Final installment in Apep Bar’s six-part Manufacturing Consent investigation. Published October 29, 2026.
In earlier installments, this series followed the evolution of public persuasion: Edward Bernays and public relations; race-targeted commercial marketing; the Smith–Mundt framework; the FBI’s documented COINTELPRO operations; and the contested history of intelligence relationships with the press. The final question is contemporary: when a digital platform decides which message each person sees, how does persuasion change?
The answer does not require a theory of an all-powerful secret committee. It requires examining the documented business models, data flows, recommendation systems and experiments that shape modern online experiences. These systems can affect exposure to information. That is not the same as proving they control what every individual believes.
1. From the mass audience to the individualized feed
A traditional newspaper or broadcast advertisement was usually delivered to a broad audience. A digital platform can instead combine signals such as viewing history, interactions and inferred interests to decide which posts or advertisements to show a particular user. An advertiser can select an intended audience; the platform’s delivery system can further determine which eligible users see the message and when. Organic content ranking and paid advertising are related but distinct mechanisms: one orders the posts and videos in a feed; the other distributes sponsored messages under an advertising system.
In September 2024, the U.S. Federal Trade Commission released A Look Behind the Screens, a staff report based on compulsory information requests issued to nine major social media and streaming companies in 2020. The FTC described extensive collection and retention of personal data, including information about people who did not use a particular service, and documented how companies used data in advertising and automated systems. It also reported inadequate user control over some uses of personal information. These are findings of an agency staff investigation, not proof that all nine firms operated identically or that every personalized recommendation was manipulative.
2. Why the business model matters
When advertising revenue depends on finding and retaining an audience, the incentives to collect data, measure engagement and improve ad delivery become powerful. The FTC’s September 19, 2024 findings identified targeted advertising as a central incentive for broad data collection among the firms it examined. The agency also described privacy-invasive tracking technologies, including pixels used to support advertising based on preferences and interests.
But an economic incentive is not evidence of a coordinated political conspiracy. Different firms, advertisers, creators, governments and ordinary users have different objectives. The defensible claim is narrower and more important: commercial systems can determine which messages are delivered to which audiences, and the details of those decisions are not always visible to the people affected.
3. The difference between influencing attention and changing minds
Algorithms can make some material more prominent than other material. Yet the leap from visibility to persuasion must be tested, not assumed.
A major 2023 field experiment by Andrew Guess and colleagues, published in Science, assigned consenting Facebook and Instagram users to chronological feeds instead of the platforms’ default ranked feeds during the 2020 U.S. election period. The change altered time spent on the platforms and the types of content users encountered. However, the researchers did not find significant changes in the measured political attitudes, knowledge or polarization outcomes over the study’s approximately three-month period. See the research publication and abstract and original study.
This result neither proves algorithms are harmless nor establishes that they secretly determine election outcomes. It demonstrates why journalists must separate measurable changes in information exposure from claims about downstream beliefs or behavior. Results from one platform, period and experimental intervention cannot settle every question about every recommendation system.
4. Persuasion can also be built into the interface
Not all influence comes from an algorithmic feed. Design choices can encourage people to subscribe, consent to tracking or make purchases. The FTC’s September 2022 report, Bringing Dark Patterns to Light, examined deceptive interface practices including disguised advertisements, confusing cancellation flows and privacy choices presented in ways that steer consumers toward data sharing. Such practices may violate consumer-protection law depending on the facts.
This matters because modern persuasion is sometimes embedded in the choice architecture of a service rather than in the words of a particular advertisement. A clear disclosure, a genuine opt-out and an understandable cancellation process are therefore questions of consumer autonomy, not simply website design.
5. What regulators have actually required
Europe’s Digital Services Act (Regulation EU 2022/2065) provides a concrete example of a regulatory response. Article 27 requires covered online platforms using recommender systems to explain their main ranking parameters and available user controls. Article 38 requires very large online platforms and search engines that use recommender systems to provide at least one option that is not based on profiling. The law also imposes advertising-transparency obligations and restrictions on advertising based on certain sensitive personal data. These are European legal obligations, not a statement that equivalent protections apply throughout the United States.
The European Commission’s DSA explanation describes options for non-personalized feeds on the largest services and additional protections for users. Whether any particular platform complies fully is an enforcement question that requires its own evidence; the existence of a law does not itself establish compliance.
6. What is documented—and what remains unproven
Documented: major platforms collect and use extensive information; advertising systems target audiences; automated recommendation systems prioritize content; design choices can steer consumer decisions; and independent experiments have shown that changing feed rankings changes what users encounter. These points are supported by the FTC reports, the published experiment and the text of the DSA.
Not established by those sources: that one hidden organization controls every platform, that every user is politically programmed by a feed, or that an algorithm alone determines a person’s vote. Those are broader claims requiring specific evidence about actors, mechanisms, intentions and measurable effects. Treating every personalized recommendation as proof of a clandestine operation obscures the verifiable problems of privacy, transparency and accountability.
7. Questions the public should be able to ask
Who paid for the message? Why was this person selected to see it? What data were used or inferred? Can the user choose a non-profiled feed, limit tracking, or understand the reason an advertisement was shown? Can qualified researchers audit outcomes while protecting personal privacy? Can regulators investigate deceptive design and unlawful data practices? These are testable questions that allow citizens to examine institutional power without assigning collective blame to any race, religion or community.
Conclusion: Consent requires information
The historical thread connecting Bernays to today’s platforms is not a claim that the same people have directed a century-long operation. It is the recurring question of who possesses the means to shape public attention, how those means are financed, and whether people can recognize and challenge attempts to influence them. The modern record supports serious scrutiny of data extraction, targeting, recommendation and deceptive design. It also demands intellectual restraint: influence is not omnipotence, correlation is not proof of causation, and suspicion is not a substitute for a document.
For the people to give meaningful consent, they need more than an endless stream of information. They need to know how that information reaches them.
Primary documents and research
- Federal Trade Commission, A Look Behind the Screens: Examining the Data Practices of Social Media and Video Streaming Services (September 2024).
- FTC, staff report announcement and findings (September 19, 2024).
- FTC, Bringing Dark Patterns to Light (September 2022).
- Guess et al., “How do social media feed algorithms affect attitudes and behavior in an election campaign?” Science 381 (2023), 398–404.
- European Parliament and Council, Regulation (EU) 2022/2065, especially Articles 27, 38 and 39.
- European Commission, Digital Services Act overview and user protections.
Editorial note: This concludes the planned six-part Manufacturing Consent investigative series. Photo credit for the featured image: Gilles Lambert / Unsplash.
