
Has social media manipulation affected the results of elections around the world in recent years, including the 2016 US presidential election that brought President Trump to power?
Finding the answer would require a complicated analysis, but the answer is not unknowable, according to a new paper by researchers from MIT.
On Friday, they laid out a detailed road map to answer the question in the journal Science.
“This is a blueprint for understanding the effects of election manipulation and interference on election outcomes,” said Sinan Aral, a professor at the MIT Sloan School of Management, who coauthored the report with colleague Dean Eckles.
Advertisement
“It can be applied to data retrospectively and proactively in near-real-time during, for instance, the 2020 election,” he said.
Aral and Eckles proposed a systematic, four-step research agenda. They called for 1. cataloguing people’s exposure to manipulative media, 2. combining exposure data with data on voting behavior, 3. assessing the effect of manipulative messages on voting behavior, and 4. computing the consequences of changes in voting behavior on election outcomes.
The paper noted that Facebook founder Mark Zuckerberg has called on the US government to regulate the manipulation of elections by social media campaigns.
But the paper argued, “We cannot manage what we do not measure. Without an organized research agenda that informs policy, democracies will remain vulnerable to foreign and domestic attacks. Thankfully, social media’s effects are, in our view, eminently measurable.”
The paper said that while Russian misinformation campaigns in the 2016 election targeting hundreds of millions of US citizens have been exposed, debate continues about whether the misinformation ultimately affected the outcome. Similar disagreements exist about the Brexit referendum in the UK and recent elections in Brazil, Sweden, and India.
“Such disagreement is understandable . . . Luckily, much of the necessary methodology has already been developed” to understand social media manipulation of elections, the paper said.
Advertisement
Aral said there are challenges to researchers getting the data that they would need to do the analysis. One challenge is the tension between concerns over the privacy of social media users and the desire to know how their opinions and votes are changing.
The paper noted that “well-intentioned privacy regulations, while important, may also impede assessments of election interference by complicating or even outlawing routine retention of data necessary to audit election manipulation.”
In a statement from the university, Aral argued that “there are privacy preserving ways to protect our data and our elections at the same time. To take advantage of them, Congress must consult experts in designing legislation that acknowledges these multiple goals and avoids their tradeoffs.”
“Hardening democracies to manipulation will take extraordinary political and commercial will,” the paper said.
“Politicians in the United States, for example, may have countervailing incentives to support or oppose a postmortem on Russian interference, and companies like Facebook, Twitter, and Google face pressure to secure personal data,” the paper said. “Perhaps only mounting pressure from legislators and the public will empower experts with the access they need to do the work that is required.”
Aral said in the interview that the proposed research could look back and answer the question of whether Russian social media misinformation nudged the 2016 election to the Republican Trump, who lost the popular vote to Democrat Hillary Clinton but garnered enough Electoral College votes to win the presidency.
Advertisement
“Given the right data, I think the answer to that question is knowable — within a certain degree of statistical confidence, obviously,” Aral said.
However, he said, right now there’s a more pressing need to safeguard the next election.
“At this point, what’s more important is what’s happening in 2020. Right now, we’re asleep at the switch. Too little is being done,” he said.