Building Infrastructure for Studying Social Media’s Role in Elections and Democracy 

When we created Social Science One over a year ago, we promised to give periodic updates on our progress in making privacy-protected Facebook data accessible to the world’s scientific community. As we reported in previous blog posts, we have made great strides in setting up infrastructure under our new “trusted third party” model for industry-academic partnerships. This work must continue. The future of social science research depends on collaborations between independent researchers and the companies that collect the most important data relating to the impact of new technology on democracy, elections, and an array of other social phenomena.

As we also reported in our previous blog posts, the data Facebook has provided thus far is not what it originally promised in the original July 2018 RFP or what we expected to be able to provide to the academic community at this point. Yet, we continue to make progress: Through our standard Social Science One RFP processes, over 80 researchers can now gain access to the Crowdtangle API – which can be useful for examining activity on public Facebook pages. We also offer support for the Ad Library, which allows for analysis of paid political advertising in the United States, India, and several other countries around the world, and access to an initial dataset of all URLs shared publicly since January 2017. 

This initial version of the URLs dataset contains information about the links themselves, such as whether they were fact-checked and in which country they were most shared. However, as currently available, it does not yet include exposure or engagement data, which would allow researchers to discern how many people and which categories of people have seen or engaged with links in their newsfeeds. After the original RFP was issued, Facebook recognized that to provide secure, privacy-protected data access to researchers in compliance with privacy laws in different countries, it needed to build an unprecedented research infrastructure from scratch. We have worked with Facebook over the past year to build that infrastructure. This infrastructure includes novel technology related to computation, privacy and statistical inference. It also includes new legal agreements, administrative arrangements, community building, and organizational relationships. 

Despite the unforeseen obstacles and the frequent criticism Facebook receives with respect to privacy, antitrust, election interference, misinformation, etc., we believe the company remains committed to this project. Facebook has allocated millions of dollars and hired a team of about twenty employees to work with Social Science One to figure out ways to make data available in safe, secure, privacy-protected ways, and in compliance with all applicable privacy laws. Our hope is that the work we have put in to build this infrastructure will pay off quickly as other data sets are made available. We will also be able to use the important lessons we have learned, and even aspects of the infrastructure, to work with other companies to enable them to contribute to the public good in this way as well.

As we emphasized when we began this project, academics will need to develop partnerships with industry if we are to answer the most pressing questions concerning the impact of new technologies on democracy and an array of social phenomena. Although we have more data than ever before, a smaller percentage of that data is available to researchers than ever before. Today, most of this extremely valuable information -- information from which academics could generate considerable knowledge about human behavior and ultimately public good -- is collected and maintained by private corporations. Governments are making critical policy decisions with long term implications based on political assumptions or unscientific folk wisdom as to what is happening on-line. Independent, critical analysis of data held by the private sector is essential to understanding crucial questions of interest to all, and to informing governments and the corporations themselves as to what policy interventions will serve the public interest.

If you are a researcher at an academic institution and would like access to the Facebook data we now have available, please apply via the RFPs available at; we have instituted fast approval processes for data access and can get you up and running quickly.

Gary King and Nate Persily