Inside Google’s Civil War

To many observers, Google’s tech workforce—notoriously well-paid and pampered with perks—hardly seems in a position to complain. And it’s a surprising tune to hear from employees of one of the titans of Silicon Valley, a place that has long worshiped at the altar of meritocracy and utopian techno-futurism. But in the past few years, the industry’s de facto mission statement—change the world (and make money doing it!)—has been called into question as examples of tech’s destructive power multiply, from election interference to toxicity on social media platforms to privacy breaches to tech addiction.

By Beth Kowitt for FORTUNE

It started in Tokyo on Nov. 1, 2018, when 100 employees walked out of Google’s office at 11:10 a.m. local time. Thirteen hours later, the elevators at the company’s New York City headquarters were so packed that workers took the stairs down to the street to protest. Google employees in Austin observed two minutes of silence for victims of sexual assault as part of their demonstration. In San Francisco, hundreds of employees gathered across from the historic Ferry Building and chanted “Time’s Up at Google” and held signs with slogans like “Workers’ Rights Are Women’s Rights” and “Free Food ≠ Safe Space.”

After Googlers in Sydney walked out, 25 hours after Asia had kicked things off, 20,000 Google employees in 50 cities around the world had joined their colleagues to protest the company’s handling of sexual harassment.

The spark that ignited the walkout was a New York Times article that had appeared a week earlier, reporting that Google paid former executive Andy Rubin a $90 million exit package, despite facing a sexual misconduct accusation Google deemed credible. (In a statement to the Times, Rubin said the story contained “numerous inaccuracies about my employment.”)

It was the first time the world had seen such a massive worker protest erupt out of one of the giants of the technology industry—and certainly the first time outsiders got a glimpse at the depth of anger and frustration felt by some Google employees. But inside the Googleplex, the fuel that fed the walkout had been collecting for months. Tensions had been on the rise as employees clashed with management over allegations of controversial business decisions made in secret, treatment of marginalized groups of employees, and harassment and trolling of workers on the company’s internal platforms. “It’s the U.S. culture war playing out at micro-scale,” says Colin McMillen, an engineer who left the company in February.

To many observers, the tech workforce—notoriously well-paid and pampered with perks—hardly seems in a position to complain. And it’s a surprising tune to hear from employees of one of the titans of Silicon Valley, a place that has long worshipped at the altar of meritocracy and utopian techno-futurism. But in the past few years, the industry’s de facto mission statement—change the world (and make money doing it!)—has been called into question as examples of tech’s destructive power multiply, from election interference to toxicity on social media platforms to privacy breaches to tech addiction. No one is closer to tech’s growing might, as well as its ethical quandaries, than the employees who help create it. “People are beginning to say, ‘I don’t want to be complicit in this,’ ” says Meredith Whittaker, who leads Google’s Open Research group and is one of the walkout organizers. Workers are beginning to take responsibility, she says: “I don’t see many other structures in place right now that are checking tech power.”

As the so-called techlash has cast a pall over the entire sector, organized employee pushback is slowly becoming part of the landscape: Amazon workers are demanding more action from the company on battling climate change; at Microsoft, employees say they don’t want to build technology for warfare; at Salesforce, a group has lobbied management to end its work with the U.S. Customs and Border Protection agency. Meanwhile, there’s not a company in the sector that isn’t grappling at some level with the ways bro-gramming culture has made tech a toxic space for women and employees of color.

But nowhere has the furor been as loud, as public, and as insistent as it has been at Google. That’s no surprise to Silicon Valley insiders, who say Google was purpose-built to amplify employee voices. With its “Don’t be evil” mantra, Google was a central player in creating the rosy optimism of the tech boom. “It has very consciously cultivated this image,” says Terry Winograd, a professor emeritus of computer science at Stanford who was Google cofounder Larry Page’s grad school adviser and would go on to serve on the company’s technical advisory board. “It makes them much more prone to this kind of uprising.” Page, now 46, and cofounder Sergey Brin, 45, intentionally created a culture that encouraged the questioning of authority and the status quo, famously writing in their 2004 IPO letter that Google was not a conventional company and did not intend to become one.

Some workers say Google’s promise to remain unconventional is in question. Interviews with 32 current and former employees revealed a demarcation between what several called “Old Google” and “New Google.” Whether there’s a clear-cut line between these eras—the company got its start in a Menlo Park, Calif., garage in 1998, when Page and Brin were still Ph.D. students at Stanford—depends on whom you ask. But there is a pattern in how they describe the change: At Old Google, employees say they had a voice in how the company was run. At New Google, the communication and trust between the rank and file and executives is in decline. Decision-making power, some say, is now concentrated at the very top of a company run by executives who are increasingly driven by conventional business metrics.

Now Google finds itself in the awkward position of trying to temper the radical culture that it spent the past 20 years stoking. Boasting more than 100,000 employees between Google and its parent company, Alphabet, executives acknowledge that the company is struggling to balance its size with maintenance of the principles, like employee voice, that were so foundational. “You can’t go through that kind of growth without the culture needing to evolve,” says Jen Fitzpatrick, a Google SVP and a member of CEO Sundar Pichai’s leadership team. (Pichai declined Fortune’s requests for an interview.) The company says it is trying to manage its ballooning diversity of perspectives and projects, as well as do a better job predicting the kinds of issues for which employees will demand full transparency. However, it adds that the activist employees are a small but vocal group, and that their opinions don’t represent those of employees at large.

“Twenty-eighteen was a different year for us—the magnitude and the nature of some of these issues is just different,” says Brian Welle, VP of people analytics at Google. The tumult was reflected in the results of the annual companywide Googlegeist survey, which was leaked to the press in February. Key metrics were down double-digit percentage points over 2017. For instance, while 74% of respondents said they had confidence in Pichai and the management team, that’s an 18 percentage point drop from the previous year.

Most challenging to Google is employees’ refusal to keep their discontent within the company’s walls, a strategy that’s been bolstered by activists’ sophisticated use of the media and the world’s fascination with the iconic company. The scene that played out at the walkout was, on one level, as familiar as a factory strike—a ­labor force flexing its collective power to send a message to The Man (in this case, CEO Pichai). But even as activists inside Google are relying on traditional labor organizing tactics, their demands are not just the typical wage or benefits ask. It’s about much more than a paycheck; employees, it’s clear, want a say in and control over the products they build.

Google has already transformed so many aspects of the way we work today. The walkout was an inflection point, a sign that the company is now poised to disrupt something even more foundational to our economic system: the relationship between labor and capital. It’s a shift that could perhaps begin only in Silicon Valley, a place that has long believed itself above such traditional business concerns—and, more to the point, only at this company, one that hired and retained employees on the premise of do no evil. Now employees seem determined to view that manifesto through their own lens and apply it without compromise, even at the cost of the company’s growth. “Who decides what is the soul of Google and what Google is?” asks Lokman Tsui, formerly Google’s go-to executive on issues of free expression and censorship in Asia and the Pacific. “Is it leadership or employees? There’s a real battle for the soul of these companies right now.”

Google’s broad mission of organizing the world’s information and making it more accessible has led the company to digitize books, mount cameras on the top of cars in order to map the world through images, and design virtual reality viewers made of cardboard.

But as the company has grown ever larger, so have its ambitions. In 2018, as Google employees found out about two new secretive projects that were underway, some questioned whether the tech giant had stretched too far beyond the bounds of its mandate in the name of expansion.

The first was the Pentagon’s Project Maven, which uses artificial intelligence to help analyze drone footage. Google became a subcontractor to the Department of Defense for Maven in 2017, but most people inside the company didn’t learn about it until the following year, when an employee wrote an unsanctioned post about the clandestine project on Google’s internal social media platform. Executives told worried employees that Maven was defensive rather than offensive. Still, some workers were concerned that Google’s technology could ultimately be used to make drone strikes more lethal, and that Maven would lead to additional deals between Google and the military. What’s more, some say management’s argument that the contract was in support of “our” military did not always resonate with a global workforce.

For Laura Nolan, then a Google engineer working in Ireland, “It was such a betrayal,” she says. “We’re pretending to be a happy company that does lovely information organizing, and then you’re building several steps toward killer drones flying around.” Nolan, who says her work would have enabled future stages of Maven, quit the company over it. Employees like Nolan didn’t expect Google to be a defense contractor like Raytheon—or even like ­Amazon, which has been open and unapologetic about working with the military.

Even before the bulk of the company learned about Maven, several senior engineers were escalating their concerns internally. Once Maven became more widely known, the resistance spread, with a group of employees writing a letter to Pichai asking that he cancel the project. In March 2018 the company tried to address concerns at its weekly all-hands meeting, known as TGIF. The gathering has been core to Google’s culture since its early days, in large part because it gives anyone the chance to question senior management. At the meeting, an employee told executives she used to work for the Department of Defense but left to avoid contributing to military technology. What, she asked, were her avenues for letting management know this was not okay? The fact that you can ask that question here is a powerful voice, Brin told her. At some companies this would have been a sufficient answer. At Google it was not.

Management continued to put together forums to try to address employee concerns and explain why they believed Maven was a worthwhile project, holding three town halls to discuss the ethics of A.I.

A group of organizers kept up the pressure, making sure there was a Maven question every week at TGIF. They tracked the number of employees who quit over the issue, handed out stickers, and made mocking memes about Maven on Google’s internal meme creator. The debate turned public in April 2018 when the original letter sent to Pichai, which would eventually garner nearly 5,000 employee signatures, was leaked to the New York Times.

In June, Google announced that it would not renew its contract for Maven and released a set of A.I. principles laying out guidelines for the future of the technology—including a vow not to use it to create weapons. Most of the employee activists viewed the announcement as a win, but speaking at a Times conference later that year, Pichai played down the influence of the internal pressures. “We don’t run the company by referendum,” he said. He explained that he had listened to people actually working on building A.I. in making the decision. He stressed, however, that the company continued to do work with the military in areas like cybersecurity.

Then, in August, just as the tensions over Maven were beginning to dissolve, The Intercept published a story revealing that Google was working on a censored search engine for China—code-named Dragonfly—that would block information related to topics like human rights and democracy. For most employees, this was the first they had heard of it. (Google says the project was exploratory and was therefore still confidential.)

Jack Poulson says he was the sixth or seventh employee to cite Dragonfly as a reason for quitting. “It was crossing a line for what it was I felt I wanted to do with my life,” says Poulson, who was a senior research scientist at Google. “I was literally profiting from a company suppressing political speech.” When, the following month, the U.S. Senate’s Commerce Committee called on Google’s chief privacy officer to testify at a hearing about data privacy, Poulson sent his own letter to the committee: “I am part of a growing movement in the tech industry advocating for more transparency, oversight, and accountability for the systems we build.”

Google had previously operated a search engine in China but pulled out in 2010 after the company got hacked. At the time, management had taken what some viewed as a moral stand, with Brin saying he saw “earmarks of totalitarianism” in the country. With Dragonfly, some employees supported the return. But for those who described the 2010 decision as a defining moment for Google’s culture, the reversal was galling. “I wondered what the heck had changed in the eight years since then,” says McMillen.

Pichai was asked that question at the New York Times conference. His response: “Our mission is to serve everyone in the world. As part of that, it’s natural we would think about users in China as well.” He added that Dragonfly was an experiment, and “nothing was imminent.”

Then a new employee, McMillen recalls the company’s 2010 decision to pull out of search in China as foundational—the literal embodiment of Google’s “don’t be evil” ethos. “As part of the perks, Google offered you the self-satisfaction of doing good in the world,” says Whittaker, who was involved in the employee resistance to both Maven and Dragonfly. “That was profound for a lot of people.” Paul Buchheit, a onetime Google engineer who’s credited with coining the mantra in the early 2000s, says “Don’t be evil” was not a magical, black-and-white standard. It was a way to pause and be reflective about the work. How did the company decide whether a given project met the criteria? “Any arbitrary employee was empowered to ask,” he says.

Because Dragonfly began in secret, some employees believed they’d been robbed of that opportunity. Nor were they convinced that Google management had asked itself the hard questions. “There was never any communication that they had thought through the ethical ramifications,” says McMillen. Workers should be able to make their own well-informed ethical decisions about giving their labor to Google, he says. Some workers indirectly involved in Dragonfly hadn’t even known what they were working on. “What are Google’s red lines around censorship and surveillance?” asks Poulson. “I researched this as much as I could as an employee and still didn’t know.”

While Maven, Dragonfly, and even the Rubin payout that gave rise to the walkout angered employees for different reasons, there’s at least one connecting thread: secrecy. The company that was built around the value of information sharing had hit a threshold where a growing number of decisions were made behind closed doors. “We’ve always had confidential projects as a company,” said Pichai at a TGIF, according to a transcript of the meeting provided to Fortune. “I think what happened when the company was smaller, you had a higher chance of knowing about it.”

But where Google management has increasingly used confidentiality as a tool to maintain control of decision-making, some of Google’s activist employees have gone in the opposite direction—turning to the media to amplify their concerns.

That’s a dramatic cultural shift for a company at which talking to the press without approval once guaranteed you’d be “viewed as a pariah,” says Liz Fong-Jones. A former Google site reliability engineer, Fong-Jones had never had a problem criticizing Google, provided it stayed within the company’s (virtual) walls.

But in January 2018, her perspective changed. The catalyst: Google engineer James Damore’s infamous July 2017 memo, an internally published 10-page document arguing that women are underrepresented in the industry owing to biological differences rather than societal factors like bias, and that the company’s diversity efforts were discriminatory. The posting by Damore, who was ultimately fired, created a furor on Google’s freewheeling message boards and mailing lists. These internal communication channels are one of the oddities of Google’s culture: The company has tens of thousands of them dedicated to everything from engineering to all things cats—run by the so-called Mewglers.

Things got even uglier when Damore sympathizers leaked comments made on the message boards by Fong-Jones, a trans woman, and other Google diversity advocates to right-wing news sites. As a result, Fong-Jones says, the group was besieged by harassment and violent threats, which, despite their repeated pleas for help, management was unable to halt. “We were asking them to stop these malicious leaks,” she says. Fong-Jones had a proven track record of getting management to listen to her. She’d successfully spearheaded an effort to get the company to end its policy requiring people to use their real names on its social media site Google Plus, convincing executives that such a policy would expose the most vulnerable users to trolling and worse. But now she felt like the lines of communication between management and employees had broken down.

It was enough for her to decide that this was a problem that would not be solved internally. In October 2017, Fong-Jones and a group of other targeted employees met with Coworker.org, an organization that usually works with low-wage workers, to help think through a PR and internal organizing strategy. “It was clear to us the company wasn’t going to do anything, and we needed to apply media pressure,” Fong-Jones says. In January she and 14 other current and former employees talked about the harassment—and Google’s response to the issue—with Wired.

Understanding that going to Wired without company approval had broken a Google taboo, members of the group published an internal post explaining their motivation—and making clear that they drew a distinction between discussing working conditions (a protected right under labor law) and leaking information about Google products or other confidential company information, which they continued to believe was off limits. Unsurprisingly, not all of their fellow employees bought the justification: “I got some negative comments along the lines of, this really sucks for you, but why did you air Google’s dirty laundry?” says McMillen, one of the then-Google employees who spoke to Wired.

One reason Fong-Jones says she takes a hard line against product leaks is that they provide management with a strong justification for sharing less information with employees. Some point to what happened last August as a prime example. Brin and Pichai were addressing the weekly TGIF meeting when it became clear that someone in the room or watching the livestream of the event was leaking what was being said to a New York Times reporter—who was tweeting the discussion, in real time, to the world at large.

One employee stood up and said “Fuck you!” to the anonymous leaker, to the applause of his colleagues. “That ruined TGIF forever,” says ­McMillen. “Nothing of interest is going to be said at TGIF anymore.”

When he left Google, Poulson says he was warned against talking to the media. “I was explicitly told that should I ever want to come back to the company, they could ignore my politics and focus on my technical contribution as long as I didn’t do something as unforgivable as speak to the press,” he told Fortune. “To be blunt, I don’t think they will be happy I’m having this phone call with you.”

Ahead of the walkout, Pichai sent out a memo to employees voicing his support and acknowledged at a conference that day that Google had not always gotten it right. “There’s anger and frustration within the company,” he said. “We all feel it. I feel it too.” At headquarters in Mountain View, CFO Ruth Porat joined the walkout with her team. Other executives simply avoided the question of whether to participate. Fitzpatrick told Fortune she had been out of the office that day and declined to revisit it when asked if she would have participated had she been on campus.

Parts of the corporate response rubbed organizers the wrong way. They viewed executives’ embrace as an attempt to recast the walkout as some sort of sanctioned company picnic. And if Porat supported the walkout, some asked, why didn’t she use her power as a C-suite executive to implement their demands?

Both McMillen and Fong-Jones quit not long after, saying they found the company’s response lackadaisical. For Fong-Jones, the biggest disappointment was the company’s unwillingness to comply with the organizers’ demand to put a worker representative on the board. “Employees are in a really good position to understand the issues,” she says. She was happy people were staying to fight, but she was burned out.

Google management has shown a willingness to listen to employees—and, in some cases, to change. The company says it had become over-reliant on TGIF and is now too big and sprawling to address every issue in the weekly one-hour meeting. It’s experimenting with adding different forums, like town halls focused on single topics, such as its recently published diversity report. “That was a realization that we came to as we started to see people raising their hands and saying, ‘My voice isn’t getting heard enough,’ ” says Fitzpatrick. And in an attempt to quell the increase in uncivil interactions on its internal platforms, its new “community guidelines” ban slurs and references to sex acts in any work document and require every online group to have a moderator, who must go through training. The company has also revamped internal reporting channels for issues like sexual harassment.

The Google organizers have taken to calling themselves the “entitled vocal majority,” after one employee publicly referred to them as the “entitled vocal minority.” No matter its size, there’s no denying the group has been impactful, playing a role in Google’s decision to not renew its contract for Project Maven. The company also has killed Dragonfly, saying there are no plans to launch search in China and that no work is being undertaken on such a project. Google has also pulled out of its sponsorship of the Conservative Political Action Conference—it irked the company’s liberal employees to see the company’s logo next to the NRA’s—and disbanded its artificial intelligence ethics council after employees published an open letter contesting the appointment of the president of conservative think tank the Heritage Foundation.

Google employees have started to flex their power beyond the company too. The one walkout demand Google met was doing away with forced arbitration, which required employees settle their disputes with the company behind closed doors. A group of Googlers has taken the fight to Washington, where it is pushing for legislation that would ban the practice. “Congress­people take meetings with Google workers that they didn’t take with Chipotle workers,” says Vicki Tardif, an ontologist at Google, who has been with the company for eight years. If they’re able to help push something through, she says, “then we’ve done that greater good that we came to Google to do.”

In April, the conflict inside the company reached a new level when Whittaker and Claire Stapleton, two women instrumental in planning the walkout, published an open letter accusing Google of retaliating against them for their organizing activities. Whittaker wrote that after the A.I. council was disbanded, she was told that in order to remain at the company, she would have to abandon her work on A.I. ethics at Google as well as at the AI Now Institute, an outside organization she cofounded. Stapleton said that after almost 12 years at Google, she was told two months after the walkout that she would be demoted and later that she should go on medical leave, even though she wasn’t sick. It wasn’t until she hired a lawyer that Google conducted an investigation and walked back her demotion, she wrote. “We’re tapping into something that’s an existential threat to Google,” Stapleton told Fortune. The company responded to their accusations that day with a statement saying there was no retaliation and that it prohibits “retaliation in the workplace and investigates all allegations.”

To some employees, the charges of retaliation are the most serious yet levied against the company. Much of the organizing efforts have been led by site reliability engineers (SREs). Their remit is to operate the most critical services Google runs. When something breaks, they’re the ones who get paged to fix it. They troubleshoot and diagnose problems, and they are expected to have opinions and questions. “You have to go probe for weaknesses,” says Fong-Jones, who was an SRE, “and also challenge people when you think something that they’re trying to railroad through is not okay.” Within the SRE world, there’s a concept called blameless postmortem—it’s a way of looking back at mistakes made without throwing anyone under the bus. “It’s a fundamental part of the culture at Google,” says Tariq Yusuf, a privacy engineer who’s been with the company almost five years. “It’s an ability to say this is a thing that’s wrong.” Retaliation, he says, removes the core barrier of being able to safely raise issues. “The whole process breaks down.”

The organizers have started to label their tactics as labor organizing, which some had previously avoided, fearing that it would be off-putting to a workforce that had traditionally aligned itself more with management. During Maven, a few employees went on “interview strikes,” declining to participate in interviewing and recruiting candidates—a form of protest they accelerated in response to the retaliation claims. On May 1, International Workers’ Day, six months after the walkout, employees embraced another old-school labor organizing strategy, staging a sit-in to address retaliation. In New York, the mood was somber, almost vigil-like. A couple hundred employees gathered to talk about the different kinds of retaliation they said they had faced: for organizing, for reporting sexual harassment. Some cried. There was even talk of forming a union. “We’re not walking back our gains,” says Whittaker, “and we’re not going to shut up.”

SUBSCRIBE TO FORTUNE