than the numbers I’ve shown you—possibly much larger. In this case, the company would have combined DDE with another powerful manipulation I’ve been studying called the targeted messaging effect (TME).
In the most extreme form of this manipulation, Google would have displayed the “Go Vote” prompt exclusively to people who were likely to vote for Democrats. Thanks to our sheep-like willingness to allow the company to track everything we do, Google knows exactly what our voting preferences are. What’s more, in order to better serve our needs (or so they say), the company shows us search results that are customized just for us.
The bottom line: Google could easily have sent “Go Vote” reminders mainly or exclusively to Democrats, and my research suggests that had they done so, virtually no one would have noticed. Even if a handful of people caught on, anecdotal observations aren’t taken seriously. To document targeted messaging of this sort, large-scale monitoring systems would have to be in place, and such systems don’t yet exist.
A highly targeted “Go Vote” prompt wouldn’t have produced a measly net gain of 49,500 Democratic voters; rather, it would have produced a net gain of more than 273,000 Democratic voters, and they, in turn, might have cast more than 4.6 million votes for Democratic candidates (17 times 273,000).
If I can perform calculations like this, so can data analysts at Google—far more accurately than I can. Would Google—quite possibly the most skillful number-cruncher in history—post a “Go Vote” prompt without first running the numbers? Not likely.
Google’s Creepy Line
The most extraordinary thing about this kind of manipulation is how little it costs. Google spent nothing on the “Go Vote” manipulation. It took a salaried programmer no more than a few minutes to change the logo and, perhaps, a bit longer to alter a few parameters to change how the logo was targeted—no cost at all to generate upwards of 4.6 million votes for the political party the company supports. In the world of politics, it doesn’t get sweeter than that.
If you think that Google would never be so rash as to hide a “Go Vote” prompt from a few Republicans, think again. Google’s former CEO, Eric Schmidt, took pride in the fact that the company frequently operated “close to the creepy line.” In 2014, he offered to run Hillary Clinton’s tech team, and in 2015, he set up a secretive company called The Groundwork for exactly that purpose. We learned recently, moreover, that Google encouraged the Hispanic vote in 2016 for the express purpose of driving votes to Clinton.
At times, Google’s search suggestions have also shown blatant political bias. Consider the suggestions the company showed users who typed “Hillary Clinton is” into the Google search box in August 2016.
Bing and Yahoo displayed what people were actually searching for—mainly negative things about Clinton—but Google displayed only “Hillary Clinton is winning” and “Hillary Clinton is awesome,” even though the company’s own data (on Google Trends) revealed that virtually no one was searching for either phrase.
Even Google’s screw-ups tend to favor Democrats. Consider the image the company showed people in July 2016 when they typed “presidential candidates” into the Google search box:
Notice anyone missing? (Hint: He’s our current president.)
I could give you dozens of examples such as this, along with analyses of tens of thousands of election-related search results I preserved in the 2016 and 2018 elections, each analysis showing significant bias favoring Democrats. (I expect to release some of my 2018 findings in March 2019 after they have been through a process of peer review.)
Without whistleblowers, warrants, or large-scale monitoring systems, however, no one can prove that Google displayed the “Go Vote” prompt to help Democrats. Google executives can simply repeat their preposterous claim that the political bias that is rampant at the company has no impact on what Google shows its users. Sadly, most left-leaning politicians and media sources will mindlessly accept the denial no matter how outrageous the numbers and images they see.
In sworn testimony before the House Judiciary Committee a few weeks ago, Sundar Pichai, Google’s CEO, did exactly that. He swept aside the entire bias issue with the carefully worded sentence, “I’m confident we don’t approach our work with any political bias.”
Based on emails and videos that have leaked from Google in recent months, as well as on statistical analyses I’ve performed, controlled studies I’ve conducted, and a large amount of data I’ve gathered in two monitoring projects, I have some doubts about the genuineness of Pichai’s testimony. I suspect, in fact, that Google personnel have long been using a number of powerful methods—some far more powerful than DDE—to shift votes and opinions for political purposes.
If Google did admit the obvious, would it even matter? Existing regulations and laws don’t prohibit companies like Google from using invisible manipulations like DDE and TME—quite the contrary, in fact. Courts have consistently ruled that Google can show users anything at all. The company has, it seems, a First Amendment right to free speech, just as you and I have.
You could conceivably argue that DDE violates campaign contribution laws, but you would probably be laughed out of court if you tried to convince a judge that Google was doing something wrong by patriotically displaying “Go Vote” reminders to millions of people.
To me, the numbers speak for themselves, and they say that Google’s “Go Vote” prompt was a subtle yet powerful vote manipulation, displayed with the express purpose of helping Democratic candidates win their races. If you continue to believe that the prompt was a public service, I suggest, respectfully, that Google has achieved its most ambitious goal: to rob you of your ability to think independently.
Benevolent on the surface, ruthless beneath, Google is rapidly undermining both democracy and human autonomy. Meanwhile, our leaders sit idly by, trapped in partisan prisons of their own making.
Robert Epstein is a senior research psychologist at the American Institute for Behavioral Research and Technology in California. With a doctorate from Harvard University, Epstein is the former editor-in-chief of Psychology Today and has published 15 books and more than 300 articles on internet influence and other topics. He is currently working on a book called “Technoslavery: Invisible Influence in the Internet Age and Beyond.” His research is featured in a new documentary called “The Creepy Line.” Follow him on Twitter @DrREpstein.
Views expressed in this article are the opinions of the author and do not necessarily reflect the views of The Epoch Times.