The Internet's Original Sin

Core thesis: The structural problem

DiResta frames disinformation not as a content problem (bad ideas) but as a structural problem rooted in platform business models. The Internet's "original sin" is the move to an ad-based revenue model that required:

1. Audience consolidation — billions of users on a handful of platforms (Facebook, Twitter, YouTube, WhatsApp)
2. Personalized targeting — detailed user data collection and behavioral profiling to make ads effective
3. Algorithmic amplification — recommendation engines and virality algorithms that optimize for engagement, not truth

These three forces converge to create an ecosystem that prioritizes what's popular over what's true. Unlike traditional information gatekeepers (newspapers, broadcast media) that had at least implicit quality filters, social platforms explicitly reward engagement, sensationalism, and divisiveness.

The evolution of online propaganda

DiResta traces how the same infrastructure built for advertisers became available to propagandists:

The advertising precedent: When targeted ads migrated to the Internet, marketers discovered that small ad buys could generate long-term audience relationships if followed by continued engagement. This "sustained engagement" model became the standard for brand-building.

The propaganda playbook: State actors, terrorists, and ideologues adopted the identical framework. A small initial spend or seeded content could build a page, then organic resharing and algorithmic amplification would reach scale. The infrastructure is identical; only intent differs.

Five categories of malign actors:

• State actors — Russia, Iran, Saudi Arabia conducting influence campaigns
• Terrorists — ISIS built a brand on social media using the same professional content strategy as commercial brands
• Domestic ideologues — conspiracy theorists, true believers using organic networks
• Spammers — financially motivated, easier to address because they lack free-speech defenders
• Professional influence operations — any government willing to run sustained campaigns

The Russian interference case study

DiResta uses Russian interference in the 2016 U.S. election as a watershed moment. What was remarkable:

• The ads themselves ($100,000 spent generating pages with 500,000+ followers) were not the primary mechanism
• The key was inauthentic engagement and organic resharing — creating compelling content that users voluntarily amplified
• Tactics included media mirages: cross-linking networks of pages to create sealed ecosystems where users thought they were encountering independent organic content (e.g., pages "about" Black culture, women's issues, LGBTQ+ rights) that were all actually manufactured by the Internet Research Agency (IRA)
• Long-term radicalization occurred through repeated exposure to increasingly extreme content within these curated environments

The Mueller indictment and Virginia state indictment showed the granular sophistication of this work — not crude propaganda but deeply researched, culturally resonant messaging tailored to specific U.S. demographic segments.

Information ecosystem consequences

The convergence of three forces produces an ecosystem with three critical failures:

1. Shared epistemological breakdown — Different populations see different Internets curated by algorithms, leading to incompatible understandings of basic facts
2. Erosion of information integrity — Engagement-optimized systems prioritize emotion and novelty over accuracy, rendering traditional epistemic norms ineffective
3. Arms race dynamics — Because the infrastructure is so effective for coordinated actors, and because most platforms lack sophisticated detection, bad actors iterate faster than platforms respond

Policy and regulatory approaches

DiResta identifies four key regulatory directions (as of MozFest 2018):

1. Ad transparency — Requirements to disclose who is running political ads, what targeting criteria were used, how much was spent. Facebook's initial Ad Archive was a step; ongoing challenge is preventing shell companies and anonymous LLCs from laundering campaign identity.

2. Bot and inauthentic-behavior labeling — Disclosing when content is computer-generated or amplified by coordinated networks. She notes this is an ongoing arms race: disclosure itself becomes a signal (absence of disclosure = trust) that bad actors learn to exploit.

3. Antitrust / platform consolidation — Possibility that breaking up dominant platforms or creating interoperability requirements could reduce the single-point-of-failure problem where one platform can determine the information diet of billions.

4. Privacy regulation — Reducing the detailed user data available for targeting, limiting the precision of microtargeting. She notes European GDPR is ahead of U.S. policy here.

The limits of individual-level solutions

DiResta critiques the faith in free-speech-solves-it approach (Mill, Brandeis): the idea that "good speech will challenge bad ideas and truth will win." This was not what happened with Russian interference, ISIS recruitment, or domestic conspiracy radicalization. She emphasizes that:

• Individual cognitive biases (believing false claims, confirmation bias) exist but are not the root cause
• Platform-level affordances (algorithms, architecture, incentives) matter at least as much as individual psychology
• The problem is fundamentally about infrastructure that enables scaled coordinated action, not individual gullibility

Questions from the audience

On NGO use of these platforms for legitimate advocacy: DiResta acknowledges the same tools (targeted ads, influencer engagement) are effective for nonprofits and advocacy. The challenge is balancing utility against abuse. Her suggestion: industry standards (similar to environmental standards) that legitimize certain practices while restricting others—e.g., requiring ad transparency, limiting microtargeting in certain contexts.

On content moderation: She notes ISIS content was surprisingly difficult to address despite clear consensus that beheading videos should not be permitted. Post-Snowden environment made tech companies reluctant to coordinate with government; civil libertarians opposed platform restrictions; security researchers struggled to explain the severity to an unconvinced audience.

On the effectiveness of Russian propaganda: She emphasizes it was not crude and laughable but sophisticated and resonant. Dismissing it as "obviously fake" misses the point: it was effective precisely because it was well-researched, culturally fluent, and designed to exploit existing social cleavages, not to deceive everyone but to polarize and radicalize specific segments.

Notes

This talk is foundational for understanding how the fake-news problem is not fundamentally about disagreement over facts but about the structural incentives of advertising-based social platforms. DiResta's key insight—that propaganda and advertising use the same infrastructure and thus require infrastructure solutions, not just content moderation—has become mainstream in misinformation research by 2026 but was less obvious in 2018.

The talk also captures a transitional moment: the realization (following 2016) that platforms are not neutral infrastructure but actively shape information flows in ways that state and non-state actors can weaponize. The "original sin" framing is both a diagnosis (the ad-based model itself is the problem) and a call for systemic redesign rather than incremental content moderation.