Defending democracy, elections

U.S. intelligence officials are clear that Russian state actors interfered with the 2016 presidential election, in part by hacking computer networks belonging to political committees, candidates, and advocacy groups connected to the Democratic Party and to Hillary Clinton’s presidential campaign. Analysts say the hackers also infiltrated voter-registration rolls and some voting software and machines in dozens of states, and continue to run disinformation campaigns on social media. Thus far, the special counsel’s office has indicted 13 members of Russia’s military intelligence agency, the GRU, for alleged involvement in the hacking.

Recognizing how serious the hacking threat remains and how outmatched those working on the front lines of elections are (especially with Congress declining to take steps to protect election systems), the bipartisan Defending Digital Democracy project at HKS has been working to help campaigns, as well as state and local officials, identify and ward off threats to their electoral integrity.

Whereas the early focus of the project was getting state and campaign officials up to speed on the range and severity of cybersecurity threats, now the work centers on training officials to become self-sufficient in defending themselves.

In the ramp-up to the midterms, the project convened a military-style table-top exercise session for 120 officials from 38 states to simulate worst-case scenarios around cybersecurity and information operations and trained participants to run their own exercises in their states.

“I’d say on the whole there’s been improvement” since 2016, said Mari Dugas, the project’s coordinator, although nobody is ready to say they’re fully comfortable with the security of their voting systems, she added.

Though information operations are not technically part of state election officials’ jobs, the spread and influence of “fake news” and disinformation has had such an impact on voters that officials have asked the project for help detecting and responding to influence campaigns that seek to undermine trust in elections or mislead voters about how, when, and where to cast ballots.

On Election Day, participating students will observe and assist officials in five states. After the election, the project will convene a small, after-action review session at Harvard with invited officials to share what they saw, any threats they found, where security gaps may still exist, and to understand better any prevailing concerns ahead of the 2020 election.

Shifting ingrained attitudes on election processes has been a challenge, because some state officials have assumed that since they’re not from swing states or their communities are small, they’re unlikely to be targets. “We’ve seen that that isn’t the case. Anybody can be a target of these things, and have been,” so helping officials create a culture of effective cybersecurity will be key, said Dugas.

The voters’ own views

Political veterans were stunned by the 2016 election largely because the victory of Donald Trump shredded what many observers thought were the immutable rules of the campaign game about what worked with voters and what didn’t.

Trump’s win was one shift in a political landscape that has been rapidly changing in recent years. Accepted wisdom about who can credibly enter the political arena, about how candidates reach and sell themselves to voters, and about how they raise money and spend campaign funds no longer hold. The voters who once had to passively endure slick TV ads of candidates chosen by the two political parties now get to pick whose online pitches they’ll entertain from inside their personal information ecosystems.

Amid such rapid change, political analysts and practitioners have struggled to keep pace, especially since 2016 showed how off-track most had gotten. Many hoping to understand how voters think and behave now recognize that polling and opinion data aren’t delivering a rich enough portrait of where the electorate is moving ideologically, and why.

“For a long time, political actors and advocates were using polling, and that set of very small sample data was giving people directional impressions about which parts of America were moving which way, and that was the key guide to strategy,” said Laura Quinn, a visiting fellow at the Ash Center for Democratic Governance and Innovation at HKS and CEO and co-founder of Catalist, a voter data “utility” service. Catalist works with progressive left clients such as the AFL-CIO union and the Democratic Congressional Campaign Committee to tailor political advocacy messaging to voter segments across the U.S. electorate.

Now, with big data part of politics, Catalist tries to drill down into the electorate by analyzing not just the commercial findings and voter file information that all campaigns use, but surveys that elicit the broader viewpoints of would-be voters. So instead of asking questions about expected topics like the Affordable Care Act or gun ownership, which cause people to self-sort into partisan groups, Catalist asks about things that don’t signal an obvious right or left response, but gets at someone’s values. These are queries like “Is it morally good or bad to leave a dog out in the rain, or for a soldier to refuse to obey a potentially illegal order from commanders?”

The respondents are clustered based on their replies, not on their gender, age, income, or even supposed ideology. “Often in politics, we’re using pure demographics as proxies, and that really is not accurate,” said Quinn. For instance, “All women do not see the world the same way.”

With values, not demographics, as the baseline sorting mechanism, she says the firm is able later to superimpose data about voting histories, demographics, and past civic behavior to get a deeper, more-longitudinal picture of how groups view the world, what motivates them, and how they’re likely to act, as opposed to standard polling’s snapshot, moment-in-time approach. “There is no persuasion without understanding.”

Using the values methodology, Catalist has broken down the roughly 230 million voting-eligible Americans into nine groups, ranging from “super seculars” on the far left, represented by comedian Samantha Bee, to “team red” on the far right, embodied by Trump superfan Sean Hannity of Fox News.

The firm Cambridge Analytica was accused of using personal information from Facebook to target pro-Trump messages at millions of U.S. voters in 2016, based on five personality traits. But Quinn believes that a values approach to sorting voters is more predictive of civic behavior.

“None of these things perfectly explains why people are the way they are. People are complicated, and they change. It’s the constellation of things that you believe that give you a sense of the person,” she said.

By using values methodology, the researchers have found that the conventional wisdom about the importance of gender and age in determining voting behavior is often wrong.

“When we see how they’re voting and how they’re behaving in all the civic data that we have, urban young people and young people in counties with major universities are behaving very differently than people their same age who are not in urban places,” said Quinn.

One strong predictor of political attitudes and behavior is marital status. “Men and women who are married are closer in their voting pattern, and single men and women are closer in their voting pattern to each other, and the gender gap between them is smaller than the gap between married people and single people,” she said.

In a changed political environment, voters are much tougher to sell in 2018. Quinn said they’ll give candidates three to nine seconds to make their case in a video pitch before deciding to listen further or turn away. Regardless of party affiliation, listeners want something that feels authentic, unfiltered, and local, she said. In the midterms, Catalist has been working with advocacy groups on ballot initiatives in Michigan, Maine, and Florida, among others, to fine-tune arguments to different voter groups.

Still, Quinn said there’s much more work to be done to understand the electorate. Unfortunately, those spending the most time on it today are practitioners focused on the next election. That’s why academia, as well as those working in political advocacy, “have a big, big role to play in reorienting the ship.”

After the election, Quinn said she’ll examine who voted and for whom, and which parts of the electorate moved in which direction, and then re-analyze the economic conditions of values groups to understand any movement. She would like to evaluate where people get their information and study voters’ changed economic positions to see how such factors might influence their views and behavior.

“If this country is going to be governable, you’ve got to be able to convince, after an election, enough of the folks who are part of the losing party that this is still a country where we can collectively make decisions and occasionally cross the aisle,” she said. “Without that ability, it’s a democracy that’s really in trouble.”