The Data Driving Democracy

Potential Solutions

Back to table of contents
Authors
Christina Couch
Project
Commission on the Practice of Democratic Citizenship
 

This section provides a brief overview of the potential solutions experts mentioned. It is not intended to suggest which solutions are the best or most viable; it is only intended to touch on some activity happening in this space that experts deemed important.

Data Sharing Initiatives

Over the past eighteen months, everyone in civic engagement research has been talking about Social Science One, although the conversation has taken multiple turns. Conceived by Gary King of Harvard University and Nathaniel Persily of Stanford Law School as a model for mutually beneficial partnerships between private-sector platforms and independent researchers, Social Science One sought to organize a commission of senior academic advisors, a small number of whom were bound by confidentiality agreements. These advisors would act as trusted third-party data brokers and work with platforms to identify and organize relevant datasets for use in mutually agreed-upon research projects. First, the datasets would be verified to ensure that they’re not cherrypicked by the platform, then Social Science One would solicit proposal requests from the outside research community that would be reviewed by ethics and peer review boards made up of an international group of anonymous reviewers. These anonymous reviewers would evaluate the proposals for merit, but there were a couple of catches: Winning proposals can’t “violate privacy or existing legal agreements/obligations” or “put a company at a competitive disadvantage.”

Gary King, director of Social Science One and the Institute for Quantitative Social Science at Harvard as well as the Albert J. Weatherhead III University Professor at Harvard, said that the commission would provide a buffer for the company, as well as a mechanism by which researchers can verify that the data they receive are real and the platform wasn’t “just selecting datasets to make the company look good.” After a project was approved, there would be no restrictions or approval processes regarding what researchers write within the context of their proposal.

Alondra Nelson from the Social Science Research Council added that, as of August 2019, there had been no indication that a scenario that would involve rejecting a proposal based on the grounds outlined above would be a realistic possibility.

Once selected, scholars would receive funding, gain “privacy-preserving” data access, and retain the right to publish without restrictions from the company.110 Grants were supported by a handful of charitable foundations, which ranged from the John S. and James L. Knight Foundation to the Charles Koch Foundation.111

Social Science One sought to take a first step toward the challenge of solving data access issues for civic engagement researchers while still protecting privacy and corporate concerns. Facebook signed on as the first partner for a research initiative dedicated to studying how social media impacts elections and the democratic process.112 The first round of Social Media and Democracy Research Grant recipients was announced in April 2019.113

But the project has hit major obstacles. As of August 2019, Facebook had not yet made the promised proprietary data available, citing the inability to protect the privacy of its users. That same month, the project funders threatened to cease their funding if Facebook could not deliver data by September 30, 2019.114 The Social Science Research Council issued a statement outlining the steps they would take, which included pausing the review process and paying out full grants to current researchers regardless of data availability.115 Since August, Facebook has released some additional data, although not enough to enable grantees to complete their research. SSRC’s statement included the intention to wind down the project by the end of 2019. A December 2019 statement from Social Science One’s co-chairs and European Advisory Committee said that the data Facebook had released to that point was of “extremely limited scientific value,” and that “there is good reason to doubt whether other useful data will be forthcoming.”116 Social Science One’s Facebook partnership highlights the hurdles researchers face when working with private platforms, and also points to the difficulty that firms have in making data available to researchers when that data are drawn from users in countries with widely varying legal regulations around privacy protection.

Outside of Social Science One, some individual researchers have formed their own platform partnerships, though these arrangements are rare. For example, Politico reported back in 2018 that Facebook provided some de-identified user data to a team led by economist Raj Chetty for research on income inequality in the United States.117 Many different individuals and organizations have laid out their own projects and recommendations for solving data access issues for social science researchers. In 2017, the Commission on Evidence-Based Policymaking recommended establishing a National Secure Data Service that would link federal database systems and streamline access to government data.118 Though other recommendations from the Commission were enacted with the passage of the Foundations for Evidence-Based Policymaking Act of 2018 (H.R. 4174),119 the final bill did not include the creation of a National Secure Data Service.

Robert M. Groves, former Census Bureau director and current Georgetown University provost, and Adam Neufeld, vice president of Innovation and Strategy at the Anti-Defamation League, outlined a model similar to Social Science One’s that would make private-sector data available through an intermediary institution that could ensure privacy.120 The Alfred P. Sloan Foundation’s Administrative Data Research Facilities Network also offers its own data sharing model.121

Endnotes

Regulation and Legal Solutions

Experts had mixed opinions on what role, if any, regulation should play in solving challenges in the field. While some said that without regulation, digital inequalities will continue to perpetuate, others expressed concern about the unintended consequences regulations might have and skepticism as to whether laws around technology, many of which are crafted without input from the research community, would be the most effective means of solving problems researchers face. Several spoke of a greater need for researchers to be involved in the lawmaking process and to ensure that their voices are heard when writing legislation around data use.

“Legitimate public concern about privacy could easily be pitted against the legitimate interests of researchers,” said Christian Sandvig. “We need to ensure that these two goals are not seen as a trade-off for opposites.”

Many experts said that regulation could potentially play a vital role in providing data access and transparency that could transform research in this field. Some specifically cited changing provisions within the Computer Fraud and Abuse Act (CFAA),122 an anti-hacking law that broadly prohibits users from accessing computers without permission. The highly controversial law as it is currently written has implications for researchers123 and makes terms of service violations criminal acts that can carry fines and prison sentences.124

Several individuals and organizations have challenged the law in court, but a few recent cases have given experts some hope that change could potentially be on the horizon. Earlier this year, the Ninth Circuit Court of Appeals ruled in the hiQ versus LinkedIn case that scraping public data “likely” does not violate the Computer Fraud and Abuse Act.125 The Electronic Frontier Foundation hailed the ruling as “a major win for research and innovation” and “an important step to putting limits on using the CFAA to intimidate researchers with the legalese of cease and desist letters.”126

Christian Sandvig along with a team of researchers, journalists, and the American Civil Liberties Union is also challenging the CFAA.127 Filed in the U.S. District Court for the District of Columbia in June 2016, the case argues that the CFAA and overreaching terms of service agreements inhibit robust algorithm auditing and prevent users from uncovering discrimination.128

Other experts called for better regulations that promote technology transparency. Users should know, for example, when a post is a paid advertisement, whether it was created by a human or bot, how users were targeted to see that post, and who is behind ad campaigns. Facebook has taken some steps toward transparency for ads about political and social issues,129 though critics who were not interviewed for this report have pointed to the company allowing ads that contain false accusations as a move away from transparency.130 Some experts said that transparency should apply to all platform activity, not just political speech. This does not mean eliminating anonymous speech, but instead making it clear that the origins of specific messages are unverifiable.

Several experts noted that regulations within the United States will most likely have limited efficacy as platforms continue to grow their base of international users. David Lazer pointed to a report for a further discussion of concerns independent academics who were not interviewed for this report have with digital platforms and the challenges of creating policies to solve those issues.131

Endnotes

Authors
Christina Couch
Project
Commission on the Practice of Democratic Citizenship
Previous Section
Barriers and Challenges
Next Section
Acknowledgments
Back to table of contents