Social media companies often keep much of their user data under tight wraps for various reasons. This makes it difficult for researchers and the media to determine whether activity on social media platforms such as viral content, political polarization, extremist and racist behavior, and other subjects are as prevalent as observers may think. Fortunately, Meta has allowed researcher access to test the extent of these malign activities. This research, which required neither a mandate from regulators nor legislation from Congress, spawned four research reports about the 2020 election cycle in the academic journals Science and Nature.

While these academic efforts are laudable, some are proposing new congressional legislation requiring social media platforms to share their data and algorithmic content with researchers. Though the journal articles cited above were written with Meta’s permission and cooperation, not every social media platform necessarily wants to participate in these studies—nor should they be forced to by legal regulations.

One popular bill currently in front of congressional lawmakers, the Platform Accountability and Transparency Act (PATA), could mandate opening the algorithmic “black box.” This legislation essentially requires social media platforms to release data to researchers. While this provision is intended to promote transparency and safeguard democracy, it raises concerns around violating First Amendment principles, respecting privacy and protecting proprietary information.

Problems with PATA

PATA would compel social media platforms to share data from their advertising libraries and divulge details about their algorithmic design. These requirements could erode a company’s competitive advantage and hamper incentives for innovation. Disseminating these trade secrets may damage the bottom line should competitors gain access to trade secrets or proprietary information.  

The risk of disclosure reduces incentives for investing in future innovation. Expanding access to such data poses significant risks and may squelch creativity by platforms. Why should employees develop new algorithmic software, products and features if these trade secrets could become public knowledge? Competitors would be able to “lift” proprietary information and steal that intellectual property if information on algorithms were made public.

PATA also presents potential data privacy and cybersecurity concerns. The more people who have access to the data, the greater the chance of a significant intrusion by malicious actors. A bad actor could pose as a “researcher,” steal the data and sell it on the dark web. Especially since there is currently no comprehensive federal data security and privacy law in place, it is unclear how a hacker would be prosecuted if a data breach occurred.

Additionally, in its current form, PATA establishes the National Science Foundation (NSF) as the sole agency to determine research projects and decide which analysts can access data from social media platforms. If the NSF approves a research project, “it requires platforms to provide qualified data and information to qualified researchers.” Unfortunately, this mandate is onerous.

Meeting in the Middle

Instead of imposing new mandates on social media platforms, researchers should work with them to identify research opportunities through voluntary agreements. For example, three of the four Nature and Science studies used voluntarily provided user data.

It may also be useful for researchers and social media platforms to establish legal safe harbors or “hold harmless stipulations to avoid litigation against researchers. This would facilitate academic research by establishing appropriate parameters while protecting user privacy and platform proprietary data. For example, a framework could be created for researcher access to viral content metrics. One of the Science papers focused on this topic and showed the release of viral content metrics did not damage Meta in any way. Since there is precedent for this type of research and it can prove valuable to analysts and their audiences, working with social media platforms to facilitate this analysis may prove beneficial. 

Researcher access to data on social media platforms remains a relevant public policy question that addresses important issues for democracy promotion. PATA offers a valuable starting point that attempts to create much-needed guidance and clarity for researchers. Studies of social media platforms can provide important policy insights. However, mandating access and not allowing platforms more input into protecting trade secrets and proprietary information is not prudent and may leave them potentially liable for privacy violations. Problems in the current PATA bill may be mitigated with a plan that first evaluates a voluntary data-sharing arrangement between companies and researchers while establishing best practices for data access.