The head of Google Search denies my claim that Google can control elections--well, sort of; if you read closely, you'll find that he denies nothing.
Google has spoken. Apparently shaken by my recent article in Politico-- "How Google Can Rig the 2016 Election" --Google's head of search, Dr. Amit Singhal, has published a critique--"A Flawed Elections Conspiracy Theory." I encourage you to read it closely.
I emphasize the word closely. On the surface, Dr. Singhal's article appears to refute my recent assertions that Google can easily flip close elections by favoring a candidate in its search rankings. If you read closely, however, you will find that Dr. Singhal has denied none of my specific claims. As I discussed recently in my appearance on "Larry King Now," here are the major points I made in my Politico article, not one of which was refuted in Dr. Singhal's reply, not even in part:
• Randomized, controlled research I have conducted with Ronald E. Robertson, published recently in one of the most prestigious scientific journals in the world--the Proceedings of the National Academy of Sciences (PNAS)--shows conclusively that search rankings that favor one political candidate can easily increase the proportion of undecided voters who favor that candidate by 20 percent or more--up to 80 percent in some demographic groups. The research also shows that this phenomenon, which we call the Search Engine Manipulation Effect (SEME), is almost entirely invisible. Dr. Singhal doesn't discuss the new research report, although there are two links to it in my Politico article. Perhaps he would prefer to pretend it doesn't exist, or at least to have readers forget it exists. Both marketers and psychologists call this form of influence "redirection."
• Google executives could easily flip a close election by adjusting the company's search algorithm to favor one candidate. I didn't accuse any executives of doing this. I simply said it was possible, and Dr. Singhal could hardly deny this, which is presumably why he didn't. That power exists, and, as long as it does, Google poses a serious threat to the democratic system of government. Google executives have more power over elections worldwide than any small group of individuals has ever had in the history of humankind.
• By adjusting the algorithm, even a single rogue employee at Google could flip an election. Again, I didn't accuse anyone of doing so. But I did point out that just a few years ago, a single rogue employee at Google--Marius Milner, who is still employed there--apparently masterminded a massive effort by the company to collect unprotected Wi-Fi data from thousands of homes and businesses around the world for more than four years using the company's Street View vehicles; Google has been fined for this betrayal of public trust, and related litigation is still in progress. In 2012, Google was fined $22.5 million by the Federal Trade Commission after officials learned that software engineers at Google had been hacking into Apple's Safari browser. In 2010, Google fired a software engineer, David Barksdale, for, among other things, spying on at least four underage teens through the company's email system. Perhaps, if he is religious man, Dr. Singhal prays daily that the engineers on his staff aren't fixing elections, but prayers are not always answered favorably. Any of Google's 37,000 employees with the right password authority or hacking skills could use SEME to fix an election. Do Google staff members ever adjust search rankings with the permission of higher-ups? Yes--500 to 600 times a year. Does Dr. Singhal deny any of this? No. Should any employee at any company have the power to flip elections? I think most people would say no.
• Finally--and this is the key point--even without human intervention, Google's algorithm, while doing exactly what it's supposed to do, routinely boosts one candidate higher in search rankings simply because of normal "organic" search activities. Google's own data, which you can link to here, show that search activity favored Obama over McCain on a daily basis for months before the 2008 election and then favored Obama over Romney on a daily basis for months before the 2012 election. The increased activity could easily have been driven by higher search rankings for Obama, and it's also possible that higher search activity pushed Obama higher in rankings. As Robertson and I show mathematically in our paper in PNAS, these two modest phenomena--search rankings boosting voter interest, and voter interest boosting search activity--could work synergistically to produce an explosion of support for one candidate shortly before election day--a digital bandwagon effect. If we could figure out the math, so could many engineers and executives at Google, including Dr. Singhal. (That is not an accusation, by the way; it's a backhanded compliment.) In his article, does Dr. Singhal deny that Google's search engine often ranks some political candidates higher than others? Not at all. How could he? It's the search engine's job to rank some results higher than others. Unless an equal-time rule were programmed into the algorithm, it would be nearly impossible for the search engine not to rank one candidate ahead of another.
As I explained in detail in an article I wrote for TIME in 2013, for most of its existence now, Google, Inc. has relied on a fundamentally deceptive business model for its income. On the surface, Google, Inc. is a bunch of cool geeks who give us lots of free services; below the surface, it's a ruthless advertising company that uses those services to collect information about us and then sells that information to vendors. At Google, we are the product, and every service the company provides is just another front to feed the advertising machine, which accounts for nearly 100 percent of the company's revenue.
After long-time Google executive James Whittaker quit in 2012, he explained that when he first arrived, Google was a super-cool corporate anomaly but that it quickly evolved into "an advertising company with a single corporate-mandated focus." Thanks to devoted, well-compensated employees like Dr. Singhal, the truth about Google remains as cloaked as a Klingon warship.
Dr. Singhal's critique of my article is classic GoogleSpeak, distracting us from what's important--the new research findings and the serious threats the company poses to free and fair elections. As usual, Singhal's statement does little more than emphasize what a cool and giving and socially responsible company Google is.
Google is now facing multi-billion-dollar fines in both Europe and India for biased search rankings--in other words, rankings that favor its own products and services over those of its competitors--and members of the European Parliament and top corporate leaders in Europe are calling openly for regulating or even dismembering the company, just as the U.S. Department of Justice gutted AT&T in the 1980s. No matter what spin you might have heard, that is the most likely reason Google recently created the Alphabet holding company--not to better organize its activities (what nonsense) but to try to protect the value of the stock held by major stockholders if and when the pixels hit the fan. After AT&T was dismantled, the value of the resulting company's stock dropped by 70 percent.
We need to blow away the colorful smoke and shatter the shiny mirrors. Research on SEME, which we have already replicated in three more large multinational experiments conducted after the PNAS article was completed, shows unequivocally that unregulated search rankings pose a serious threat to the democratic system of government. Rankings that favor one candidate for any reason (in other words, even sans "conspiracy") drive massive numbers of votes to that candidate, with no one the wiser and no one able to counter the effect. That's where we need to focus, not on how cool Google is.
If Google were really the socially responsible, do-no-evil company it pretends to be, here is what it would do. Come to think of it, here is what it should do right now:
• It should challenge its employees, perhaps in consultation with top researchers (hint hint), to figure out how to protect vulnerable voters from SEME, perhaps by implementing an "equal time" rule in its search rankings, just as other forms of media use equal time rules to make sure political opponents share stages and air time.
• It should express public concern about the potential harm that SEME can do rather than pretend, absurdly, that ordered lists of search results are innocuous. If they are so innocuous, why are companies spending billions of dollars trying to creep up another notch, and why is the company facing massive fines for favoring its own products and services in those lists?
• It should explain, openly and honestly, how it currently chooses to rank election-related information, and it should make available to investigators internal data that will allow them to determine how candidates worldwide may have been favored in past elections, both in generic search rankings and in rankings customized for individuals.
Meanwhile, our data suggest that the search engine Dr. Singhal oversees is determining the outcomes of upwards of 25 percent of the national elections in countries in which Google dominates search. I'd like to think that if I were in Dr. Singhal's shoes, I'd be concerned about such a possibility; I certainly wouldn't brush it off, no matter how much I was being paid.
A Ph.D. of Harvard University, Robert Epstein is senior research psychologist at the American Institute for Behavioral Research and Technology and the former editor-in-chief of Psychology Today magazine. Follow him on Twitter @DrREpstein. This article is reprinted by permission from Ora.TV, where it first appeared.