Online Symposium: Regulatory Threats and Methodological Challenges
In recent years, CELE has been investigating the use of threats by public officials as a governance mechanism for Internet platforms. This research question arose from growing concern over the use of informal governance mechanisms that, unlike formal rule-making processes, remain somewhat outside public scrutiny. If effective, these mechanisms could have profound consequences for how information flows on the Internet without proper accountability. Our research led us to a broad conceptualization of regulatory threats, which we define as any type of expression or public/private action by public officials with regulatory power over others, where they clearly or implicitly express, suggest, or imply their desire to see the behavior of their subject move in a particular direction. The regulatory response to noncompliance does not need to be explicit but may be implicit in the very dynamics of the policymaking process at play. We developed this definition to some extent by comparing it to previous research.
Unlike other areas, platform regulation is, for a number of reasons, more susceptible to regulatory threats. The regime proposed in the 90s to govern the development of platforms—particularly content platforms—and the internet as a whole was one of legal immunity. Except in matters of copyright, the principle of non-liability for third-party content prevailed in the United States, Europe, and some other countries by statute, and in others, including most of Latin America, by free trade agreements and case law. The system of immunities implicitly incentivized corporate self-regulation regarding content moderation and curation on the internet. As a result, this system made corporate self-regulation processes particularly susceptible to pressures from both civil society and governments alike.
Contrary to Wu, we do not exclude beforehand “mere policy guidelines, studies, reports and similar materials” from our definition. It is quite possible that these are part of a threatening pattern. Unlike Halfteck, we believe that regulatory threats are a more precise concept than legislative threats. And contrary to Bambauerwe do not believe that the definition of what constitutes a threat should be normatively burdened to suggest that threats are inherently bad and should necessarily be limited as a governance mechanism. The broader definition we propose serves primarily a descriptive purpose, reflecting how threats operate in practice based on the more detailed accounts produced through judicial investigations in the United States. It is also consistent with how other disciplines have theorized and researched regulatory threats in different sectors.
Studying how these mechanisms work in practice proved to be extremely difficult.
Our goal was to map the actions of public officials and then the reactions of corporations. For the latter, we scraped the policy blogs of Google, X (formerly Twitter), and Meta (formerly Facebook). We also use data on changes to terms of service from the Letra Chicaproject, built by CELE and Linterna Verde from Colombia. We also captured the ToW Center for Digital Journalism database for platforms and publishers, which tracks journalistic production related to Internet governance. For the former, we gathered data from U.S. congressional bills and from CELE's Legislative Observatory (which covers nine Latin American countries). We also captured the Federal Trade Commission database (including commissioners' speeches) and the American Presidential Projectdatabase of presidential speeches. To cover Europe, we relied on data from the Eur-Lex database and—to include the European Commission—we took data from the Press Cornersection of its website. To conduct the analysis, we used a combination of search terms and regular expressions during the data analysis and processing stages. We simplified and coded the information according to certain criteria (e.g., actions and reactions, source database, and tags based on regular expressions of internet governance issues such as disinformation, gender-based violence, hate speech, etc.). We placed all the information on a timeline and visualized it. The processed data can be found in here.
The outcome of this analysis was somewhat limited and played a minor role in our research. While it was useful to organize the information chronologically and to observe the ongoing actions and reactions of platforms and public officials in the database, it proved extremely difficult to establish meaningful links between different events. While causality was difficult to prove from the outset, even identifying significant correlations was challenging because the data collection techniques captured vast amounts of events, and reducing them using filtering techniques did not make the database any more intelligible. The timeline proved useful for navigating specific periods or events (e.g., the platforms’ response to the Covid-19 crisis in 2020 and 2021), especially when compared with the data discovered in cases like Biden v. Missouri. However, as it stands, the database has limited utility. That said, threatening dynamics thrive both in discussions about regulation and when enforcement is at stake. For instance, recent letters from former Commissioner Thierry Breton to corporations in the context of the DSA enforcementhave placed the mechanism under scrutiny in the European Union. The pattern of public interactions shifted from enforcement threats to actual pre-enforcement decisions, such as the initiation of regulatory procedures, investigations,formal requests for information, and direct exchanges with corporate officials where DSA violations were explicitly announced. TikTok's decision to withdraw TikTok Lite from the European market came months after a formal investigation was announced and was celebrated as a direct result of that decision.
Civil society organisations in Europe have complained numerous timesabout these dynamics, and the European Commission has gently distanced itself from Breton's tactics. In the United States, Mark Zuckerberg's letter to the House Judiciary Committeeacknowledged the pressure the company faced from public officials worldwide. Referring to communications from the Biden administration during the Covid-19 pandemic, Zuckerberg said government pressure was misguided and regretted not being more forthright about it. He also mentioned that the company was ready to push back if something like this happens again.
In our view, we are therefore facing a significant methodological challenge in investigating an important mechanism of Internet governance. As our project draws to a close, we aim to address these challenges through an online symposium, which will take place here on the CELE blog , over the coming days. The questions we pose to researchers who have reflected on these issues are as follows.
- Are threats of regulation an important governance mechanism on the internet?
- How can regulatory threats be distinguished from normal rule-making processes (in the early stages) or enforcement dynamics (relevant, for example, in the context of DSA implementation in Europe)?
- If regulatory threats are part of the dynamics of internet governance, are they problematic from a normative perspective (as suggested by First Amendment literature in the United States)? If so, what are those normative concerns?
- Should concrete case studies be developed? If so, do you have any hypotheses about a regulatory process that could be particularly insightful under the framework of regulatory threats being discussed?
We open our discussion with a post by Daphne Keller, which you can find here in its original language.
