Algorithms of Fear

Each week, it seems, brings some new incendiary rhetoric in our presidential election cycle or some new global political disruption emerging from the overlapping social, economic, technological, and environmental systems that govern our lives. Each disruption seems to shock us; and the most severe of them become political milestones, referred to across our digital networks in a kind of digital shorthand: Zika, Trump's wall, Ferguson, Flint. And now, of course, there's Orlando.

Yet despite all of the displays of solidarity and outrage that digitally trail each new disruption or debate, civic technologists are becoming convinced they need to do much more, much faster, to help make the Internet a greater force for civility, resiliency, and broader civic engagement.

One big reason for that, say many civic technologists, is Donald Trump and his high-boil candidacy for the White House. Another reason is that we now live in a world run by algorithms—computer programs that make decisions or solve problems for us. They decide our credit scores; they determine stock prices and even the movies we watch. And especially in this Trump-driven presidential election cycle, say civic technologists, media companies are using content algorithms to drive readers to a lot of highly emotional content that is triggering human behaviors meant to accentuate our differences and stoke our fears—in part, to better hold our attention; people tend to share more content if it makes them afraid or angry, they say.

The trouble is, fear is the most powerful enemy of reason in a democratic society, says Micah Sifry, co-founder of the Personal Democracy Forum (PDF), an annual conference that follows how the Internet is influencing politics, governance, and advocacy. "We should be helping those online learn how to navigate their differences other than driving them into daily or hourly flame wars," he said at the annual PDF gathering in New York over the weekend. Sifry believes it's time to create new algorithms and data tools to help find new ways "to ease polarization when there are differences" online.

"The issue of how technology is now affecting our awareness and our responses to triggers in the news, or our personal triggers, and what that is doing to our politics, is a really critical issue right now," Sifry said in an interview. "Digital media that we were all celebrating a few years ago as being better, more social, more personal, and more decentralized than the old broadcast media are now being re-centralized through channels designed by giant companies like Facebook and Google." Bots that have the effect of accentuating our differences online might be good for the media business, Sifry says, "but they're not so great for democracy."

Wael Ghonim, the Egyptian activist and former Google executive who helped to catalyze the 2011 Tahrir Square pro-democracy demonstrations in Egypt, says US social media companies are using what he calls "mobocratic algorithms" that give us a "missing middle" between the extremes of "likes" and "comments"—which Ghonim says tend to quickly devolve into flame wars on social and digital media platforms. In an interview, Ghonim said, "We who use the Internet now 'like' or we flame—but there's [very little] now happening [algorithmically] to drive people into the more consensus-based, productive discussion we need to have, to help us make civic progress. Productive discussions aren't getting the [media] distribution they deserve. We're not driving people to content that could help us, as a society ... come together without a flame war."

danah boyd, another PDF speaker and social media scholar, said another problem with today's content algorithms supports concerns about fear-mongering. Many content algorithms, she said, also carry data sets "rife with racial bias, and if you're not fixing that, you are building racist systems."

Hossein Derakhshan, an Iranian-Canadian blogger imprisoned in Tehran from November 2008 to November 2014 and the author of the essay, "The Web We Have to Save," says Internet technologists need to "re-imagine new algorithms with new kinds of social values, such as diversity and equality." Can it happen? Derakhshan says yes. "DJs do this all the time," he said. "Their selections are not based on what's new and what's popular. They play old tracks we all loved in the past, and they surprise us with great stuff we would never have listened to otherwise. Maybe it's time for states to intervene and push big social networks to open up their algorithms, and to allow third-party developers to create different kinds of algorithms as plug-ins. What we need now, more than ever, is ... to be surprised."

One of the biggest challenges facing civic technologists today, Sifry says, is a lack of enough funding to build someof the tools needed to "feed us into consensus and conflict resolution," rather than "merely feed our own echo chamber of somebody's flame war." There is also a kind of "disruption fatigue" setting in among some civic activists, Sifry says. "The danger is that we're not building new systems and tools fast enough." The danger, he says, is that "people might eventually give up the notion of having a democratic debate (online) and ... will start saying they've had enough: 'Give me a strongman who will solve it for me' or ...'I'm tired of this disruption for social good.'" To help civil society create better tools faster, Sifry and his PDF co-founder, Andrew Rasiej, used their conference to launch Civic Hall Labs, a new nonprofit that aims to build technology for the public good through a collaborative, multi-disciplinary approach.

Ghonim is optimistic and happy for the wake-up call issued at the conference—to build the tools we need to promote a more respectful public discourse. He agrees that building a more democratic, civil space online won't be easy, but believes it's technically doable. "You can build algorithms and experiences that are designed to get the best out of people, and you can build algorithms and experiences that drive out the worst. It's our job as civic technologists to build experiences that drive th ebest. We can do that. We must do that now."

—By Marcia Stepanek

(This article originally ran in the Stanford Social Innovation Review and is being reprinted here with permission.)