The Problem Was Solved

In 2012, American doctors wrote 260.5 million opioid prescriptions. Enough for every adult in the country to have a bottle. The overprescription had been building for fifteen years, fueled by pharmaceutical marketing that told doctors the drugs were safe for chronic pain. By the mid-2010s, the crisis was visible. Addiction, overdose, death. The response was structural. In 2016, the CDC published its Guideline for Prescribing Opioids for Chronic Pain.¹

The prescriptions fell. By 2024, opioid dispensing had dropped 52 percent. 260.5 million prescriptions became 125.7 million.

In the same period, deaths from illicitly manufactured fentanyl rose from 3,105 to 72,776.²

The prescriptions fell. The deaths multiplied.

The people who needed opioids did not stop needing them when the prescriptions disappeared. The dependency had been created by the original overprescription era of the late 1990s and 2000s. Doctors had been told the drugs were safe. Patients had been given them for back pain, for dental procedures, for recovery from surgery. By the time the CDC published its guideline, millions of Americans had a dependency the healthcare system had created and the healthcare system was now shutting off.

The demand reorganized around illicit supply. Illicit supply was fentanyl, because fentanyl is cheaper to produce, more potent per gram, and easier to smuggle than heroin. A kilogram of fentanyl can be synthesized from precursor chemicals for a fraction of the cost of harvesting an equivalent amount of heroin from poppies. The supply chain itself reorganized around the economics of the new drug. By 2023, synthetic opioids accounted for 69 percent of all overdose deaths in the United States.³

The CDC acknowledged this in its revised 2022 guidelines. Policies drawn from the original guideline, the agency wrote, had "contributed to patient harm, including untreated and undertreated pain, serious withdrawal symptoms, worsening pain outcomes, psychological distress, overdose, and suicidal ideation and behavior."⁴

The fix worked. The system reorganized. The reorganization killed 72,776 people in a single year.

In December 1945, Alexander Fleming stood in Stockholm to accept the Nobel Prize for the discovery of penicillin. In his lecture, he described an experiment. Expose bacteria to concentrations of penicillin not sufficient to kill them, and the bacteria develop resistance. "The time may come when penicillin can be bought by anyone in the shops," he said. "Then there is the danger that the ignorant man may easily underdose himself and by exposing his microbes to non-lethal quantities of the drug make them resistant."⁵

He gave a hypothetical. Mr. X has a sore throat. He buys some penicillin and gives himself not enough to kill the streptococci but enough to educate them. He infects his wife. She gets pneumonia and is treated with penicillin. The streptococci are now resistant. The treatment fails. Mrs. X dies. "Who is primarily responsible for Mrs. X's death?" Fleming asked. "Why Mr. X whose negligent use of penicillin changed the nature of the microbe."

By 2019, antibiotic-resistant infections killed 1.27 million people per year. More than HIV. More than malaria. An additional 4.95 million deaths were associated with resistant infections that year.⁶ Eighty years separated Fleming's warning from the body count. The mechanism he described in Stockholm operated across those eighty years without interruption. Bacteria exposed to antibiotics at sub-lethal doses evolved resistance. The resistance accumulated. The miracle drug that had saved hundreds of millions of lives reorganized the microbial world it was designed to control.

The man who invented the fix described its reorganization on the day he was celebrated for creating it.

He was describing a property of every fix applied to a living system.⁷

The reorganization

On March 11, 2011, a magnitude 9.1 earthquake struck off the northeast coast of Japan, triggering the tsunami that caused the Fukushima nuclear disaster. Within days, German Chancellor Angela Merkel ordered the immediate closure of Germany's seven oldest nuclear reactors. Over the next twelve years, the remaining plants were shut down one by one. On April 15, 2023, the last three reactors went offline. Nuclear risk in Germany dropped to zero.⁸

The grid reorganized. Germany had removed roughly half its zero-carbon baseload power. Nuclear had provided roughly a quarter of the country's electricity. Renewables could not scale fast enough to fill the gap. Coal and gas filled it instead. A study published in the Journal of the European Economic Association found that the phase-out replaced nuclear generation primarily with coal-fired production and net imports. The lost nuclear capacity was not replaced by wind or solar. It was replaced by lignite and hard coal, the dirtiest fuels on the grid.⁹

Over 70 percent of the resulting social cost was mortality. Air pollution from the additional coal burning, the particulate matter released when fossil fuels are combusted, kills an estimated 1,100 people in Germany every year. The researchers estimated the total social cost at $12 billion per year.

A 2025 analysis found that if the nuclear plants had remained operational, Germany's electricity prices in 2024 would have been 23 percent lower.¹⁰ Nobody gamed the grid. The grid is a physical system. It reorganized around the gap nuclear left. The reorganization was atmospheric.¹¹

The prescription fix reorganized demand at the biological level. The nuclear fix reorganized an energy grid at the atmospheric level. The next case reorganized at the level of human development.

In the spring of 2020, schools across the United States closed to reduce COVID-19 transmission. The closures worked. In-school transmission fell during the peak pandemic waves. The schools stayed closed, in many districts, through the 2020–2021 academic year. Some large urban districts remained primarily remote for more than a year.¹²

The 2023 National Assessment of Educational Progress tested thirteen-year-olds across the country. Reading scores had fallen to levels last seen in 1971. Math scores had fallen to levels last seen in the 1990s. More than fifty years of reading progress and more than twenty years of math progress, erased.¹³

The 2024 assessment, covering fourth and eighth graders, confirmed that recovery had not arrived. 33 percent of eighth graders scored below the Basic threshold in reading, the highest proportion in the history of the exam. The bottom 10 percent of students saw their losses grow 70 percent larger between 2022 and 2024. Fourth-grade reading fell another two points. Eighth-grade math was flat.¹⁴

Congress allocated $189.5 billion in federal recovery spending through three rounds of Elementary and Secondary School Emergency Relief funding.¹⁵ No state saw reading gains in either grade between the 2022 and 2024 assessments.

The mental health data ran alongside the academic data. CDC data showed that emergency department visits for suspected suicide attempts among girls aged twelve to seventeen were 50.6 percent higher in early 2021 than in the same period of 2019. Among boys the same age, the increase was 3.7 percent. The disproportion was not random. Girls' social development during adolescence depends more heavily on peer connection. The closure removed the infrastructure of that connection.¹⁶

The fix protected children from a virus. The system reorganized around the missing developmental infrastructure. You cannot add back missed developmental time the way you add back missed instruction. The $189.5 billion tried. It has not worked.¹⁷

The opioid fix reorganized demand. The nuclear fix reorganized a grid. The school closures reorganized a generation. Each fix succeeded at what it targeted. Each system reorganized at a level the fix could not reach. The level was different every time. Biological. Atmospheric. Developmental. The fixer could not see it because the fixer was looking at the level the fix addressed.

Some fixes go further. They succeed so visibly that the visibility becomes the product.

The performance

COVID prompted widespread adoption of test-optional admissions policies. By 2024, nearly two thousand colleges and universities no longer required SAT or ACT scores. The policy was designed to increase access and equity.¹⁸

A Harvard-based research team studying admissions at Ivy-Plus colleges found that standardized tests are four times more predictive of college academic performance than high school grades. Students from less advantaged backgrounds with similar test scores showed no calibration bias. They performed at the same level as their peers from wealthier families.¹⁹ A Dartmouth study examined what happened when the test became optional. Disadvantaged students with SAT scores above 1400 were 3.6 times as likely to receive an admissions offer if they submitted their score than if they did not.²⁰ A separate study of Ivy-Plus admissions found that applicants from private nonreligious high schools were twice as likely to be admitted as those from affluent public high schools, holding test scores, race, gender, and parental income constant.²¹

The fix removed the one metric that could level the playing field. Without scores, admissions shifted weight to the remaining inputs: recommendation letters, extracurricular ratings, interviews, the intangible category called "personal qualities." Each correlates more strongly with socioeconomic status than test scores do. A student from a well-funded suburban high school arrives with a guidance counselor who writes detailed, polished recommendations. A student from an underfunded school arrives with a counselor who serves four hundred students and writes from a template.²²

The equity fix widened the equity gap. The evaluation system reorganized. Unstandardized metrics favor unstandardized advantages.

The test was visible. Measurable. Comparable across every applicant from every school. The recommendation letter is opaque. The extracurricular record is shaped by access. The interview is shaped by cultural fluency. The fix removed the visible measurement and the system reorganized around the opaque ones.²³

Carbon offset markets performed a different kind of visibility. The voluntary carbon offset market, led by Verra, the world's largest carbon credit certifier, peaked near $2 billion in 2021. Companies purchased credits to offset their emissions. Fund a forest protection project, receive a credit, claim the environmental benefit.²⁴

In January 2023, a nine-month investigation by The Guardian, Die Zeit, and SourceMaterial found that more than 90 percent of Verra's rainforest offset credits were likely phantom credits with no real climate benefit. A Cambridge University study found that the threat to the forests had been overstated by approximately 400 percent. Of twenty-nine Verra-approved projects examined, only eight showed evidence of meaningful deforestation reduction.²⁵

The market collapsed. Voluntary carbon credit prices fell more than 80 percent. Total market value dropped 56 percent in a single year. Verra's founding CEO stepped down.²⁶

The structural problem runs deeper than fraud. Offset markets require a counterfactual. How much deforestation would have happened without the project? The counterfactual is not difficult to measure. It is impossible to measure. You cannot observe what would have happened in the absence of an intervention. You can only estimate it. And the entity estimating it is the entity that profits from a higher estimate. The system produces inflated claims because the measurement itself cannot produce honest answers. The product the market sells is a number that represents something that did not happen in a world that does not exist. The question the market is built to answer is unanswerable within the market's own structure.²⁷

Gucci purchased the credits. Shell purchased the credits. easyJet purchased the credits. Salesforce purchased the credits. Each claimed the environmental benefit. The benefit did not exist. The performance of environmental responsibility was the product. The environment was the cost.²⁸

The equity fix made equity less visible by removing the one measurement that crossed socioeconomic lines. The environmental fix made environmental responsibility less real by creating a market whose product could not be verified. The privacy fix, enacted the same decade, completed the pattern. It performed the asking of a question whose answer was already determined.

The European Union's General Data Protection Regulation took effect in 2018. It required explicit cookie consent. Give users control over their data. Ask before tracking.²⁹

The system reorganized into consent theater. When a "Reject All" button appears at equal visual prominence to "Accept All," about 60 percent of users reject cookies. Only 17 percent of websites provide that option without visual manipulation. When rejection requires one additional click while acceptance requires one, 90 percent accept. The asymmetry is designed. The button that benefits the company is larger, brighter, more prominent. The button that benefits the user is smaller, grayer, hidden behind a second screen. Seventy-seven percent of major German websites use these manipulative design patterns. Fifty percent of users accept cookies out of habit, not informed choice. Seventy-two percent of young respondents report exasperation with consent popups.³⁰

The fix mandated the question but let the entity that benefits from consent design the dialog. The regulator required the asking. The company designed the answering. The result was predictable from the structure.³¹ The internet experience degraded for every user on every visit to every website. Privacy did not improve. The performance of privacy improved.

The equity performance. The environmental performance. The privacy performance. Each satisfies the demand that something be done. Whether the fix achieves its stated goal is secondary to whether it appears to. A performance needs a stage. A stage needs staff. Staff need a budget. A budget needs a threat.

The institution

The Transportation Security Administration was created on November 19, 2001. The Aviation and Transportation Security Act federalized airport screening. The purpose was to prevent another attack by screening passengers and luggage at every commercial airport in the country.³²

In 2015, the Department of Homeland Security's own Red Teams smuggled mock explosives and banned weapons through TSA checkpoints in sixty-seven of seventy attempts. A 95 percent failure rate. The TSA's acting administrator was reassigned.³³ By 2017, the failure rate had improved to approximately 80 percent.³⁴

The agency spends approximately $10 billion per year. Cumulatively, hundreds of billions have been spent since 2001.³⁵ The TSA intercepted 1,508 firearms in the first quarter of 2023, 93 percent of them loaded. The full-year figure was 6,737. But these are overwhelmingly accidental carries by travelers who forgot a weapon in their bag. They are not intercepted terrorist plots.³⁶

TSA PreCheck charges $78 for reduced screening. No shoe removal, no laptop removal, no liquid restrictions. Paying money buys you less security scrutiny, structurally undermining the stated security purpose by creating a paid bypass layer.³⁷

The Government Accountability Office reviewed the agency's operations and found that the TSA "has no efforts underway to systematically evaluate potential cost and effectiveness tradeoffs across all countermeasures." The agency that screens every passenger at every airport in the country has never systematically measured whether its screening works.³⁸

The TSA is succeeding at what it became. Its budget, headcount, and institutional survival depend on the continued perception of a threat level that justifies ten billion dollars a year. The security is the performance. The institution is the product.³⁹ The problem it was created to solve is the resource it cannot afford to lose.

The fix created a stakeholder in the problem. The stakeholder needs the problem to persist. The fixer became the obstacle to resolution. The incentive is structural. No one inside the institution needs to be corrupt or incompetent for this to operate. The structure produces the outcome.

The machine

After the First World War, France built the Maginot Line. Construction ran from 1930 to 1936 at a cost of approximately three billion francs. The fortification stretched along the Franco-German border: reinforced concrete walls up to 3.5 meters thick, retractable steel turrets, underground railways connecting bunkers, air filtration systems, and ammunition stores designed to sustain months of siege. It was the most sophisticated defensive infrastructure in military history.⁴⁰

France built it because France had learned from the previous war. The learning was complete. Twenty million casualties across four years of static warfare, trenches stretching from the Swiss border to the English Channel. The Maginot Line was the structural conclusion. Fortify the border. Deny the adversary the ability to repeat what had happened between 1914 and 1918. A triumph of analysis applied to the wrong model.

Germany went through the Ardennes. On May 10, 1940, armored divisions under Heinz Guderian moved through the forest that French planners had considered impassable to tanks. The advance bypassed the Maginot Line entirely. Six weeks later France signed an armistice.⁴¹

The Maginot Line treated the adversary as a machine that would repeat its previous behavior. The adversary was an organism. It adapted. The lesson France learned so completely became the specific vulnerability the enemy exploited. The fix channeled the attack rather than preventing it.

Every case in this piece follows the same pattern.

The opioid fix treated demand as a machine. Turn off the supply and demand stops. Demand is an organism. It reorganized around illicit fentanyl. The nuclear fix treated the grid as a machine. Remove one source and the gap stays empty until renewables fill it. The grid is an organism. It reorganized around coal, and the coal kills 1,100 people a year. The school closures treated development as a machine. Pause it, resume it later, fund the resumption. Development is an organism. The missed windows do not return. $189.5 billion could not buy them back. The TSA treated security as a machine. Screen every passenger, prevent every attack. Security is an organism. It reorganized into an institution whose survival depends on the threat.

Every fix treats the system as a machine. Every system is an organism. The organism reorganizes. The reorganization is the next crisis.⁴²

Prohibition fixed alcohol consumption. The Eighteenth Amendment took effect on January 17, 1920. For thirteen years, criminal organizations built national distribution networks, bribery systems that reached from local police to federal agents, and enforcement structures to supply the demand the law could not eliminate.⁴³ The Twenty-First Amendment repealed Prohibition on December 5, 1933. The demand for illegal alcohol disappeared overnight. The infrastructure did not. The organizations that Prohibition created shifted to gambling, narcotics, prostitution, labor racketeering, and extortion. The Five Families structure in New York, formalized during Prohibition, operated for decades after repeal. The Mafia did not dissolve when alcohol became legal. The institution the fix created outlived the fix by half a century.⁴⁴

The confidence that follows each successful fix is the most dangerous part of the cycle. The prescriptions fell. Problem solved. The plant closed. Problem solved. The test was removed. Problem solved. The consent was obtained. Problem solved. Each success generates the certainty that permits the next failure to grow unobserved. The belief that the problem was handled is the mechanism that prevents anyone from looking at what the fix is building underneath.⁴⁵

The pattern does not require modern complexity. It is as old as recorded law.

Hammurabi's Code, composed around 1754 BC, fixed penalties for surgical failure. If a surgeon operated on a nobleman with a bronze lancet and the nobleman died, the surgeon's hands were cut off.⁴⁶ The penalty addressed incompetent surgery. The system reorganized around the penalty. Surgeons refused complex cases. Patients who needed difficult operations could not find a surgeon willing to perform them. The fix for bad surgery eliminated surgery.

Diocletian's Edict of Maximum Prices, issued in 301 AD, fixed price ceilings on more than 1,400 goods and services. The penalty for exceeding them was death. Goods disappeared from open markets. Sellers refused to sell at the mandated prices. Commerce moved underground or stopped entirely. Gold prices rose 250 percent in denarii as the currency's purchasing power collapsed around the edict.⁴⁷ The fix for high prices eliminated commerce.

The French Revolution's Law of the General Maximum, enacted September 29, 1793, fixed the price of grain, flour, meat, and other essential goods at approximately one-third above 1790 levels. Farmers responded to the price ceiling by switching to crops that were not covered by the law. The supply of controlled goods fell. The bread shortage worsened.⁴⁸ The fix for hunger produced hunger.

Same mechanism. Same failure. Four thousand years. The systems were different. The fix was different. The reorganization followed the same structural logic every time. After each one, someone said: we fixed it.

And then one domain broke the pattern.

The exception

On March 27, 1977, two Boeing 747s collided on the runway at Tenerife in the Canary Islands. 583 people died. A KLM captain, one of the most experienced pilots in the airline's fleet, initiated takeoff without clearance while a Pan Am 747 was still on the runway. The co-pilot and flight engineer had information that could have prevented the collision. Neither spoke up forcefully enough to override the captain.⁴⁹

The investigation revealed a structural failure. The authority gradient in the cockpit was so steep that junior crew members could not challenge a senior captain's decisions, even when those decisions were killing everyone aboard.

The fix that followed did something different from every other fix in this piece. Instead of targeting the output, replacing the pilot, adding a rule, increasing a penalty, aviation changed the environment in which pilots operate. NASA convened a workshop in 1979. United Airlines launched the first comprehensive program in 1981. The industry called it Crew Resource Management. Authority gradients were flattened. Junior crew members were trained to challenge captains. Communication protocols were redesigned around the assumption that humans make errors and the system must catch them.⁵⁰

Black boxes did not fix crashes. They changed the feedback loop. Every failure, every near-miss, every anomaly produced specific, recoverable knowledge about what the system needed to learn. The information flowed from the crash site back into training, procedure, aircraft design, and air traffic control protocols. The system processed its own failures and converted them into structural improvements. No other domain in this piece has that feedback loop.⁵¹

Checklists did not replace human judgment. They redesigned the interaction between human judgment and system complexity. The pilot still flies the aircraft. The checklist catches the things the pilot's mind, operating under cognitive load and time pressure, will predictably miss. A surgeon washing hands before an operation does not need less skill. The handwashing protocol catches the predictable failure of human attention.⁵²

On February 12, 2009, Colgan Air Flight 3407 crashed on approach to Buffalo Niagara International Airport. All forty-nine aboard and one person on the ground died. It was the last major fatal crash of a U.S. commercial airline.⁵³

In the nine years that followed, zero passengers died on a Part 121 scheduled commercial flight. Nine years. Millions of flights. Hundreds of millions of passengers. Zero fatalities. In April 2018, an engine on Southwest Flight 1380 disintegrated during cruise and a fragment struck a cabin window. One passenger, Jennifer Riordan, was killed. Captain Tammie Jo Shults and First Officer Darren Ellisor, trained in CRM protocols, performed an emergency descent and landing at Philadelphia. 148 people survived. The system caught the failure. The crew's response was praised as exemplary CRM.⁵⁴

Every fix in this piece targeted the output. Reduce prescriptions. Close plants. Close schools. Remove the test. Offset the emissions. Screen the passengers. Each fix adjusted a parameter. The parameter moved. The system reorganized around the adjustment.

Aviation targeted the feedback structure. It changed how the system processes its own failures.⁵⁵

A parameter fix adjusts a number. The system reorganizes around the new number. A feedback fix changes how the system learns. The system reorganizes around the learning. The first kind of fix is consumed by the organism. The second kind becomes part of how the organism grows.⁵⁶

The opioid guidelines adjusted a parameter, the number of prescriptions. The system reorganized around the number. The CRM redesign changed a feedback structure, how the cockpit processes disagreement and error. The system reorganized around the learning. The guideline was consumed. The redesign endured.

The difference is the difference between every failure in this piece and the one domain that broke the pattern. The fix that works does not say "don't make mistakes." It says "when mistakes happen, the system catches them before they compound." It does not replace the broken part. It changes the conditions in which every part operates. It treats the system as an organism that will reorganize, and it designs the reorganization.

The opioid guidelines reduced prescriptions. The answer to every pilot error before Tenerife followed the same logic. Fire the pilot. Add a rule. Increase the penalty. Aviation stopped doing that. Aviation asked a different question. What environment would make the error less likely and the recovery more likely when it occurs? The answer was structural. Flatten the authority gradient. Change the communication protocols. Install the feedback loop. Let the system learn from every failure instead of just punishing the person who failed.

260.5 million prescriptions became 125.7 million. 3,105 deaths became 72,776. The prescriptions fell. The deaths multiplied.

1,100 people die in Germany every year from the air pollution produced by the coal that replaced the nuclear plants that were closed to save lives. 33 percent of American eighth graders cannot read at a basic level after schools closed to protect them. Ninety percent of carbon offset credits purchased to protect the environment were phantom credits that protected nothing. Ten billion dollars a year screens passengers through checkpoints that fail 95 percent of the time. In 1754 BC, the surgeons stopped operating. In 301 AD, the goods disappeared from markets. In 1793, the bread shortage worsened.

The fix was applied. The metric improved. The system reorganized at a level the fix could not reach. The reorganization became the next crisis. After each one, someone said: we fixed it.

The question was never whether the fix worked. The fix always works. The prescriptions fall. The plant closes. The test is removed. The consent is obtained. The passengers are screened. The question is what the fix is building. What system forms around the fix that the designers of the fix never intended and cannot see. The answer is always the same. An organism that includes the fix as part of its architecture and has vulnerabilities the old system did not have.

The antibiotic you took this year. The cookie banner you clicked through this morning. The test that was removed from the application your child submitted. The plant that was shut down in the country where your electricity comes from. The fix was applied. The metric improved.

The problem was solved. The system reorganized.