The First Wave: Six Months of Social Media Platforms Responding to COVID-19
By Matt Bailey, Owen Bernstein, and Hiba Ismail
As the world rapidly responded to the COVID-19 pandemic in the first half of 2020, so too did the social media platforms. As the world closed down, social media usage shot up, with an increasing number of people connecting and getting news and information about the virus for more hours every day. The platforms made public commitments, formed partnerships with the WHO and other health organizations and agencies, and contended with the impact of the pandemic on their own organizations and workforces. This flurry of policies, operational changes, and features announced by the platforms sought to address not only the public health crisis but also the “infodemic” of disinformation and invective that has accompanied it. It also responded to unprecedented levels of use from communities and countries stuck at home, and the need to keep on top of the PR cycle in the context of renewed scrutiny and a dip in public opinion.
Due to the outsized impact that these platforms have on free expression and open discourse, PEN America has watched these developments very closely. For the first six months of 2020, PEN America tracked policy, operational, and functional updates major social media companies made in response and relation to the pandemic. We found that while all of them moved quickly to adjust policies, the platforms pursued their own distinct priorities and strategies. Facebook and Twitter moved to stop the spread of conspiracy theories and refine their advertising policies, while LinkedIn, Snapchat, and Reddit have generally focused less on responding directly to COVID-19 and instead provided updates related to the operation of their own platforms and user interests.
Here, we share our preliminary findings and also outline how a lack of substantive transparency from the companies (necessitating this kind of survey in the first place) is an impediment to tracking how such policy changes impact digital free expression.
The Data
The dataset provided here tracks relevant actions and announcements by 13 of the largest and most influential tech and social media companies and subsidiaries from January through June 2020. Apple (including its operating systems and devices), Facebook (including the main app, Instagram, and WhatsApp), Google (including Ads and YouTube), Microsoft (including LinkedIn), Nextdoor, Pinterest, Reddit, Snapchat, TikTok, and Twitter are included. While not comprehensive, the data represents an effort to capture as many of the most relevant updates as possible within the given time period. In many cases, we provide links to reporting that help contextualize the press release language surrounding the updates. We have not attempted to track research activities and informal remarks by executives in media interviews, nor to perform investigation of changes made behind the scenes. We have also made judgment calls about which announcements are relevant to include. For example, Facebook’s major announcements in May about the launch of “Facebook Shops” are not included; although one could make an argument that the timing of that launch was relevant given the economic downturn and its impact on small businesses, we did not feel this was among the most directly relevant changes made during the period under review.
Explore a simplified version of the dataset here. The complete dataset includes significantly more fields and can be found here as a Google Sheet or downloaded as an Excel-compatible .csv file.
A field reference describing each column in the table, as well as additional fields available in the download, is provided below.
Who’d Like Some Hot Takes?
In their efforts to address the challenges of the pandemic, social media platforms have at times worked together, borrowing each other’s practices and initiatives to supplement their own policies while also differing greatly in the content and focus of their updates. In early March, Google, Microsoft, Twitter, and Facebook issued a joint statement announcing their combined commitment to support users and fight misinformation. Later that month, Twitter, Facebook, and Instagram released updates to prevent profiteering off of panic by banning advertisements for face masks and hand sanitizer. WhatsApp, TikTok, and many others worked together to limit the viral spread of disinformation by limiting forwarding, providing pop-up notifications on fact-checked content, and removing accounts that violated company policies.
The platforms have, of course, also pursued their own priorities and strategies. Facebook and Twitter have each focused heavily and publicly on stopping the spread of conspiracy theories, coordinated disinformation campaigns, and hate speech, while at the same time maintaining a steady clip of new feature releases and public good campaigns. Others, including LinkedIn, Snapchat, and Reddit, have generally focused less on responding directly to COVID-19 and instead provided updates related to the operation of their own platforms and the thematic interests of their users. Pinterest has stood out by attempting to comprehensively redirect users searching for COVID-related terms to authoritative content—keeping the platform focused on well-being and positivity, including through what they’ve termed “compassionate search” features.
In large part, the dataset tells a story about how quickly the platforms have iterated on their pandemic response strategies—and their approaches to telling their own story—throughout the crisis.
A Response that Anticipated the Global Reporting Curve
While individual platforms adopted different responses at different paces, the graph below tells a rough story of the acceleration of the crisis and the responses of the platforms. In mid-March, as COVID-19 continued to spread and more comprehensive data about the virus became available, social media platforms greatly increased the number of updates they were releasing. These updates were efforts to quickly curb the spread of misinformation and profiteering. Many companies also began promoting trustworthy information, including via dedicated educational pages and free advertising provided to the WHO and other authorities. After the initial surge in platform updates, the rate of new policies and other announcements declined but stayed higher than pre-March levels. This is perhaps indicative of both the platforms’ continued efforts to respond to the evolving challenges of the pandemic as well as their understanding of the need to demonstrate their commitment to action publicly.
More Users, More Updates
The rate of social media platforms’ responses was also loosely correlated to the size of the company. Platforms with large user bases were able to respond with a higher volume of updates. For example, Facebook, with over two billion monthly users, made over 50 relevant updates on its main platform in the first six months; Snapchat, with approximately 250 million monthly users, made fewer than 20 relevant updates. The correlation here is rough and complicated by the somewhat apples-to-oranges comparison of an aggregated company like Apple with the disaggregated social properties of Facebook, Inc. One interesting outlier, however, is TikTok, which scrambled as a politically embattled and rapidly growing company during this time to establish foundational policies on issues like privacy for minors while simultaneously responding to world events.
Focusing on, and Messaging, Community Help
We have tagged each item in the dataset for its relevance to seven themes, which themselves emerged from the analysis:
Community Help: Updates that provide information, monetary support, inclusivity, accessibility, or helpful interface tools to users, employees, or communities in general.
Privacy: Updates that maintain or improve the platform’s security, prevent the usage or sharing of users’ information, allow increased access or sharing of data, or provide tools that help users protect users’ information.
Misinformation: Updates that prevent the sharing of, limit the spreading of, or provide labels on misinformation.
Information: Updates that increase transparency or provide resources or tools to users that promote the sharing of credible information.
Advertisements: Updates that change the way advertisements are posted by advertisers or interacted with by users, as well as updates that prevent certain types of advertisements. Additionally, updates that change the platform’s monetization structure in general are included.
Content Regulation: Updates that prohibit sharing or access to certain information, language, or content or that remove previous restrictions on content sharing.
Coronavirus: Updates that are direct or indirect responses to the coronavirus pandemic.
The thematic distribution of each platform’s updates also provides interesting correlations and a few telling outliers. For all the noise regarding disinformation and the removal of inappropriate content, the platforms uniformly chose to emphasize providing updates related to Community Help over other themes. Platforms likewise generally chose to prioritize updates seeking to provide access to quality or authoritative information, for example by releasing new features and portals in partnership with health authorities. One interesting outlier in this analysis is LinkedIn, which has focused heavily on supporting local businesses through monetization and advertising options and has remained largely silent on misinformation and content regulation. TikTok’s scramble to establish foundational policies on content regulation can be seen here as well.
Conclusion: Words and Action
We hope this database will prove useful for those continuing to track social media platforms, advocate for needed reforms, and understand the intertwined futures of public health response and civil rights online.
While we’ve provided links to reporting and analysis of these updates in the dataset, what it ultimately shows us best is how these platforms are presenting and talking about themselves. Very real action has been taken by each of these companies to provide support where it is needed, to respond to extremely rapid shifts in the global threat environment, and respond to changing user behaviors on their platforms. At the same time, it is extremely important to understand these updates as also strategic and meant to control for business risks related to public perception.
Press releases and resources like this one are no substitute for real-time transparency in the form of granular open data and the structures of accountability that can be built on top of it.
However, we hope this resource will be a small piece of the work needed to build better systems to support free expression—even and especially when we are confronted with catastrophes. We need to ensure that the responses of advocacy organizations, elected officials, and the public are as factually grounded as possible, and that those facts come as easily and cheaply as possible when resources are most in demand.
Finally and relatedly, we can no longer afford to pay attention only to the largest platforms. As it’s becoming increasingly clear in the long windup to the 2020 U.S. elections, disinformation campaigns are targeting more platforms, as part of more baroque cross-platform campaigns, than before. We cannot afford to allow smaller and more niche platforms to remain out of sight and out of mind. If the first six months of COVID-19 have provided a stress test of our global social media emergency response system, one lesson we have learned is that even the mid-sized players might be critical infrastructure for free expression and democracy. The need to track what is happening and being done by an expanding number of relevant platforms means that we need an approach to transparency that scales.
Field Reference
These notes are useful to interpret the data. Note that many of these fields are only included in the downloadable versions of the dataset.
The company that the update was released by or applies to.
The date the update was released or applied.
Each update is classified as “Feature,” “Operations,” or “Policy.” These three categories are not mutually exclusive; many updates could be characterized by more than one category. That said, we characterized each update based on the category it most closely conformed to.
- Feature: A change to the structure, form, or function of the platform. The introduction of new tools that change or expand a user’s experience operating on the platform.
- Operations: An action performed by a platform that does not relate to the technical operation of the platform or the user’s experience operating on the platform. This includes lawsuits, public statements, investments, and research commitments among other events.
- Policy: The establishment of new rules, regulations, or guidelines that affect a user’s experience operating on the platform.
Recorded as “Yes” for updates that provide information, monetary support, inclusivity, accessibility, or helpful interface tools to users, employees, or communities in general.
Recorded as “Yes” for updates that maintain or improve the platform’s security, prevent the usage or sharing of users’ information, allow increased access or sharing of data, or provide tools that help users protect users’ information.
Recorded as “Yes” for updates that prevent the sharing of, limit the spreading of, or provide labels on misinformation.
Recorded as “Yes” for updates that increase transparency or provide resources or tools to users that promote the sharing of credible information.
Recorded as “Yes” for updates that change the way advertisements are posted by advertisers or interacted with by users, as well as updates that prevent certain types of advertisements. Additionally, updates that change the platform’s monetization structure in general are included.
Recorded as “Yes” for updates that prohibit sharing or access to certain information, language, or content or that remove previous restrictions on content sharing.
Recorded as “Yes” for updates that are direct or indirect responses to the coronavirus pandemic.
Key development measures the importance of each recorded update. Updates recorded as “Yes” were chosen either because of their relevance to the COVID-19 pandemic, misinformation, and content regulation trends that existed across multiple platforms or because of their expected significant impacts on the platform.
Geography records the region in which each recorded update will be applied.
- Unspecified: The company did not specify where the feature, operation, or policy will be applied.
- Global: The company has specified that the feature, operation, or policy will be applied “globally,” “around the world,” or “everywhere.”
- Specific country/countries/continent: The company has identified a specific country or a number of countries in which this feature, operation, or policy will be applied.
A link to the primary source of information concerning the update, typically the company’s newsroom or a company press release.
A link to a secondary source of information concerning the update, typically coverage of the update by sources other than the company itself.
A link to a tertiary source of information concerning the update, typically coverage of the update by sources other than the company itself or more opinionated coverage.
A brief description of the content of each recorded update.
Project Team
Matt Bailey serves as PEN America’s digital freedom program director, focusing on issues ranging from surveillance and disinformation, to digital inclusion that affect journalists and writers around the world.
Owen Bernstein is a sophomore at Harvard College studying government, data science, and East Asian studies. You can find him on GitHub and LinkedIn.
Hiba Ismail is a senior at Columbia University studying political science. You can find her on Twitter at @hibaiismail.