The birth of the digital public sphere
This article is contributed by Ahmed Medien.
Online forums and social marketplaces have become a large part of the internet in the past 20 years since the early bulletin boards on the internet and AOL chat rooms. Today, users moved primarily to social platforms, platforms that host user-generated content. These platforms comprise the new online public squares where ideas and information are exchanged and debated by anyone. Platforms which offer free services to their users decide on the rules of engagement, and as is the norm, these rules have changed and evolved throughout the past decade and a half. From little to no moderation, platforms have introduced algorithms and guidelines that have shaped public conversations throughout the world (e.g., ethnic violence in Myanmar, BREXIT, the “Stop the Steal” campaign) by their successful implementation or lack thereof. This post summarizes the evolution of content moderation rules and community guidelines of four popular international platforms: Facebook, Twitter, Instagram, and YouTube.
The Timeline of Content Moderation
2004: Facebook launches for university students at select US schools
2005: YouTube launches
2007: Facebook becomes available to the world
2009: Major platforms apply PhotoDNA technology to flag and remove child pornography image (CSAM) online
2009: YouTube and Facebook have become global social platforms, experience blocking in at least 13 countries around the world
2010–2019: Launch of standardized community guidelines
2010: Instagram launches
2010–2011: Social media platforms such as Facebook, Twitter, and YouTube play a major role in carrying the voices and reporting from regional protests in North Africa and the Middle East known as the Arab Spring
2010: Facebook releases its first set of Community Standards in English, French, and Spanish
2011: YouTube makes an exception to allow violent videos from the Middle East if they are educational, documentary, or scientific in nature in response to activists in Egypt and Libya exposing police torture
2012: Facebook acquires Instagram
2012: Twitter launches its first transparency report
2012: YouTube removes, blocks “Innocence of Muslims” video in several Muslim countries
2012: Twitter institutes “country withheld tweet” policy (soon after blocked content in Russia and Pakistan)
2012: Documents from Facebook’s content moderation offices leaked for the first time (Gawker)
2013: Facebook launches its first content moderation transparency report
2010–2019 Launch of standardized community guidelines: Misinformation, terror-linked and organized hate content implodes online
2014: ISIS, terror-linked content, and online radicalization become a major issue on social platforms in several countries
2014: The beheading video of American Journalist James Foley appears online amidst a big wave of terror-linked content
2014: YouTube reverses its policy on allowing certain violent videos
2014: Platforms apply a new rule against ¨Dangerous Organizations¨ linked to terrorism
2015–2016: Twitter changes its content moderation rules on harassment after a remarkable harassment case against the stars of the rebooted Ghostbusters
2016: Major platforms are under amateur and state actor-linked information manipulation campaigns during the 2016 US presidential election
2016: Facebook launches a fact-checking program on its platforms and partners with IFCN fact-checking organizations
2016: Platforms fail to stop campaigns of misinformation that spur ethnic violence in Myanmar against the Rohingya minority
2017–2018: Launch of the Global Internet Forum to Counter Terrorism (GIFCT)
2016: FB live starts to attract an increasing number of suicides and live shootings .
2018: Video of the shooting of Philando Castile in the USA is broadcast live on Facebook.
2018: TikTok launches in China
2018: Twitter removes 70 million bot accounts to curb the influence of political misinformation on its platform
2018: YouTube releases its first transparency (enforcement of community guidelines) report
2018: Facebook forms an Oversight Board to rule over restoring removed content
2018: Facebook allows its users to appeal its decisions to remove certain content
2019: Christchurch terrorist attack (originally broadcast live on Facebook Live) leads to the Christchurch Call to (eliminate terrorist content)
2019: Twitter allows its users to appeal its content removal decision
2019: TikTok launches internationally and attracts a new growth of international audiences
2019: The first emergence of the novel coronavirus in China
2020-Present: New rules to moderate content online expand internationally to counter an ever international phenomena of hate speech, election misinformation, and health misinformation
2020-Q1: Novel coronavirus known as COVID-19 spreads around the world; COVID-19 misinformation soon follows on major social platforms
Major social platforms intervene to ban health information that contradicts government and official sources
Major platforms start to label COVID-19 related misinformation at scale
2020 (Q2+Q3): Facebook Oversight Board chooses its first cases
Twitter and Facebook start labeling the posts of US President Donald Trump
Facebook introduces a slew of new content policy shifts that address:
- Holocaust denial content
- US-based organizations that promote hate
- Organized militia groups
- Conspiracy theories
Major platforms introduce new vetted content around the topic of election integrity in the US after repeated claims of voter fraud
Twitter launches transparency center
2020-Q4: Misinformation around election integrity and fraud intensifies in the US promoted by the US President Donald Trump
Major platforms start fact-checking posts by the US president and other similar users
Facebook starts labeling content in India
Platforms introduce new rules to counter Covid-19 vaccine misinformation
2021-Q1: Jan 6, Facebook suspends the account of US President Donald Trump for violating its community guidelines and inciting violence
Jan 6, Twitter deletes some tweets of US President Donald Trump and locks his account
Jan 8, Facebook, Twitter, YouTube, Instagram ban the account of US President Donald Trump for the remainder of his term in office, block his posts from other accounts
Facebook oversight boards make first rulings on the 6 cases they selected
Facebook Oversight Board announces it will rule on the ban of Donald J. Trump
Facebook amends its groups’ moderation rules, adds new grounds for removal
The rules of Content Moderation are complex
As private companies increasingly take on the public forum’s role, users and businesses operating on these platforms may find their rules increasingly complex. Between curbing the rise of hate speech, disinformation (which sometimes can amount to national security threats), and protecting the fundamental right of expression for their users, platforms find themselves making consequential decisions to operate free-thought-exchanging forums within a profit-driven business model and the grace of new regulatory frameworks that may impede on their ability to expand internationally.
Trust, safety, and accountability are new rules that major platforms have committed to operating by following the Santa Clara principles. With the rise of disinformation and false content on the internet in general and the larger online information ecosystem of which social platforms are a part, trust is the currency with which social platforms and their online communities can thrive. Trust is the evidence users are engaged in a healthy debate online. They trust the information they read and the users they interact with. Accountability is the other side of the same coin. In a competitive, complex world of online spaces and competing ideals, social platforms must ensure they are accountable to their users by providing them with the metrics and the feedback when guideline enforcement mechanisms are applied to protect them from unfettered exposure to harmful online speech and online content.
AI and the first line of defense for Content Moderation
The scale of content moderation on modern internet platforms and the promise of social means that we have got to delegate some of the content moderation work to algorithms and machines. While humans must stay in the loop, AI content moderation systems leverage the scale of large datasets of classified speech to filter harmful online content. The largeness of platforms like Facebook, Twitter, Tiktok, and YouTube and the hundreds of millions of pieces of user-generated content every day makes it imperative to implement AI systems for content moderation. Still, big and small platforms alike will benefit from these solutions in the long run. AI content moderation systems do not eliminate human moderators’ role, because contextual knowledge and judgment continue to be important. Human moderators also contribute to building up new training data that AI algorithms will act on instead of relying on copies from old AI models.
A particularly robust AI content moderation system that adapts to several languages can help community managers and users in the pre-moderation phase. The AI system can both detect and hide offensive content but also educate users on the community rules they have agreed to before they commit a violation and create a healthier online environment conducive for healthy and constructive conversations and exchange of information.