Who Benefits? How Incentives Shape the Stories We Are Told

Every story has a supply line.

In the Marine Corps, nobody pretended ammunition appeared because the mission was righteous. Fuel, food, maps, radio batteries, spare parts, transportation, and permission all had to move before the unit could move.

Public narratives work the same way.

A story reaches you because someone funded the reporting, granted access, selected the angle, approved the language, survived the legal review, published it, distributed it, ranked it, and measured whether you reacted.

That does not make the story false.

It means the story has logistics.

“Who benefits?” is one of the most useful questions in systems analysis. It is also one of the easiest questions to abuse. Used well, it reveals incentives, dependencies, conflicts, and silences. Used badly, it becomes a shortcut from suspicion to guilt.

Someone benefited.

Therefore they planned it.

Someone made money.

Therefore the claim is false.

Someone lost money.

Therefore they must be the hero.

That is not pattern recognition.

That is motive wearing a fake mustache and trying to pass as evidence.

The mature question is not simply:

Who benefits?

It is:

How could that benefit have shaped what was funded, noticed, framed, repeated, punished, or left out—and what evidence shows that it actually did?

The Question Behind the Question

When people ask who benefits, they are usually asking something deeper:

Why did this version of reality become the version I was given?

Why did one study receive headlines while another disappeared into a database?

Why did one war become a moral emergency while another remained background noise?

Why did one health risk dominate public conversation while a quieter risk received a paragraph on page fourteen?

Why did one expert become the face of certainty while dissenting specialists were treated as irresponsible before their arguments were examined?

Sometimes the answer is corruption.

Sometimes it is cowardice.

Sometimes it is ordinary professional judgment.

Sometimes it is limited time, limited evidence, or limited public attention.

And sometimes the answer is that the system rewards one story more reliably than another.

That distinction matters because incentives do not need to command people like puppets. They can shape the field on which people make sincere decisions.

A journalist can believe a story matters and also know that the headline needs clicks.

A scientist can pursue honest research while choosing among questions that funding agencies are willing to support.

A politician can sincerely support a bill that also helps donors, employers, or future allies.

A platform engineer can improve engagement without intending to inflame a country.

A creator can believe every word they say while learning, one post at a time, that outrage pays better than nuance.

The incentive does not have to invent the belief.

It only has to reward its repetition.

Every Story Has a Supply Line

A public narrative usually passes through several gates.

Funding

Who pays for the institution, study, campaign, newsroom, nonprofit, documentary, expert panel, or platform?

Funding does not automatically purchase the conclusion. It does influence which projects can begin, how long they can continue, which risks can be absorbed, and which questions never receive resources.

Access

Who gets interviews, documents, briefings, credentials, invitations, leaks, early data, or a seat in the room?

Access can improve reporting. It can also create dependency. The source who can end your access may not need to dictate your sentence. The possibility of losing access can become an editor inside your head.

Selection

Which events become stories?

Newsrooms, agencies, researchers, and platforms face more information than they can process. Selection is unavoidable. But selection determines what enters public consciousness and what remains effectively invisible.

Framing

What is the problem called?

A protest can be framed as civil resistance or public disorder. A policy can be framed as protection or control. A layoff can be framed as restructuring. A civilian death can become collateral damage. The factual events may remain the same while the moral architecture changes.

Distribution

Who decides what appears first, what is recommended, what is searchable, what is demonetized, and what is buried beneath three pages of results?

A story that technically exists but cannot be found does not have the same public power as a story delivered to millions of people before breakfast.

Measurement

What counts as success?

Clicks. Watch time. Subscriptions. Donations. Poll movement. Sales. Citations. Grants. Quarterly revenue. Votes. Reduced liability. Institutional approval.

Whatever gets measured begins training the people inside the system.

Consequences

What happens to the person who tells the inconvenient version?

Maybe nothing.

Maybe they lose access, promotion, funding, advertising, professional standing, or the willingness of colleagues to be seen with them.

Again, none of these gates proves a coordinated deception.

Together, they form the supply line through which public reality travels.

Concentrated Benefits, Distributed Attention

Mancur Olson’s work on collective action helps explain why small organized groups can shape public outcomes more effectively than large diffuse populations.[1]

A company that stands to gain $100 million from a rule has an obvious reason to hire lawyers, lobbyists, researchers, public-relations firms, and policy specialists.

Ten million people who may each lose ten dollars have much less reason to organize. Most will never learn the rule’s name. Even those who notice may reasonably decide that the time required to fight it costs more than the loss.

The same imbalance applies to narratives.

A narrow group with a large stake can fund years of messaging.

The public has jobs, children, rent, grief, appointments, and approximately nine minutes before somebody needs the bathroom.

This does not mean the better-funded story always wins.

It means repetition, access, professional packaging, and persistence are easier to purchase than public attention.

The party with the greatest interest in shaping a story is often the party most capable of staying in the room after everyone else has gone home.

That is not mystical.

It is logistics again.

The Agenda Is a Form of Power

Media influence is often imagined as mind control: a broadcaster speaks, the public obeys.

The evidence is more complicated.

In their classic 1972 study, Maxwell McCombs and Donald Shaw compared the issues emphasized in campaign coverage with the issues voters considered most important. Their work helped establish agenda-setting theory: media attention can influence which subjects become salient to the public, even when it does not dictate exactly what people must believe about them.[2]

That is a subtler form of power.

You do not have to convince the public that your answer is correct if you can determine which question occupies the entire week.

You do not have to erase an issue if you can make another issue feel urgent enough to consume the available attention.

You do not have to lie about an event if you can decide whether it becomes a national crisis, a local story, or no story at all.

Agenda-setting does not require a central committee.

Editors make choices. Reporters chase developments. audiences reward some topics. Political actors manufacture events. Press offices provide ready-made material. Competing outlets copy what is already moving. Algorithms notice the momentum and amplify it.

By the end of the day, the story can appear universally important because every institution is covering it.

Every institution may be covering it because every other institution is covering it.

A synchronized outcome is not always evidence of synchronized intent.

Sometimes it is a feedback loop wearing a press badge.

What the Platform Is Paid to Notice

In 2025, Meta reported approximately $196.2 billion in advertising revenue out of roughly $201.0 billion in total revenue—about 97.6 percent.[3]

That fact does not prove that every decision at Meta is made to manipulate users.

It does tell us what keeps the machinery running.

The Federal Trade Commission’s 2024 examination of major social-media and video-streaming companies concluded that targeted advertising powered the business model of many of the companies it studied and accounted for most of their revenue. The report described an incentive to increase engagement because greater engagement supports the data collection and targeting on which advertising depends.[4]

The platform does not need to hate you.

It needs you to remain measurable.

Your attention becomes useful because it can be predicted, segmented, and sold.

That changes the question a platform asks.

Not necessarily:

Is this true?

But:

Will this hold attention?

Not necessarily:

Will this make the public wiser?

But:

Will this produce another interaction?

Truth can perform well under that system.

So can beauty, humor, education, and genuine human connection.

So can rage.

The machine is not morally offended by any of them. It observes what keeps people on the platform and learns.

Fear Is Efficient

Human attention is not evenly distributed.

A 2023 registered report in Nature Human Behaviour analyzed randomized headline experiments involving about 105,000 headline variations, 5.7 million clicks, and more than 370 million impressions. The researchers found that each additional negative word in an average-length headline increased the click-through rate by 2.3 percent, while positive words reduced it.[5]

That does not mean negative stories are false.

Some things are bad.

Some dangers deserve alarms.

The problem begins when the reward system cannot distinguish necessary warning from manufactured agitation.

A newsroom, influencer, campaign, or activist group does not need to begin with a plan to terrify people. It can begin with a sincere message and gradually learn which version travels farther.

The careful headline underperforms.

The angrier headline wins.

The thumbnail gets darker.

The language becomes absolute.

The audience becomes less informed but more activated.

Then the system calls the result “what people want.”

That is only partly true.

People made choices.

The menu was also designed, tested, ranked, and refined around the choices most profitable to repeat.

Our nervous systems bring the vulnerability.

The business model builds the obstacle course.

Funding Can Shape the Question Before It Shapes the Answer

People often imagine biased research as a scientist changing numbers after a sponsor enters the room carrying a sack with a dollar sign on it.

Real influence can be less theatrical.

Which comparison is chosen?

Which dose?

Which population?

Which endpoint?

How long is the study?

Which adverse effects are emphasized?

Which results become the abstract conclusion?

Which unfavorable project is never completed, never submitted, or never promoted?

A 2017 Cochrane review covering 75 papers found that industry-sponsored drug and device studies more often reported efficacy results and overall conclusions favorable to the sponsor’s products than studies with other funding sources. The review also found that ordinary risk-of-bias tools did not fully explain the association.[6]

This is evidence that sponsorship can be associated with research outcomes.

It is not evidence that every industry-funded study is false.

Industry funds expensive research because industry develops products. Many studies are rigorous. Many researchers act with integrity. Some sponsored findings are independently replicated and clinically valuable.

The lesson is not:

Ignore funded research.

The lesson is:

Funding belongs inside the evaluation, not outside it.

Ask whether the methods are strong.

Ask whether the data are accessible.

Ask whether the comparison was fair.

Ask whether independent studies agree.

Ask whether the conclusion outruns the results.

The conflict does not settle the science.

It tells you where to inspect the seams.

The Algorithm Is Not a Hypnotist

It is tempting to imagine algorithms as remote-control devices installed directly behind the eyes.

That story is too simple.

In a large 2023 experiment during the 2020 U.S. election, researchers moved consenting Facebook and Instagram users from algorithmically ranked feeds to reverse-chronological feeds. The change substantially affected how much time users spent on the platforms and what content they saw, but it did not significantly change polarization, political knowledge, or other key attitudes over the three-month study period.[7]

That finding matters.

It challenges the claim that changing one feed ranking automatically rewires a person’s politics.

But it is not the end of the question.

A 2026 randomized field experiment on X found that switching some users from a chronological feed to the platform’s algorithmic feed increased engagement and shifted several policy attitudes in a more conservative direction. The researchers found that the algorithm promoted more conservative content and fewer posts from traditional news organizations. Switching the algorithm off did not produce an equal and opposite effect, which the authors linked partly to persistent changes in whom users chose to follow.[8]

A separate 2025 study used a browser extension to reorder posts containing antidemocratic attitudes and partisan hostility. Downranking that content improved participants’ views of the opposing political party over a ten-day experiment, without removing political posts.[9]

These studies do not prove that every platform controls users in the same way.

They show why broad claims fail.

Different platforms have different owners, ranking systems, user populations, content cultures, and business objectives. Different outcomes may respond over different time periods. Engagement can change without attitudes changing. Attitudes can shift without party identity changing. Removing an algorithm may not reverse effects already produced through new follows, habits, or communities.

“The algorithm” is not one thing.

It is a set of design choices.

And design choices have incentives behind them.

The Strongest Counterargument

There is a danger in reducing every institution to its revenue model.

Journalists are not vending machines.

Scientists are not grant-seeking robots.

Readers are not cattle.

Public servants are not all captured.

Audiences can reject manipulation. Competitors can expose errors. Professional ethics can resist pressure. Reputational costs can punish dishonesty. Editors sometimes publish expensive stories that anger advertisers. Researchers produce findings sponsors dislike. Employees become whistleblowers. Institutions contain internal conflict because institutions contain human beings.

A profitable story may also be true.

A popular story may also matter.

A donor may support a cause because the cause is worthy.

An advertiser may fund a publication without controlling its editorial decisions.

A platform may optimize engagement while also investing in safety, quality, and user satisfaction.

The mixed evidence on feed algorithms should make us cautious about easy conclusions. Changing exposure can have large effects in one setting and limited measurable effects in another.[7][8][9]

This counterargument does not erase incentives.

It keeps incentives in their proper place.

An incentive is a pressure.

It is not a confession.

The fact that a person could benefit from a claim does not prove the claim is false.

The fact that a person could suffer from a claim does not prove the claim is true.

The fact that an institution makes money from attention does not mean every story it distributes was selected by an evil committee.

It means we should examine whether the incentive shaped the process—and look for evidence of the mechanism.

How to Follow the Incentive Without Losing the Truth

“Who benefits?” should open the investigation.

It should not close it.

1. State the claim without the motive

Write the factual claim first.

Then write the alleged motive separately.

Claim: A platform ranked certain political content more prominently.

Motive: The owner wanted to alter political attitudes.

The first may be measurable. The second requires additional evidence.

2. Identify every form of benefit

Money is only one incentive.

Also look for:

• Attention
• Access
• Status
• Institutional legitimacy
• Legal protection
• Career advancement
• Reduced liability
• Market share
• Political power
• Ideological satisfaction
• Avoidance of embarrassment

3. Map the supply line

Who funded the work?

Who produced it?

Who supplied the information?

Who approved it?

Who distributed it?

Who ranked it?

Who measured success?

Who could punish deviation?

4. Look for the mechanism

Do not stop at “they benefited.”

Did influence operate through a contract, grant, algorithm, editorial policy, access agreement, campaign contribution, legal threat, employment relationship, or shared professional culture?

The more specific the mechanism, the more testable the claim.

5. Compare what was rewarded with what was repeated

Which stories received more promotion, money, airtime, or institutional support?

Did negative, partisan, or simplified versions outperform careful ones?

Did the institution change behavior in response?

6. Seek disconfirming cases

Did the same institution publish information against its financial interest?

Did the funded research produce unfavorable results?

Did the platform amplify content that harmed its preferred narrative?

Disconfirming evidence helps measure how strong the incentive really is.

7. Separate structural pressure from conscious intent

A person can respond to incentives without recognizing them.

A system can produce bias without issuing orders.

Do not accuse individuals of secret intent when the evidence shows only structural pressure.

Do not excuse a harmful structure merely because no one admits intending the outcome.

8. Ask what would change your mind

What evidence would make the incentive explanation less likely?

If no imaginable evidence could do that, you do not have an investigation.

You have a faith commitment with spreadsheets.

9. Assign the evidence level

Use the ladder:

Documented fact → supported interpretation → inference → open question → speculation

“Meta earns most of its revenue from advertising” is documented fact.[3]

“This creates an incentive to increase engagement” is supported by the FTC’s analysis and the structure of the business model.[4]

“This specific post was amplified to change my political belief” would require evidence about ranking, exposure, intent, and effect.

Keep the levels separate.

10. Follow more than one incentive

The institution has incentives.

So does the whistleblower.

So does the critic.

So does the influencer exposing the system.

So do you.

If one side benefits from trust and the other benefits from distrust, examine both business models.

We Are Part of the Incentive System

It is comfortable to imagine narratives as something powerful institutions do to innocent people.

The truth is less flattering.

We train the system.

We click.

We share before reading.

We reward the person who confirms us.

We punish nuance with indifference.

We complain that media is sensational while forwarding the most sensational clip of the day.

We say we want truth, then unsubscribe when the truth becomes complicated.

Independent media is not automatically free of these pressures.

A creator who depends on audience support may become captured by the audience. The audience may not demand a lie directly. It may simply reward certainty, enemies, urgency, and endless escalation.

Yesterday’s brave question becomes tomorrow’s content franchise.

The person who built a platform by telling followers that everyone else is lying may discover that admitting uncertainty is bad for retention.

That is why Conspire With Me has to live under the same standard it applies to everyone else.

We sell products.

We make music.

We want attention.

We want this work to grow and fund good things.

Those are incentives.

The answer is not pretending they do not exist. The answer is making them visible and refusing to let monetization require panic, false certainty, or permanent dependence.

The audience should leave more capable—not more afraid to think without us.

That is the test.

The Question Worth Keeping

Keep asking who benefits.

Just do not let the question become lazy.

Follow the money.

Then follow the access.

Follow the metric.

Follow the career path.

Follow the legal exposure.

Follow the audience demand.

Follow the emotional reward.

Follow what happens to the person who refuses.

Then return to the evidence.

Because benefit can reveal motive.

Motive can suggest a mechanism.

A mechanism can guide an investigation.

But only evidence can carry the claim across the finish line.

Thread Note

Every story has a supply line.

Some are funded by money.

Some by access.

Some by fear.

Some by the audience’s hunger to be told that it was right all along.

Do not mistake the supply line for proof that the cargo is false.

Do not pretend the cargo arrived untouched by the journey.

Ask who benefits.

Ask how.

Ask what changed because of that benefit.

Then test the answer hard enough that even your preferred villain has a chance to be innocent.

The goal is not to become suspicious of every story.

It is to become literate in the forces that help a story survive.

Research Notes & Further Reading

[1] Mancur Olson. The Logic of Collective Action: Public Goods and the Theory of Groups, 1965. Foundational analysis of why concentrated interests can organize more effectively than diffuse publics.
https://users.ssc.wisc.edu/~peoliver/SOC924/Articles/olsonlogic.pdf

[2] Maxwell E. McCombs and Donald L. Shaw. “The Agenda-Setting Function of Mass Media.” Public Opinion Quarterly, 1972. Foundational empirical work on how media emphasis can influence which issues the public considers important.
https://doi.org/10.1086/267990

[3] Meta Platforms, Inc. “Meta Reports Fourth Quarter and Full Year 2025 Results,” January 28, 2026. Reports 2025 advertising revenue of $196.175 billion and total revenue of $200.966 billion.
https://investor.atmeta.com/files/doc_financials/2025/q4/Meta-12-31-2025-Exhibit-99-1-FINAL.pdf

[4] Federal Trade Commission. A Look Behind the Screens: Examining the Data Practices of Social Media and Video Streaming Services, September 2024. Describes targeted advertising, data collection, engagement incentives, and transparency concerns across major platforms.
https://www.ftc.gov/reports/look-behind-screens-examining-data-practices-social-media-video-streaming-services

[5] Claire E. Robertson, Nicolas Pröllochs, Kaoru Schwarzenegger, Philip Pärnamets, Jay J. Van Bavel, and Stefan Feuerriegel. “Negativity Drives Online News Consumption.” Nature Human Behaviour, 2023. Randomized headline experiments found that additional negative language increased click-through rates.
https://doi.org/10.1038/s41562-023-01538-4

[6] Andreas Lundh, Joel Lexchin, Barbara Mintzes, Jeppe B. Schroll, and Lisa Bero. “Industry Sponsorship and Research Outcome.” Cochrane Database of Systematic Reviews, 2017. Systematic review of associations between industry sponsorship and favorable drug and device study outcomes.
https://www.cochrane.org/evidence/MR000033_industry-sponsorship-and-research-outcome

[7] Andrew M. Guess and colleagues. “How Do Social Media Feed Algorithms Affect Attitudes and Behavior in an Election Campaign?” Science, 2023. Replacing Facebook and Instagram algorithmic feeds with chronological feeds changed exposure and platform use but did not significantly change key political attitudes during the study.
https://doi.org/10.1126/science.abp9364

[8] Germain Gauthier, Roland Hodler, Philine Widmer, and Ekaterina Zhuravskaya. “The Political Effects of X’s Feed Algorithm.” Nature, 2026. Randomized field experiment finding that exposure to X’s algorithmic feed affected engagement, content exposure, account following, and several political attitudes.
https://doi.org/10.1038/s41586-026-10098-2

[9] Tiziano Piccardi, Martin Saveski, Chenyan Jia, Jeffrey T. Hancock, Jeanne L. Tsai, and Michael S. Bernstein. “Reranking Partisan Animosity in Algorithmic Social Media Feeds Alters Affective Polarization.” Science, 2025. Demonstrates that changing the rank of hostile political content can alter attitudes toward political outgroups.
https://doi.org/10.1126/science.adu5584