Letter Urging Meta to Maintain CrowdTangle Tool Through Upcoming Elections

Mark Zuckerberg, CEO, Meta 1 Hacker Way
Menlo Park, CA 94025

cc:

Nick Clegg, President of Global Affairs

Adam Mosseri, Head of Instagram

Roy Austin, Vice President for Civil Rights and Deputy General Counsel Chris Cox, Chief Product Officer

Guy Rosen, Chief Information Security Officer

Miranda Sissons, Director of Human Rights 

May 7, 2024

Delivered via electronic mail.

Dear Mr. Zuckerberg:

The undersigned civil society organizations and researchers urge you to fulfill your civic obligations by maintaining the full functionality of CrowdTangle beyond Inauguration Day through at least January 31, 2025, to ensure this vital tool can be used to ensure the integrity of this important election cycle. The gaps in functionality and access between CrowdTangle and the announced Meta Content Library, coupled with a lack of access for key partners, will lead to substantial vulnerabilities during a key election year. We applaud the efforts of Mozilla to pen an Open Letter To Meta in March 2024 outlining similar concerns and highlighting a few additional matters for your consideration.[1] We also applaud the validation of the European Commission in stating that tool

s like CrowdTangle are essential for real-time election-monitoring and that access to these tools during election periods should be expanded rather than deprecated, due to the risk of damage to civic discourse and electoral processes.[2]

As you know, on March 14, 2024 Meta announced plans to phase out CrowdTangle[3], a critical tool employed by a vast network of journalists, researchers, and election observers to scrutinize the flow of information during election cycles in the United States and globally. Set for discontinuation on August 14, 2024, this move comes in the midst of the first presidential election in the United States following the historic attack on the U.S. Capitol that sought to interfere with the peaceful transfer of power. It also follows the tumultuous 2020 presidential elections, which saw an explosion in problematic content and prompted Meta to create an entirely new label for content that delegitimized the outcome of the election or discussed the validity of voting methods, of which we expect more this election cycle.[4] Finally, this election cycle is the first marked by the widespread availability of generative AI tools, which will only exacerbate concerns about problematic content from 2020’s election cycle. This decision jeopardizes essential pre- and post-election oversight mechanisms and undermines Meta’s transparency efforts during this critical period, and at a time when social trust and digital democracy are alarmingly fragile.

Meta’s choice to sunset CrowdTangle will obstruct external insights into the dynam- ics of discourse on Facebook and Instagram, particularly during a year marked by crucial elections with increased threats of political violence.[5] This obstruction poses a severe risk to the efforts of civil rights groups, activists, journalists, and election officials to identify and mitigate political misinformation, incitements of violence, and the online targeting of vulnerable communities. It stands as a formidable barrier to the preservation of electoral integrity and the information ecosystem, which Meta has repeatedly purported to care deeply about.[6] The work of independent researchers and civil society organizations should be valuable to Meta as additional capacity to bolster these efforts – but the closure of CrowdTangle, despite calls from the public to preserve and expand access to the platform, seem to indicate differently. The deprecation of CrowdTangle also poses additional risks to marginalized communities by stripping key civil rights groups of the access and opportunity to research the impact of digital platforms and content on these groups.

In the absence of larger transparency mandates, CrowdTangle has been an essential tool in helping researchers parse through the vast amount of information on the platform and identify harmful content and threats. Researchers have used CrowdTangle in a variety of ways, such as to identify threats to candidates[7], denial of war crimes in Ukraine[8], and branded terrorist content from the Islamic State.[9] Without CrowdTangle, researchers’ ability to identify and alert others about mis/ disinformation, imminent threats to public officials or elections, or terrorist groups organizing on the platforms will be significantly diminished.

While Meta has touted the launch of the Meta Content Library[10] for its access to all public content and the inclusion of Reels and Events, we are very concerned about the restrictive nature of accessing and using data in the Meta Content Library[11], as well as the overall usability, including:

■ More restrictive data retention policies and exportrules: In general,the Meta Content Library has much more restrictive and limiting rules around data use and retention, despite the fact that the Library is composed entirely of the same type of publicly available data in CrowdTangle. At the moment, the Meta Content Library significantly restricts or gatekeeps data access and downloads from its database. Moreover, it only allows users to query a limited public Facebook dataset and a curated set of Instagram public figures and public pages versus CrowdTangle, which allows researchers and users to download any available public data in their system. Moreover, there are limitations that make traditional research and monitoring much more difficult and in some cases, impossible. For instance, the Meta Content Library requires users to refresh all data queries every 30 days. That represents a major challenge for collecting and storing data for research, especially compared to the use of CrowdTangle which doesn’t require any query refreshing over any period of time and supports deeper, longer-term analysis.[12]

■ Much more limited interfaces and fewer insights: The Meta Content Library interface has much more limited overall functionality and, in some cases, no longer has some important data. At the moment, the Meta Content Library interface is limited to searching for posts; CrowdTangle offers multiple interfaces, including the ability to conduct research at an account-level, see trends over time and more.[13] For instance, CrowdTangle’s Intelligence interface makes it possible for researchers to see the growth of an account (or groups of accounts) over time. In order to replicate that functionality in the Meta Content Library, users would have to build and import their own Python code library (as well as delete all the data they’ve collected every 30 days). Moreover, even within the post search interface, there are important insights that are no longer available. For instance, the Meta Content Library does not include any aggregation of overall trends of a keyword or phrase, something that is available in CrowdTangle. In aggregate, the missing functionality at the interface level will make the overall Library significantly less usable, especially for any users who don’t have the capacity to design their own code libraries.

• Lack of clarity around or elimination of team collaboration functions: Initially, the Meta Content Library did not feature the ability to collaborate with colleagues through its interface, which was unlike CrowdTangle and would represent significant challenges for individuals who work within teams to collaborate on their research. While the Beta update now allows for some ability to share searches, it is still unclear to which extent users can share searches and if this will allow the same collaboration that CrowdTangle did.[14]
• Diminished search functionality: The Meta Content Library does not support the same level of flexibility and robustness in the keyword search abilities as CrowdTangle. Moreover, depending on the specific query, the Meta Content Library may also only return a sample of potential matching content rather than all the matching results. This means that running the same query twice could yield different results. Furthermore, the Meta Content Library does not link to individual Facebook posts in results of searches, limiting the ability of researchers to explore that content within the interface.[15]

■ Exclusion of the for-profit news industry: The lack of access for for-profit newsrooms[16] will hinder the ability of most newsrooms to have access to real- time data about what’s happening on Facebook and Instagram around elections and other important topics.[17] It’s true that there are some commercial tools available, but those services don’t offer the same functionality as CrowdTangle and are often prohibitively expensive for many newsrooms. This will likely have a disproportionate impact on allowing the outside world to monitor narratives within smaller communities or marginalized groups that might not otherwise have academic researchers or nonprofit news outlets dedicated to their issues.

■ Restrictive application requirements: The Meta Content Library requires applying on an individual project basis, whereas CrowdTangle allows access at an organizational level. This has the potential to introduce onerous wait times to effectively study new & emerging risk areas in a timely manner and adds more work for civil society groups that already have limited resources.

■ Structural issues that create barriers for scientific research: There are a number of problematic issues with the overall structure of the data in the Meta Content Library compared to CrowdTangle. For instance, as mentioned above, when search queries yield a large set of positive matches, the query will return a sample of matching results versus the complete set of positive matches.
This means that running the same query twice could yield different results and makes it much harder to draw definitive conclusions for a lot of research questions.[18] This is significantly different from CrowdTangle, which returned all the matching results from its entire system even if a query is being run against a more limited overall dataset.[19]

■ Rushed roll-out process: It can take months, if not years, for researchers to get fully onboarded into complicated data toolkits, including building entirely new API integrations, setting up custom queries, familiarizing themselves with new data dictionaries, and more. Yet, by making this announcement, Meta is asking thousands of researchers around the world to transition to a new system in a few months.

We see the potential for the usefulness of the Meta Content Library, but only if its rollout does not leave large gaps, many of which are highlighted above. While plugging these gaps may be on a future roadmap, that does not remedy the pressing issues present in 2024’s high-profile and high-risk election cycle. Above all, given the extremely short time frame planned for Crowd Tangle to Meta Content Library transition, differences between the two tools, and delays inevitable in developing and rolling out new software tools, we do not believe that the Meta Content Library can be an adequate replacement for the role that CrowdTangle plays in enabling civil society and academia to conduct critical research on threats to and around the current election cycle. 

For the sake of transparency, accountability, and continuity, we strongly urge Meta to:

1. Maintain the full functionality and operations of CrowdTangle, including the ability to add new users, until at least January 31, 2025, ensuring coverage through the upcoming election and inauguration in the United States.
2. Ensure existing organizations already using CrowdTangle and entities such as civil society organizations, journalists, researchers, and eligible media outlets, have access to the future Meta Content Library at the organization level through direct inclusion or a streamlined application process.

CrowdTangle has long stood as a paragon of transparency in digital platform opera- tions, offering real-time insights into the dissemination of disinformation, hate speech, and voter suppression efforts on Facebook. This tool has not only been instrumental in addressing challenges to civic discourse and democracy, but has also supported research into human rights abuses, public health crises, and more. Its imminent closure, announced without a ready or equivalent alternative, undermines the spirit of transparency and the broader needs of election monitoring.

Given the gravity of the threats to US democracy and the urgency brought on by your announced timeline for closure of CrowdTangle, we respectfully request that you announce plans to delay the transition as soon as possible and well ahead of your current August deadline.

Please respond to the specific deficiencies outlined in this letter with details on how the Meta Content Library will address them in time for the upcoming election within 21 calendar days of receipt.

Sincerely,

Center for American Progress


Institute for Strategic Dialogue


Accountable Tech


Advancing Justice - AAJC


Alondra Nelson, Institute for Advanced Study


Anna Marchese, Columbia World Projects, Columbia University

Asian and Pacific Islander American Vote

Center for Countering Digital Hate

Center for Democracy and Technology

Color Of Change
Common Cause

Cybersecurity for Democracy


DatastrategIA and Universidad de Navarra

David Silva, Kent State University

Digital Democracy Institute of the Americas (DDIA)

Digital Forensic Research Lab (Atlantic Council)


Dr. Emma L Briant, Monash University


Emerson T. Brooking, Digital Forensic Lab Atlantic Council

Free Press

Friends of the Earth Action


GLAAD


Human Rights Campaign


Human Rights Watch

InfoEpi Lab


Issue One


Jan Zilinsky, Technical University of Munich


Japanese American Citizens League (JACL)


Jeff Allen, co-founder of Integrity Institute


Jeff Hancock, Stanford University


Jennifer Stromer-Galley, Syracuse University


John Perrino, Stanford Internet Observatory


Kaicheng Yang, Northeastern University


Kapor Center


Kathleen Carley, Carnegie Mellon University


Media and Democracy Data Cooperative, Josephine Lukito

Media Matters for America


Michael Best, Georgia Institute of Technology


National Conference on Citizenship

National Hispanic Media Coalition


New America’s Open Technology Institute


Niti Mishra, Pompeu Fabra University (UPF)


NYU’s Center for Social Media & Politics


Public Citizen


Public Knowledge


Rebekah Tromble, George Washington University


Renee DiResta, Stanford Internet Observatory


Rose Jackson, Atlantic Council’s Democracy + Tech Initiative

Salla-Maaria Laaksonen, University of Helsinki


Sikh American Legal Defense and Education Fund (SALDEF)

Tech Justice Law Project


Tech Policy Press


The Tech Oversight Project


Union of Concerned Scientists