The Data Was Known

You know processed food is bad for you. You know you should save more. You know you should exercise more. You know you sit in front of a screen longer than your body was designed for. You have the information. You have had it for years.

You make the same decisions.

The Surgeon General's report on smoking was released on January 11, 1964. The advisory committee chose a Saturday, to minimize the impact on the stock market.¹ Warning labels appeared on every cigarette pack sold in America by January 1, 1966.² Television advertising was banned in 1971. Sixty years of the most sustained public information campaign in American history followed.

The smoking rate in 1965 was 42.4 percent. By 2024, it had fallen to 9.9 percent, the first time below ten percent since surveys began. Twenty-five million Americans still smoke.³

The decline looks like evidence that information works. The curve bent when the structure changed. The warning label appeared in 1966. The television advertising ban followed in 1971. For two decades after the ban, the rate declined slowly. The steep drops came later, synchronized with structural interventions: clean indoor air laws starting in the mid-1970s, workplace smoking bans expanding through the 1990s, federal and state tax increases that raised the cost at the point of purchase. A ten percent increase in cigarette price reduces consumption by four percent in high-income countries, five percent in low-income ones.⁴ Between structural interventions, the rate flattened. The information was constant throughout. It was on the pack the entire time.

The institution that regulates tobacco knew this. The Family Smoking Prevention and Tobacco Control Act, signed in 2009, mandated graphic warning labels with photorealistic images of diseased lungs. The FDA issued its final rule in March 2020, eleven years after the mandate. The tobacco industry sued the same month. A district court struck it down in December 2022. The Fifth Circuit reversed in March 2024. The Supreme Court rejected the industry's challenge in November 2024. A preliminary injunction on January 13, 2025 postponed implementation once more.⁵ Six separate legal proceedings over five years. Seventeen years of institutional energy focused on getting an image onto a pack of cigarettes. Evidence from Canada and Australia shows that graphic labels produce modest, diminishing effects on the populations that see them.⁶ In the same period, no comparable federal fight for tobacco tax increases. The Philippines raised taxes on cheap cigarette brands by 341 percent in 2012. Cigarette sales fell 28 percent in three years. The largest reductions among the poorest. Revenue from the tax funded universal health coverage.⁷ The tax changed the behavior and funded the alternative. The label did neither.

The institution fought for labels. The mechanism that works, taxes, required a political fight beyond the institution's reach.

The label was on every cigarette pack for sixty years. What happens when you put the same kind of information on every food product in America?

The Nutrition Labeling and Education Act was signed on November 8, 1990.⁸ Standardized nutrition labels appeared on most packaged food: calories, fat, sodium, sugar, protein. Updated in 2016 to include added sugars for the first time, realistic serving sizes, and a larger calorie display.⁹

Obesity when the labels were mandated, measured by the National Health and Nutrition Examination Survey: 22.9 percent.¹⁰ Most recent measured data, NHANES 2021-2023: 40.3 percent.¹¹ The information was placed on every product sold in the country. The rate roughly doubled.

The label is readable. The decision does not consult it. The decision happens in an aisle with three children, forty-five minutes before dinner, and a budget that favors whatever is on sale. The label addresses deliberation. The purchase is automatic, shaped by habit, marketing, shelf position, package design, and the memory of what the household will eat without complaint. The label is one input among dozens. It would need to override all of them at the point of decision. It overrides almost none.

Menu calorie labeling became mandatory for chain restaurants in May 2018. The largest study, using internal Starbucks transaction data from New York City's calorie posting law, found a six percent reduction in calories per transaction.¹² The reduction came entirely from food. Beverage orders did not change. The deliberate food decision responded to the label. The automatic beverage habit did not. Across broader meta-analytic evidence, calorie labeling reduced intake by roughly eighteen calories per meal.¹³ On an eight-hundred-calorie fast food order, that is two percent. Calorie-only labels showed no statistically significant effect.¹⁴

The label was on the food. The calories were on the menu. What happens when you teach people the underlying math?

Financial literacy programs have been running in the United States since 1997. The assumption: if people understood compound interest, they would save more. A meta-analysis of 201 studies found that financial literacy interventions explain 0.1 percent of the variance in financial behavior.¹⁵ Effects that did appear decayed to zero within twenty months. The study drew a precise distinction: people who are financially literate save more. Teaching financial literacy to people does not make them save more. The knowledge and the behavior are correlated. They are not causally connected. People who are financially literate do save more. The temptation is to conclude that the literacy caused the saving. The meta-analysis separated the two: when researchers isolated the causal effect of interventions that taught financial literacy, the effect on behavior was indistinguishable from zero. The conditions that produce savings, stable income, parental modeling, low financial stress, also produce financial literacy. Both are outputs of the same structural input. Literacy does not cause saving. The structure causes both.

Plans with automatic enrollment in 401(k) retirement accounts: ninety-four percent participation. Plans with voluntary enrollment: sixty-four percent.¹⁶ Same people. Same plan. Same information available. The default changed. The outcome changed. Thaler and Benartzi's Save More Tomorrow program: eighty percent of participants increased their savings rate from 3.5 percent to 13.6 percent in less than four years.¹⁷ They did not learn about compound interest. They signed a form.

The data was there. The behavior was somewhere else.

Where the decision lives

The smoker knows it is harmful. The cigarette is doing a job the information campaign does not address.¹⁸ Stress relief after a twelve-hour shift. Social bonding on a break with coworkers who smoke. A sensory ritual that marks the boundary between work and rest. Chemical dependency reinforced ten thousand times. The warning label targets the health risk. The job the cigarette performs has nothing to do with the health risk. The smoker already knows it will kill them. The smoker is hiring the cigarette for the five minutes of calm between obligations. Information about mortality does not provide a substitute for those five minutes. Nicotine replacement therapy does. It increases the chance of quitting by 50 to 60 percent.¹⁹ The therapy competes with the cigarette for the same job. The label provides information the smoker already has.

The person who does not save understands compound interest. The arithmetic is not complex. The person is navigating a landscape of rent, utilities, transportation, food, and a paycheck that does not stretch to the end of the month. A financial literacy program can demonstrate that saving six percent of income produces a comfortable retirement in forty years. The person who cannot save six percent this month has not failed to understand the lesson. The paycheck does not permit it. Teaching compound interest to someone who has no money left after rent is not education. It is performance.

You are inside this. You know the research on sleep and screens. You read it on a screen, at night, before going to sleep later than you planned. The information was available at the exact moment the behavior occurred. The behavior won. The screen is brighter, closer, and more immediately rewarding than the article about the screen.²⁰

The institution's tool

A label costs nothing. Reformulating food costs billions. A financial literacy class costs thousands. Capping fund fees costs the financial industry everything. A tobacco tax increase is a political fight. A warning label is a regulatory action.²¹

The institution chooses the intervention that satisfies the demand for action without requiring structural change. The label, the campaign, the disclosure requirement. "We warned them." "We educated them." "We disclosed it." Each satisfies the accountability requirement. The label makes the problem visible to the institution.²² The institution can point to it. The surgeon general can announce the advisory. The agency can mandate the disclosure. Each produces something the institution can point to. The structural intervention, a tax increase, a design mandate, an enrollment default, produces nothing visible. It changes the environment quietly. The credit is diffuse. Nobody campaigns on "I changed the default checkbox on the enrollment form." Everyone campaigns on "I launched a nationwide awareness initiative." The behavior that the label was supposed to change lives in a layer the institution cannot see: habit, craving, stress, convenience, chemical dependency, identity, price, access. The more visible the information, bigger labels, graphic warnings, mandatory disclosures, the wider the gap between what the institution can measure and what produces the behavior.

The FDA exists to regulate product information. Its tool is the label. If labels do not change behavior, the FDA's primary tool is irrelevant to the outcome it exists to improve. The school exists to teach. If teaching does not change behavior, the school's primary function is disconnected from the outcome. The newspaper exists to report. If reporting does not change behavior, the newspaper's purpose is open to a question it cannot afford to ask.

The institution keeps investing in information because the alternative threatens its reason to exist. The school board that acknowledges its financial literacy curriculum does not change savings behavior has no answer for the next budget meeting. The public health department that acknowledges its anti-smoking campaign does not reduce smoking has no justification for the campaign budget. The alternative threatens the institution's reason to exist.

On May 23, 2023, the Surgeon General issued an advisory on social media and youth mental health. "A profound risk of harm."²³ On June 17, 2024, the same office proposed tobacco-style warning labels on social media platforms in a New York Times op-ed.²⁴ The same mechanism as cigarette labels in 1966. Proposed again, fifty-eight years later, by the same office, in a new domain.

Ninety-five percent of American teenagers aged thirteen to seventeen use a social media platform. Forty-six percent report being online "almost constantly," up from twenty-four percent in 2015.²⁵ Perception shifted: forty-eight percent of teens now say social media has a "mostly negative effect" on their peers, up from thirty-two percent in 2022. Awareness increased. Usage did not decrease.

The Surgeon General proposed the mechanism the office controls. Structural intervention, platform design requirements, age verification with real enforcement, liability reform, requires Congress. The label requires a press conference.

In April 2021, the Department of Health and Human Services launched "We Can Do This," a vaccine confidence campaign. Three hundred and seventy-seven million dollars in advertising. Over 1.5 billion dollars allocated for vaccine literacy and confidence. Seven thousand ads in fourteen languages.²⁶ The most information-saturated public health campaign in American history.

Roughly one in five Americans never received a single COVID-19 vaccine dose.²⁷ The partisan gap was twenty-eight percentage points: ninety-one percent of Democrats vaccinated, sixty-three percent of Republicans.²⁸ The gap tracked identity and trust. Among those who said they would "definitely not" get vaccinated, Facebook was the primary information source for thirty-six percent. Personal physicians ranked as the most trusted messengers across all groups, but the people most resistant to vaccination were the least likely to consult them. The campaign reached everyone. The behavior followed who you were, not what you knew.

The institution protects its tool. The institution is made of people. The people have their own reason to believe.

The teacher who believes teaching changes outcomes can keep teaching. The journalist who believes reporting changes behavior can keep reporting. The parent who believes explaining changes the child can keep parenting. The alternative, that structure changes behavior and information does not, removes the informer from the causal chain. The legislator who raises the tobacco tax is in the causal chain. The regulator who mandates auto-enrollment is in the causal chain. The city planner who changes the food environment is in the causal chain. The teacher who explains nutrition is not.

Nobody volunteers for irrelevance.

The people who design information campaigns are, disproportionately, people for whom information worked.²⁹ They went to college. They read labels. They save for retirement. They changed their own behavior in response to knowledge. They assume the mechanism was the knowledge.

It was not. The college graduate who quit smoking after reading a health article quit inside a social network where smoking was stigmatized, a workplace where it was banned, and an income that absorbed the price increases. The article arrived inside a structural environment already pushing toward cessation. The article received the credit. The structure did the work. The person designing the next anti-smoking campaign designs it based on what they believe worked for them. What worked for them was not the campaign.

The assumption that information produces behavior change is held most strongly by the class for whom it appeared true. The cost of the assumption, decades of campaigns that do not reach the behavior, lands on the class for whom it is not.

What changed the behavior

Seatbelt usage in 1983: fourteen percent.³⁰ The National Ad Council had been running "Buckle Up" campaigns for more than two decades. Observational surveys in the 1970s found belt use in major cities at roughly ten percent.³¹ Twenty-five years of the same message. The message was not unclear. Seatbelts save lives. This was communicated on television, on billboards, in driver's education courses, in public service announcements featuring mangled cars and bereaved families. The information was visceral, repeated, and universally available. Usage stayed between eleven and fourteen percent.

In 1984, New York became the first state to pass a mandatory seatbelt law. Before penalties took effect, usage jumped from below twenty percent to forty percent on the announcement alone. With penalties: sixty-seven percent.³² By 1987, twenty-nine states had followed.³³ Usage jumped from fourteen percent nationally to forty-nine percent in a decade. By the end of 1985, compliance fatigue set in. Usage in New York dropped to around fifty percent. The law was new, enforcement inconsistent. It looked like the structural intervention might fade the way campaigns do. It did not. As more states adopted laws, as enforcement became routine, as a generation grew up buckling without being told, the rate climbed through the 1990s and 2000s. By 2024: 91.2 percent.³⁴

The information did not change. Seatbelts save lives. That was known in 1960. That was known in 1983. The structure changed. A fine at the point of decision. In primary enforcement states, where police can stop a driver solely for not wearing a belt: 92.2 percent compliance. In secondary enforcement states: 89.5 percent.³⁵ The two-point structural difference in enforcement tracks a consistent compliance gap across every year measured. The law did in a decade what the information could not do in twenty-five years.

Philippines, 2012. Tax on cheap cigarette brands increased 341 percent. Cigarette sales fell 28 percent in three years. Largest reductions among the poorest.⁷ Price changed the behavior. The label did not.

UK food manufacturers reformulated their products to avoid red traffic-light labels on packaging.³⁶ The label changed the producer's behavior. The consumer did not switch away from red-labeled products. The manufacturer changed the product before the consumer had to choose. The information on the label reached the boardroom, where it became a structural incentive to reformulate. It did not reach the shopping cart, where it would have needed to override habit, price, and craving at the point of purchase.

Information changes what people know. Structure changes what people do.

Retirement savings showed the same pattern. When the employer changed the default from opt-in to automatic enrollment, thirty percentage points more of the workforce participated. The change required no education. It required a different checkbox on the enrollment form.

The diagnostic is specific. The seatbelt fine is present at the moment of buckling. The tobacco tax is present at the moment of purchase. The auto-enrollment default is present at the moment of starting a new job. The manufacturer reformulates before the consumer reaches the shelf. Each structural intervention operates at the point of decision or before it. The warning label, the calorie count, the financial literacy lesson, these operate at a point far from the decision. The smoker does not consult the label when the craving arrives. The shopper does not consult the calorie count when the hand reaches for the familiar box. The employee does not recall the compound interest lesson when the paycheck is gone before the month is over.³⁷

Structure is power. Information is what the people without structural power have left.³⁸

The legislator who raises the tobacco tax changes smoking rates. The regulator who mandates auto-enrollment changes savings rates. The city planner who places the grocery store in the food desert changes diets. The teacher who explains nutrition, the journalist who reports the data, the awareness campaign that reaches ninety percent of the population, these change what people know. They do not change what people do.

Teachers produce literate citizens. Journalists create accountability. Advocates change the terms of the conversation. These functions exist independently of their effect on individual behavior at the point of decision. The finding is narrower than it appears. Information changes knowledge. Behavior lives in a different layer. The two were never the same thing. The belief that they were has directed decades of institutional energy into the visible layer, the label, the campaign, the disclosure, instead of the structural layer where the outcome is determined.

The people who care most about information campaigns, who design them, fund them, evaluate them, and defend their effectiveness, are the people with the least structural power. Educators. Journalists. Advocates. Public health officials. Writers. They invest in information because information is the tool they control. The structural interventions that change behavior, taxes, defaults, bans, design mandates, belong to legislators, regulators, and engineers. The people with the information and the people with the structural power occupy different rooms. The belief that information is power keeps the first group convinced their room is the one that matters.

You knew before you read this. You knew processed food was not helping. You knew you should save more. You knew about exercise, and about sleep, and about the screen keeping you up at night.

In sixty years of tobacco warnings, thirty-four years of nutrition labels, and twenty-seven years of financial literacy programs, the information has not reached the place where the decision lives.

The label is on the pack. The calories are on the menu. The lesson is on the whiteboard. The decision is somewhere else entirely.