Q&A with Sinan Aral: Can We Protect Democracy from Social Media?
In an August 30 article in Science, MIT IDE lead, Sinan Aral, and MIT Assistant Professor, Dean Eckles, wrote about the critical issue of “Protecting Elections from Social Media Manipulation.” The social media researchers present a new perspective on measuring manipulation of elections — one of the biggest threats to democracy in the modern era.
In the article, they propose a four-point action agenda and discuss the political, legal, and ethical implications of undertaking such actions. “Achieving a scientific understanding of the effects of social media manipulation on elections is an important civic duty,” according to the report. “Without it, democracies remain vulnerable. The sooner we begin a public discussion of the trade-offs between privacy, free speech, and democracy that arise from the pursuit of this science, the sooner we can realize a path forward.”
Aral recently spoke with MIT IDE Editorial and Content Manager, Paula Klein, about the report and in particular, what data scientists and the technology community can do and how they can refocus their efforts. Excerpts from the conversation follow:
IDE: You raise the question: “To what extent are democratic elections vulnerable to social media manipulation?” What is the most important takeaway of your report?
Sinan Aral: Based on much previous research we know we are very vulnerable. The main point here is that while we know a lot about the exposure to fake news during the 2016 election –for instance, 126 million Americans read fake news on Facebook; 20 million saw fake news from Russia on Instagram, and 10 million tweets with 6 million followers were also targeted — not much has changed.
In all, 27% of voting-age Americans were subjected to deliberate misinformation in the weeks leading up to the election, especially in swing states. What we don’t know are the effects, if any, on voting behavior, turnout, and choice.
That information is knowable, however; we have the tools to understand what’s happening and the effects, but we’re not using them. Those of us studying social media influence have developed a number of statistical research methods, such as causal inference, for understanding the impact of online communications on behavior. The methods have been written about by me and others and tested on everything from exercise habits, to dating, to voting, and news consumption. The methods are well developed and understood and they’ve become quite good and flexible; nevertheless, we haven’t applied them to protecting our democracy.
This paper is a call to action for policymakers, platforms, and data scientists, to focus on safeguarding democracy. It is a perspective that says, ‘here’s what we know and don’t know; and here’s what to do about it.’
IDE: You remind us that “we can’t manage what we don’t measure.” Now that measurements are known, what are the best ways to manage and remedy election manipulation?
SA: We are suggesting a four-step process: 1) To understand the reach of social media messages, and to catalog the manipulation of these messages, 2) To measure the exposure — who saw what and why, 3) To infer some effects on behavior from that exposure based on causal inference. In other words, if we can fully understand the behavior changes caused by social media interactions we may be able to predict — and prevent — what may happen as a result. As noted, over the last 10 years, Dean Eckles and I have published major papers on this topic and will continue this work. I started doing this research in 2001 as a graduate student when digital social networks barely existed! 4) To connect all of this understanding to voting behavior to counteract the problem. We need to know how effective was the misinformation, where, and to whom?
IDE: In addition to the critical importance of this work for protecting basic democratic principles, how can those involved with social media — executives, data scientists, policy makers — specifically begin to change current practices?
SA: There isn’t one single most effective way to make change; we need a combination of things and we need broad participation. Citizens and experts alike have to put pressure on Congress and on platform providers.
Dean [Eckles] and I are speaking to members of Congress, but interest has been less than we had hoped and expected. It’s unfortunate because I watched [Facebook CEO Mark] Zuckerberg testify, and I saw how unprepared legislators are on the subject of technology — and they admit it! Neither they nor their staffs grasp the nuances. There’s a big disconnect between what they know and what has to be done and they aren’t consulting us and others for help. That’s a mistake.
Congress is about to undertake a series of critical regulatory efforts on privacy and antitrust and yet, the people making these laws are flying blind.
The whole purpose of my upcoming book is to bring these issues into the public square. Today, platform and tech discussions are relegated to conference halls at MIT. I’m writing a book that I hope will bring the general public into the discussion and raise awareness of the threats.
[The book, titled The Hype Machine, How Social Media Disrupts our World, our Democracies and our Public Health, is due out from Random House in fall 2020. ]
IDE: Do you think manipulation influenced the 2016 U.S. election results and others? Are you hopeful or pessimistic about remedies?
SA: There’s no doubt social media disinformation is going to continue in the 2020 elections; news manipulation and misinformation are just getting worse and more sophisticated in the U.S. and in other countries, too, and it’s spreading to video and audio. It’s imperative that the nation be prepared to intervene, but I don’t know if we will. Voting is a foundational right that props up all the rest of our rights, yet it’s taken for granted.
Digital technology companies have to realize that social media threats are not just about elections; there are tremendous business implications in areas such as fraud and loss of corporate reputation. Because damage can spread so quickly, there are financial risks, too. Companies need to be thoughtful and ready to address fraud and democracy at the same time.
Why aren’t we doing more? Corporate and political will seem to be lacking. Lots of bills about election integrity are languishing in Congress, so there’s no public debate, though we do see growing interest in our work (See note).
Even the most promising research tools won’t help if they’re not applied to election manipulation. We also see a lack of access to data and a lack of cooperation from the platforms that are worried about compromising consumer privacy. There are many reasons things aren’t getting done, but we have to move past them. Without an organized, research-based agenda that informs policy, democracies will remain vulnerable to foreign and domestic attacks.
[NOTE: Sinan Aral’s paper “The Spread of True and False News Online” (co-authored with Deb Roy and Soroush Vosoughi), which was published on the cover of Science in March 2018, has not only made the Top 10 of the Altmetric Top 100 list, it has been ranked by Altmetric as the second-most-talked-about academic paper of 2018 in any scientific discipline.]
Read and watch previous reports on this topic by Sinan Aral on our site here, here, and here.
Aral was also interviewed on MSNBC on the new report. Watch the videos here and here. Also, read the interview in the Los Angeles Times and watch the report on Yahoo Finance.
