Read the original article here.

---

Internet Censorship in the Age of the Dead Internet: A Detailed Educational Resource

In the contemporary digital landscape, the concept of a "Dead Internet"—an online environment increasingly populated and shaped by automated bots, AI-generated content, and coordinated influence operations rather than genuine human interaction—adds a complex layer to understanding Internet censorship. As governments, corporations, and other actors seek to control the flow of information online, their efforts must now contend with, and sometimes utilize, this non-human presence. This resource explores the mechanisms and implications of Internet censorship, highlighting how its dynamics are shifting in an internet where bots may not only be the target but also the tool and the obfuscation layer for control.

What is Internet Censorship?

Internet censorship refers to the control or suppression of what can be accessed, published, or viewed on the Internet, often enforced through legal means or technical mechanisms. This can range from blocking specific websites or types of content to restricting the ability to publish information online. Censorship can be applied by state governments, organizations (like schools or libraries filtering content), or even through individual or organizational self-censorship driven by fear of repercussions or a desire to conform.

Definition: Internet Censorship The legal or technical control and suppression of content accessible or publishable on the Internet. This includes restricting access to information and placing limitations on what information can be made available online.

The extent of censorship varies significantly across countries, from moderate filtering to comprehensive restrictions on news, social discussion, and information access. It often intensifies during periods of political sensitivity, such as elections, protests, or riots, as seen during events like the Arab Spring or more recent political unrest globally. Beyond direct content blocking, censorship can also manifest through legal means, such as using copyright claims, defamation lawsuits, or allegations of distributing obscene material to suppress content indirectly.

Views on Internet censorship are divided. While many believe access to the Internet and freedom of expression online are fundamental human rights, a significant portion of the global population surveyed has also expressed support for some form of online content regulation. The challenge lies in balancing the desire for free expression with concerns over harmful or illegal content.

The Unique Challenges of Censorship in a Borderless Medium

Internet censorship shares challenges with traditional media censorship (newspapers, TV, etc.) but is uniquely complicated by the borderless nature of the internet. Information hosted outside a country's jurisdiction can still be accessed by its residents. This necessitates technical censorship methods to block access to content that censors cannot physically or legally control at its source.

Early optimism about the internet's inherent resistance to censorship ("The Net interprets censorship as damage and routes around it") has been tempered by the reality of sophisticated technical control methods developed over time. Experts now acknowledge that controlling information online is feasible, especially when significant resources are devoted to building comprehensive censorship systems.

Technical censorship employs various methods:

• Blocking and Filtering: Can be based on static blacklists of prohibited sites/content or dynamic, real-time analysis of traffic. Blacklists might be manual or automated, often non-transparent.
• Points of Control: Censorship can be enforced at various points in the network infrastructure, from the international internet backbone to individual devices.
• Transparency: Censorship is often non-transparent. Users might encounter fake "Not Found" errors instead of explicit blocking notifications, making it harder to identify that censorship has occurred. This opacity can be amplified in a "Dead Internet" where it's already difficult to discern the source or intent of content (human, bot, censored, removed).

Despite these sophisticated methods, total censorship remains difficult due to the distributed nature of the internet. Technologically savvy users can often find ways to circumvent blocks, though this is not universally accessible or without risk. The term "splinternet" is sometimes used to describe the fragmentation of the global internet due to national firewalls and varying censorship regimes, creating distinct online ecosystems.

Content Suppression Methods: Technical Approaches

Technical censorship involves using technology to prevent access to undesirable online resources. Effectiveness varies depending on the sophistication of the methods and the resources available to both the censor and those seeking to circumvent it.

Definition: Technical Censorship The use of technological means to block, filter, or disrupt access to online content or services.

Entities enforcing censorship identify targets based on keywords, domain names, and IP addresses. These targets are often compiled into blacklists from various sources, including government agencies.

Common technical methods include:

• Internet Protocol (IP) Address Blocking: Access to specific IP addresses is denied. This can be inefficient as a single IP might host multiple websites, blocking innocent sites alongside targets (over-blocking). It affects various protocols like HTTP, FTP, etc.
  • Context: This method is less precise but straightforward. In a "Dead Internet," blocking IP ranges might inadvertently block large numbers of bot-controlled sites or services hosted on shared infrastructure, potentially adding to the "deadness" or altering the bot landscape in unforeseen ways.
• Domain Name System (DNS) Filtering and Redirection: Connections to blocked domain names are prevented by either failing to resolve the name or returning an incorrect IP address (DNS hijacking). Affects all IP-based protocols. Circumvention involves using alternative, uncensored DNS resolvers.
  • Context: DNS manipulation can be a primary method. Bots relying on specific DNS settings might be easily blocked, while sophisticated bots or human users using alternative DNS servers can bypass it. The success depends on the bot's design and the user's technical skill.
• Uniform Resource Locator (URL) Filtering: Scanning the full URL string for keywords, regardless of the domain. Affects HTTP. Circumvention involves using escaped characters in the URL or encrypted protocols like VPN/TLS/SSL.
  • Context: More granular than IP or domain blocking. Could target specific bot-generated pages or forums. Encrypted traffic (increasingly common, even for bots) makes this harder.
• Packet Filtering: Examining data packets for controversial keywords and terminating TCP connections if detected. Affects TCP protocols. Circumvention involves using encrypted connections (VPN/TLS/SSL) or altering packet size (MTU/MSS) to avoid triggering filters.
  • Context: Deep Packet Inspection (DPI) is resource-intensive. It could potentially detect patterns associated with bot communication or specific forbidden topics discussed by bots. However, sophisticated bots often use encryption or novel communication patterns to evade detection.
• Connection Reset: Blocking future connection attempts after an initial one is blocked by a filter. Can have cascading effects. Circumvention involves ignoring the reset packet.
  • Context: This punitive measure can disrupt both human and bot activity. It's a blunt instrument that reinforces the "deadness" by making parts of the network unreliable.
• Network Disconnection: Completely cutting off internet access, either regionally or nationally, by disabling routers or cables. Circumvention is difficult, potentially requiring satellite access.
  • Context: The most extreme form. Renders all online activity (human and bot) impossible in the affected area. In a "Dead Internet," this would effectively freeze the entire digital ecosystem in that region.
• Portal Censorship and Search Result Removal: Major platforms (like search engines) exclude specific websites or content from results or listings. This makes content invisible unless the user knows the direct address.
  • Context: Highly relevant to the "Dead Internet." If search engines and platforms are dominated by curated or algorithmically prioritized content (potentially influenced by state actors or corporate interests), dissenting or non-standard human content becomes much harder to discover amidst the noise, regardless of technical blocking. Bot-generated content could be used to flood search results, pushing legitimate content down, acting as a form of censorship by obscurity.
• Computer Network Attacks: Using Denial-of-Service (DoS) attacks or website defacement to disrupt access. Often used by non-state actors but can be state-sponsored or used alongside other methods.
  • Context: Bots are often the primary tool for DoS attacks (botnets). State or non-state actors can deploy vast networks of compromised machines or purpose-built bots to silence opponents by making their sites inaccessible.

Over and Under Blocking: A persistent issue with technical censorship is the difficulty in precisely targeting only undesirable content. IP blocking, for example, can block entire servers hosting multiple sites (over-blocking), while complex content might slip through keyword filters (under-blocking). This imprecision is exacerbated by the dynamic and vast nature of online content, including bot-generated material.

Use of Commercial Filtering Software: Many governments utilize commercial filtering software developed by companies in various countries (e.g., SmartFilter, Websense, Netsweeper). These tools, initially marketed for businesses and schools, are repurposed for national censorship. This industry has faced criticism for enabling human rights violations. Companies like Cisco have been accused of building infrastructure used for state censorship and surveillance (e.g., China's Golden Shield). These filtering programs often rely on category-based blocking, which can be inaccurate, leading to unintended censorship. The lack of transparency from vendors about their filtering criteria and blacklists means that decisions about what constitutes "acceptable speech" are often outsourced to private companies with little public oversight, adding another opaque layer in the digital environment.

Content Suppression Methods: Non-Technical Approaches

Censorship is not limited to technical means. Traditional methods are also employed online:

Definition: Non-Technical Censorship The use of legal, economic, social, or coercive measures to control or suppress online content and expression.

• Laws and Regulations: Legislating what content is illegal or must be removed.
• Requests and Pressure: Formal or informal demands to ISPs, publishers, or authors to alter, remove, or block content.
• Financial Influence: Bribes to include, remove, or slant information.
• Legal Persecution: Arrest, prosecution, fines, or imprisonment for publishing or distributing content.
• Civil Lawsuits: Using defamation or other civil actions to pressure content removal.
• Physical Actions: Confiscating or destroying equipment.
• Institutional Control: Closing down publishers or ISPs, or withholding/revoking licenses.
• Boycotts: Organizing boycotts against entities hosting or publishing targeted content.
• Threats and Violence: Intimidation, attacks, or violence against publishers, authors, and their families.
• Economic Coercion: Threatening job loss.
• Paid Content and Influence Operations: Paying individuals (or using bots/AI) to create content or comments supporting specific positions or attacking opponents without disclosing the source of funding. This "astroturfing" or state-sponsored commenting (like the "50 Cent Party" or "Russian web brigades") directly contributes to the "Dead Internet" phenomenon by flooding platforms with inauthentic, often propagandistic, content designed to manipulate public opinion and drown out genuine dissent.
• State-Created Content: Governments creating their own online publications to shape public opinion and counter independent voices.
• Access Limitations: Restricting internet access through prohibitive costs or deliberately underdeveloped infrastructure.

Platform Censorship and Deplatforming

Web service operators (social media companies, hosting providers, etc.) play a significant role in controlling online content, often based on their own terms of service.

Definition: Deplatforming The act of suspending, banning, or otherwise restricting access to platforms or services for individuals, organizations, or content deemed controversial or in violation of platform policies, effectively silencing their online presence.

Platform censorship, including deplatforming, is a growing concern. Companies reserve broad rights to remove content or terminate user accounts, often without detailed explanations, using vague terms like "at our sole discretion" or "for other reasons." While policies often target hate speech, violence, or illegal content, the interpretation and enforcement can be controversial. High-profile cases, like the coordinated ban of Alex Jones and InfoWars, highlight the immense power platforms wield over online speech.

Context for the Dead Internet: Platform censorship affects both human users and potentially sophisticated bots. If a human's account is removed, their content (including any bot-assisted activity or AI-generated posts) disappears. As platforms evolve to detect and remove bots and inauthentic activity, their actions blur the lines between censoring undesirable content (from humans or bots) and censoring inauthentic accounts. This creates further opacity: was content removed because it was "hateful" (a policy violation) or because it was posted by a suspected bot network (an authenticity violation)? This ambiguity makes it harder for humans to understand the rules of engagement and easier for platforms or state actors (potentially influencing platforms) to remove content under various pretexts.

Circumvention: Navigating the Censored Landscape

Internet censorship circumvention involves bypassing technical filtering and blocking to access restricted material. It relies on the fact that blocking typically affects access, not the existence of content elsewhere on the global internet.

Definition: Internet Censorship Circumvention The process by which users bypass technical or non-technical restrictions to access or publish content that is otherwise censored.

Circumvention methods include:

• Proxy Websites and Servers: Routing traffic through an intermediate server that can access the blocked content.
• Virtual Private Networks (VPNs): Encrypting internet traffic and routing it through a server in a different location, effectively masking the user's IP address and bypassing local filters. VPN use is widespread for privacy and circumvention.
• Sneakernets: Physically transferring data via drives or other media.
• The Dark Web: Using anonymizing networks like Tor to access content not available on the surface web.
• Circumvention Software Tools: Dedicated applications designed to facilitate access to blocked sites.

While effective for tech-savvy users, circumvention can be challenging for the general public. Solutions vary in ease of use, speed, and security. Crucially, using circumvention tools carries risks in many censored environments, potentially leading to legal penalties for the user. Efforts by governments, like the US initiative to deploy "shadow" internet systems for dissidents, highlight the strategic importance of circumvention. Even physically moving to an uncensored location has been observed as a form of circumvention ("Internet refugee camps").

HTTPS Adoption: The widespread adoption of HTTPS (secure HTTP) encrypts much web traffic, making it harder for simple filters to inspect content and block based on URL keywords. However, censors can still block entire domains based on the unencrypted domain name visible during the connection setup. Emerging standards like Encrypted Client Hello (ECH) aim to encrypt more of this initial handshake information.

Context for the Dead Internet: Circumvention tools are used by both humans seeking free expression and bots seeking to bypass restrictions for various purposes (scraping, spamming, influence operations). As censors get better at detecting and blocking circumvention methods, it impacts both human users and bots. The arms race between censors and circumvention tool developers is complicated by the potential for bots to be involved on either side – as mass users of circumvention tools or as automated systems designed to detect and disrupt them.

Common Targets of Internet Censorship

Motivations for censorship fall into several overlapping categories:

1. Politics and Power: Suppressing political opposition, criticism of the government or authorities, content related to minority groups or religions deemed a threat, lèse-majesté (offending symbols of state/royalty), and politically sensitive historical events. State-sponsored bot activity (like the "50 Cent Party") often targets this category, aiming to bury critical human voices or propagate pro-government narratives.
2. Social Norms and Morals: Blocking content considered offensive, undesirable, or harmful based on societal standards. This includes hate speech (racism, sexism, homophobia), illegal drug use, pornography, child pornography (a globally censored category), gambling, violent or criminal content, blasphemy, defamation, and sometimes even political satire or social issue discussions.
3. Security Concerns: Filtering content related to national security threats, including malware, sites promoting insurgency, extremism, or terrorism. DoS attacks, often carried out by botnets, can be used under the guise of security concerns or as a censorship tactic against perceived threats.
4. Protection of Existing Economic Interests: Blocking new online services that threaten established industries, such as VoIP (Voice over IP) services impacting telecommunications monopolies or file-sharing sites impacting copyright holders. Allegations exist that copyright enforcement is sometimes used as a pretext for broader site-blocking measures.
5. Network Tools: Censoring the tools and platforms that facilitate communication and information sharing, including social media, media sharing sites, blogs, email providers, search engines, and circumvention tools themselves. Blocking access to platforms where human interaction should occur contributes significantly to the feeling of a "Dead Internet," driving human conversation elsewhere or silencing it entirely.
6. Information about Individuals (Right to be Forgotten): The concept, particularly developed in the EU, allowing individuals to request the removal of search results linking to outdated or irrelevant personal information. This raises complex questions about balancing privacy rights against freedom of information and the potential for individuals to "whitewash" their past, effectively censoring historical information. Court rulings on its global applicability highlight the conflict between national laws and the global internet, with potential chilling effects on speech.

Definition: Right to be Forgotten The concept that individuals have the right to request that certain outdated or irrelevant personal information be removed from search results, even if the original information remains online.

Resilience to Censorship

A user's ability to resist censorship depends on several factors:

• Awareness: Knowing that censorship is occurring motivates users to seek alternative information and circumvention methods. Lack of transparency (like fake error messages) reduces this awareness.
• Demand: Strong desire for the censored information (e.g., political news during protests vs. general entertainment) increases motivation to circumvent.
• Ability and Resources: Having the technical skills, access to circumvention tools (which may have costs), and reliable internet access to bypass filters.
• Social Networks: Wider, more diverse social networks can provide information about censorship and circumvention methods.
• Content Type: Entertainment content might be more resilient than political content due to higher demand or less stringent targeting.

Context for the Dead Internet: In a bot-filled environment, it's harder for human users to develop awareness of censorship. Is content missing because it was censored, or because no human ever posted it? Is a viewpoint prevalent because it's genuinely popular, or because bot networks are promoting it while human dissent is filtered out? The noise and opacity of a "Dead Internet" can act as a form of censorship itself, overwhelming human voices and making genuine dissent harder to find and trust, regardless of explicit blocks.

Internet Censorship Around the World

Internet censorship is practiced globally, with varying intensity and methods. It's particularly stringent in regions like East Asia, Central Asia, and the Middle East/North Africa. However, even democracies censor content like child pornography, hate speech, or specific illegal material, often with public support.

China is consistently cited as having one of the most sophisticated and extensive censorship systems (often called the "Great Firewall"). It targets political content, religious groups, foreign news sites, social media platforms (like Facebook, Twitter, YouTube), and actively uses censorship to control crowd formation or collective action, regardless of political stance.

International Concerns: The global nature of the internet clashes with national laws. Recent court cases, like EU rulings on the "Right to be Forgotten" and defamation, raise concerns about the potential for national laws to dictate content availability globally, creating a "chilling effect" where platforms or publishers censor content permissible in some countries to avoid legal issues in others with stricter laws.

Internet Shutdowns: An extreme form of censorship is the complete or partial disconnection of internet access. This has been used during protests, political unrest, and even to combat cheating on exams. Notable examples include national shutdowns during the Arab Spring in Egypt and Libya, and more recently in Sudan, Ethiopia, and a particularly large-scale shutdown in Iran during fuel protests in 2019. These shutdowns silence all online activity, human and bot, in the affected area.

Measuring Censorship: Various organizations monitor and report on Internet censorship globally (OpenNet Initiative, Freedom House, Reporters Without Borders, V-Dem, Access Now). They use different methodologies, including technical scanning, expert surveys, and tracking of reported incidents like shutdowns, to assess the level and types of censorship. These reports often highlight the lack of transparency by censoring states, which makes it difficult to fully understand the scope and nature of restrictions.

The Arab Spring and Recent Events

The Arab Spring in 2011 demonstrated the power of social media (Facebook, Twitter, YouTube) for organizing protests and disseminating information, which in turn led to increased censorship and even internet shutdowns in affected countries like Egypt and Libya. This highlighted the vulnerability of centralized network infrastructure to state control.

More recently, events like the Russian invasion of Ukraine have seen state-level blocking of platforms like Twitter and Facebook, and discussions around censoring foreign state media perceived as propaganda. Such events continue to drive the use of VPNs and other circumvention tools as users attempt to access unfettered information.

Conclusion: Censorship in a Bot-Driven World

Internet censorship is a multifaceted challenge involving technical, legal, economic, and social controls. While methods evolve, the fundamental goal of controlling information flow persists.

In the context of "The Dead Internet Files," censorship becomes even more complex:

1. Bots as Tools: Bots and AI are increasingly used not just for DoS attacks but for sophisticated influence operations, astroturfing, and potentially even automated content filtering or surveillance.
2. Bots as Noise: The proliferation of bot-generated content makes it harder for genuine human voices to be heard, acting as a de facto form of censorship by obscuring valuable information and potentially manipulating perception without overt blocking.
3. Opacity Increases: The blend of human and bot activity makes it harder to distinguish between censorship (blocking human dissent), platform moderation (removing spam/inauthentic accounts), and the natural drowning out of content by automated noise. False error messages and non-transparent filtering mechanisms fit seamlessly into a landscape where digital reality is already blurred.
4. Impact on Human Connection: By censoring platforms where human interaction occurs or by flooding them with inauthentic content, censorship directly contributes to the feeling of a "Dead Internet"—an environment less conducive to genuine human connection, discussion, and free expression.

Understanding Internet censorship in the 21st century requires acknowledging the pervasive and complex role of automated systems. As the internet continues to evolve, so too will the methods and implications of control, raising critical questions about the future of human communication and access to information online.