Fake News, Social Media and Covid-19

 

I.Introduction

Just as the COVID-19 pandemic changed every aspect of our lives, it also changed the cyberspace. In particular, this is evident in aspects relating to content moderation of user-generated content on social media platforms. In this short entry I will briefly describe the changes that tech giants introduced to their content moderation practices in light of the coronavirus and examine some of the potential pitfalls of the new order of content moderation.

Three preliminary remarks are worth noting. First, social media platforms legally and substantively differ from traditional content providers in that the content they present is generated by users, usually without editorial discretion or fact-checking. As such, social media giants such as Facebook )FB( and Twitter, are generally considered intermediaries providing the platforms for users to communicate among themselves. Second, critically, digital platforms also differ from traditional content providers in the amount and volume of content generated and the number of users on the platform. If the New York Times website sees 31 million monthly readers, Facebook has 2.5 Billion users each month, and every single hour, 30,000 (thirty thousand) hours of content are newly uploaded to YouTube. Third, this massive volume of user generated content moves us to the second issue – the importance and limits of digital content moderation. Very generally, content moderation refers to a set of practices that online platforms, use to screen, rank, filter, and block user-generated content. All digital platforms moderate, whether it is a closed Facebook page with 100 users or the entire Facebook platform, the role of content moderators is to decide whether to promote or demote posts, which content to recommend to which users or hide from them, what restrictions to apply to content and more. Content moderation shapes what will be seen and heard in the online community.[1] 

While there is a general agreement that without content moderation, human or machine-made, the cyberspace would be a very hellish place, many questions remain in need of debate and perhaps regulation: what content gets moderated, when is it moderated (ex-ante or ex-post) and by whom (human or Artificial Intelligence (AI)). The following will briefly address these   questions.

II.Covid-19, fake news and content moderation

In February 2020, facing the surge in global cases of the COVID-19 virus, Director-General of the World Health Organization (WHO) said: “We’re not just fighting an epidemic; we’re fighting an infodemic.” The director general referred to fake news that “spreads faster and more easily than this virus” and can hamper an effective public health response and create confusion and distrust among people.

In an unusual response, online search and social media platforms like Facebook, Google, Instagram, Pinterest, Twitter, TikTok, YouTube, reddit and others took firm proactive steps to counter the spread of rumours by users of their platforms. This included taking down misleading information about Corona such as that the virus cannot survive in the hot weather, that taking a high dose of chloroquine medication can protect you, and that drinking bleach can prevent the virus.  It also included the broadening of the definition of what is harmful content in relation to the corona.

On March 19, Twitter announced it is banning any kind of information shared on its platform that contains misinformation about the COVID-19 coronavirus. Twitter broadened their definition of “harm” to include “content that goes directly against guidance from authoritative sources of global and local public health information.” Twitter also expanded its dedicated search prompt feature to ensure that searches on its service for information about COVID-19 yield “credible, authoritative content at the top of search” (see, Twitter public policy). Lastly, it included proactively promoting content from trusted sources such as the WHO and government agencies. Pursuantly, a search of the term Corona in google search engine in Israel for example will yield as top results links to official and verified information providers such as the WHO, the Israel heath ministry website and local health care providers (such as Clalit and Maccabi) and as of this week also information about vaccinations.

Social media platforms not only took proactive measures to promote verified information, but they were also unprecedently transparent in regard to these measures, updating their users in the ongoing changes introduced to the service and the steps taken by the platforms to regulate content. Following the suit of industry giants, we began to see unison steps in smaller platforms, more subversive ones, like reddit and even less political ones such as Pinterest. The decision by social media platforms to aggressively filter out unfounded medical advice, hoaxes and other false information that is thought to risk public health is not an obvious call. In order to understand the mini-revolution that took place under the premise of corona, we should look back at the 2016 United States (US) presidential election. 

For years, digital platforms who see their role as intermediaries have resisted intervening in user generated content. Mark Zuckerberg famously and persistently argued that Facebook is not the arbiter of truth. And, indeed, Facebook doesn’t have to be that arbiter. Under Section 230 of the US communication decency act from 1996, intermediaries are afforded a broad immunity from civil liability for content uploaded by users. In the absence of stricter regulation, digital platforms have two main incentives for moderating content: 1) their business model which incentivizes them to protect their services from content that will drive away users; and, 2) the fear from imposed regulation by government agencies.

Following the 2016 presidential elections that brought to the fore the concern over the political impact of fake news and misinformation, also fueled by the data scandal of Cambridge Analytica, there has been a rise in the public scrutiny of how platforms are making decisions regarding user-generated content. Legislators in the US and Europe responded in calls to regulate the cyberspace and expand intermediary liability for User generated content. US courts also have begun to pull away from their broad reading of Section 230 and attend more carefully to the ways in which platforms curate, manipulate and repackage user-generated content.

At the eve of the COVID 19 pandemic, content moderation was going through a legitimacy crisis. On February 18, Facebook published its white paper regarding regulation of content moderation. So, in many respects, the Corona pandemic erupted just at the very moment public scrutiny over content moderation reached a critical mass. Cynics among us will perhaps argue that the seemingly altruistic move by digital platforms to combat disinformation and promote verified news about COVID is an attempt to counter the tech-lash and win favorable headlines for responsible behavior.

It is hard to argue against the importance of curtailing misinformation about sham treatments, hoaxes and other potentially harmful behaviors especially when those are endorsed by the US president. Nevertheless, if this role that tech companies took upon themselves in response to Corona will redefine the new normal, we expect to see proactive content moderation spreading into other content areas. And maybe we already began seeing it with the 2020 election, as big social media sites took steps to mitigate election-related extremism and disinformation, culminating in a faceoff between Trump and twitter and the great migration to alternative platforms like Parler.

III.Algorithmic content moderation

The corona crisis served as a major catalysator for another key change in the digital platform ecosystem – the move relying on machine learning and AI tools to fight misinformation, fake news, and other harmful content, in hopes that fast and automated computer systems can deal with the constant need to scale up operations. Following the backlash of 2016, tech companies engaged with third party companies to moderate content, hiring close to 30,000 content moderators employed in poor and trauma inducing conditions. However, like in many other sectors, the spread of the coronavirus sent these workers home, pushing the industry forward with ML algorithms compensating for the loss of human labor due to restrictions on work.[2]

In many respects, the crisis pushed social media companies to experiment in what they long dreamed of: using machine-learning tools and AI — not humans — to police posts on their platforms. The scale of user generated content does not really leave much option but to rely on ML at least in some of the tasks. What became apparent in 2020 is that solely relaying on AI is still far from good enough. While AI can speed up time-consuming steps such as going through the vast amount of content published online every day and flagging material that might be false, AI can’t make the final judgment calls. As content moderators argued recently, “Without our work, Facebook is unusable, Its empire collapses. Your algorithms cannot spot satire. They cannot sift journalism from disinformation. They cannot respond quickly enough to self-harm or child abuse. We can.”

When the social media giants announced the changes, they acknowledged the algorithms might struggle to differentiate between legitimate and illegitimate content. And indeed, the effects were almost immediate. Facebook and Google roughly doubled the amount of potentially harmful material they removed in the second quarter of 2020 compared with the three months through March. But while far more content was flagged and removed for allegedly breaking the companies’ rules on what could be posted online, in some areas dangerous and possibly illegal material was more likely to slip past the machines. And other content that should not have been blocked, was removed. For example, in Syria, where campaigners and journalists rely on social media to document potential war crimes, scores of activists’ accounts were closed down overnight — often with no right to appeal those decisions.

Alongside the risk of more false positive removals of legitimate content, the algorithmic content moderation replaces the previous ex-post moderation that was based on user and government flagging of problematic material with ex-ante removal of information. Shifting to ex ante moderation effectively censors the content, stripping the online communities from their agency, preventing them from debating on what is harmful content. Algorithmic content moderation as many other processes of algorithmic decision making is also far less transparent. The black box propriety nature of content moderation, and the lack of accessibility to review the data that was removed grants the private companies enormous power to shape the new public square that the cyberspace is.

IV.Conclusion: Implications of the new content moderation order

At the end of 2020, after almost a year of fighting a pandemic and an infodemic, one can safely conclude that the coronavirus resulted in two major shifts to content moderation:

First, the willingness of tech giants to proactively moderate user-generated content on their platforms, and second, the way content is being moderated (mainly by AI). Coupled together, these changes contribute to the drafting of a new order in content moderation and free speech on the cyberspace more broadly. As we hopefully look toward a post-COVID future, we need to assess the challenges that this new order poses to basic rights such as freedom of speech, right to information, right to truth and more.

In reshaping what freedom of expression means in 2021, we must move beyond a binary view of two equally harming options: no moderation at all which means throwing away the cyberspace as we know it, or wild moderation of content based on the platforms’ business models, adopting instead a hybrid way for content moderation regulation.