The Problem Was Always There

In 2012, Intel was the most advanced chipmaker on earth. Not by a narrow margin. By years. The company operated on a two-year rhythm it called Tick-Tock. Every cycle, a new manufacturing process, then a new chip architecture. A decade of consecutive generations without missing a beat. TSMC, the Taiwanese foundry that made chips for Apple and Qualcomm, was at least a full generation behind, further if you counted Intel's transistor architecture advantage. Samsung was further back. Intel did not just make the best processors. It made them in factories nobody else could match.¹

In 2025, TSMC is worth $1.6 trillion. Intel is worth $87 billion. TSMC is eighteen times more valuable than the company that led the industry a decade earlier.²

The story people tell about Intel is that the world changed and Intel did not. Mobile replaced desktop. Intel missed the smartphone. The PC market declined. The story sounds structural, permanent, unavoidable. It also starts in the wrong place.

Around 2005, CEO Paul Otellini turned down Apple's offer to make the chip for a phone that did not yet exist. Two years later, the iPhone launched. Apple went to Samsung, then to TSMC. That revenue stream, the largest consumer electronics product in history, funded TSMC's next generation of factories.³

In May 2013, Brian Krzanich became CEO. Within his first year, he made the decision that would determine the next decade. Intel's 10-nanometer manufacturing process would not use extreme ultraviolet lithography. EUV was the technology the entire industry was converging on, the light source that could print features smaller than anything current equipment could manage. Krzanich chose to rely on complex workarounds using older equipment instead. He believed Intel's engineers could muscle through without it.⁴

They could not. The 10-nanometer process was promised for 2016. It shipped in limited volume in 2019. Between 2014 and 2021, Intel released six generations of processors on the same 14-nanometer node. Seven years on a single manufacturing process. In the same window, TSMC shipped 7-nanometer, 5-nanometer, and 3-nanometer. Three leaps while Intel stood still.⁵

In 2016, Intel formally abandoned Tick-Tock. The two-year rhythm that had defined its dominance for a decade was replaced with a three-step cycle that was, in practice, an admission that the engine had broken.

Three decisions in a narrow window. Declining the iPhone chip. Betting against EUV. Failing to recover when 10-nanometer slipped. Each one identifiable. Each one traceable to specific people in specific years. The decline has a start date. The start date is not ancient. It is 2013.

Every company has a story about itself. Intel's story, by 2024, was that the industry had moved on. That mobile had replaced desktop. That Intel was always a PC company that never adapted. The word "always" did work that the timeline would not support. Intel was not always behind. Intel was ahead by years, and then it was not, and the transition happened inside a window narrow enough to fit on a whiteboard.

The question nobody asked

Most explanations for most problems begin with the problem as it exists today and work backward until they reach something that sounds old enough to be a root cause. If a company is struggling with retention, someone will trace it to the founding culture. If a neighborhood is unsafe, someone will trace it to decades-old demographic patterns. If a system is broken, someone will say it was always broken.

The instinct is understandable. Old causes feel structural. They feel permanent. And permanent causes relieve everyone currently in the room of responsibility. If the problem has always been there, nobody made it. If nobody made it, nobody can be asked to fix it.

There is a simpler diagnostic. Before accepting any explanation for why something is broken, check the timeline. When did the metric actually change? If the problem is recent, the cause is recent. If the cause is recent, the ancient explanation is wrong.⁶

Find when the number actually moved. Check whether the conventional explanation existed before or after the change. If the explanation predates the change by decades or centuries, the explanation does not explain the change.

It is the simplest question anyone can ask of any problem. It is also the one that is almost never asked.

The numbers that moved

In 1960, 51 percent of Black women between the ages of 15 and 44 in the United States were married and living with their husbands. By 1994, that figure was 25 percent. An absolute majority, 56 percent, had never married at all.⁷

The children tell the same story. In 1960, 22 percent of Black children were born to unmarried women. By 1994, 70 percent were. The number has barely moved since. In 2023, it was 69 percent.⁸

The homicide data tells a third version. Before the 1960s, the murder rate among Black men was declining. Between 1950 and 1961, the nonwhite male homicide rate fell from 46 to 34 per 100,000. A 26 percent decline in twelve years. After 1963, the trend reversed. By 1974, the overall homicide rate had risen more than 80 percent. By 1980, it reached 10.2 per 100,000, a postwar peak.⁹

Three measures. Marriage. Children. Violent death. Each one moving in the same direction, across the same period, after decades of moving in the opposite direction. The pattern is not subtle. It is not ambiguous. It shows up in the Census Bureau and the CDC and the FBI's crime reports. The only question is what caused it.

The conventional explanation for all three patterns is the same. Legacy of slavery. The institution that ended in 1865 is offered as the cause of disintegration that accelerated a full century later.

The timing does not fit.

If slavery caused family dissolution, the dissolution should have been worst in 1870 and improved over time as the institution receded into history. The opposite happened. The numbers were better in 1960 than at any point since, and then they collapsed. Whatever caused the collapse operated in the twentieth century, not the nineteenth. The explanation that sounds oldest is not the explanation that fits the timeline.

After emancipation

When slavery ended, freed men walked across states to find family members who had been sold away. Newspapers across the South and North carried "Information Wanted" advertisements. A man in Tennessee searching for a wife sold to Mississippi. A woman in Virginia looking for children taken to Alabama. The Freedmen's Bureau documented thousands of these searches. The historian Herbert Gutman found that in as many as three-fourths of the slave families he studied in plantation records, all children had the same mother and father.¹⁰

The contrast is painful. In the immediate aftermath of slavery, when conditions were hardest, when freed people owned nothing and had no legal infrastructure, the family held. A century later, in objectively better material conditions, with legal protections their ancestors could not have imagined, the dissolution accelerated.¹¹

The timing test does not say what caused the change. It says what did not cause it. The institution that ended a hundred years before the numbers moved cannot explain numbers that moved a hundred years later. The explanation that everyone carries, the one that has been repeated so often it feels like settled history, fails the simplest check available. When did it actually start?

Something changed in the 1960s. What it was remains debated. That the timing points there and not to the antebellum South is not debated by anyone who has looked at the numbers.¹²

The other start dates

If the timing test only worked in one domain, it would be a finding about that domain. It works everywhere.

Five percent

The belief: healthcare has always been expensive. The timeline: in 1960, the United States spent 5 percent of its GDP on healthcare. By 2024, that figure was 18 percent. The economy grew enormously across those six decades. Healthcare grew faster. It consumed three and a half times as much of the economy as it did when the Beatles were still in Hamburg.¹³

The ancient explanation is that healthcare is inherently costly, that modern medicine requires expensive technology and highly trained professionals. All of that is true. None of it explains why spending was 5 percent in 1960 and 18 percent in 2024, because the explanation existed before the change. Medicine was already expensive and technically demanding in 1960. It was also 5 percent of GDP.

The timeline points to specific policy decisions. During World War II, the federal government exempted employer health insurance from wage controls and income taxes. Coverage jumped from 12 million Americans to 70 million in a decade. The consumer was decoupled from the price. In 1965, Medicare and Medicaid were created. Spending went from 5.0 to 6.9 percent of GDP in five years. In 1997, Congress froze Medicare-funded residency training slots at 1996 levels. The American Medical Association had recommended a 25 percent reduction, fearing "physician oversupply." The cap remains largely in place today.¹⁴

The problem was not always there. The problem has three start dates, and each one corresponds to a specific piece of legislation that rearranged who pays, who decides, and how many doctors get trained.

Three and a half times income

The belief: housing has always been expensive in Toronto and Vancouver. The timeline: from 1990 through 2004, the national house price-to-income ratio in Canada sat between 3.5 and 4 times the median household income. By 2024, it was 7.7 times nationally. In Toronto, 11.8 times. In Vancouver, 11.6 times.¹⁵

The ancient explanation is that these are desirable cities with limited land. That was also true in 1998, when houses cost a quarter of what they cost now.

In 2006, Canada Mortgage and Housing Corporation began insuring 30-year amortizations, then 35-year, then 40-year. That same year, Ottawa allowed zero-down-payment insured mortgages. Immigration targets rose from under 300,000 in 2015 to 500,000 by 2023. Housing starts stayed at roughly 220,000 per year. Between 2019 and 2023, Canada gained 933,500 additional households while housing stock rose by only 763,500. A cumulative shortfall of half a million homes in less than a decade.¹⁶

"Housing has always been expensive in Toronto" is a sentence that erases a timeline the data preserves. A couple in 2000 could buy a home for four times their income. Their children, earning more in real terms, face twelve times. The explanation everyone offers is that Toronto is a desirable city. It was also a desirable city in 2000.

Eight years

The belief: the stock market has always been short-term. The timeline: in 1960, the average holding period for a share on the New York Stock Exchange was roughly eight years. By 2020, it was five and a half months.¹⁷

In 1975, the SEC eliminated fixed-rate brokerage commissions after 183 years. In 2001, decimalization replaced fractional pricing and created the conditions for high-frequency trading. Today, algorithmic trading accounts for roughly half of all U.S. equity volume. Zero-commission platforms removed whatever friction remained. In the 1960s, more than half of NYSE volume came from individual investors, people who called their broker, paid a fixed commission, and held the certificate. By 2020, much of it came from algorithms holding positions for seconds in server rooms in Mahwah, New Jersey. Each change has a date. The short-termism was not built into the structure of markets. It was built into markets by specific regulatory and technological decisions across specific decades.

Two point nine percent

The belief: productivity growth has always been slow. The timeline: from 1947 to 1972, U.S. labor productivity grew at 2.9 percent per year. From 2005 to 2022, it averaged 1.6 percent. The first slowdown arrived in 1973, the year OPEC quadrupled oil prices. Growth dropped to 1.5 percent and stayed there for twenty-three years. A temporary revival to 2.9 percent during the internet boom ended around 2005. The second slowdown has persisted since.¹⁸

"Productivity has always been slow" is wrong twice. It was not slow in 1960. It was nearly double the current rate. And the transitions from fast to slow have identifiable start dates, each one aligned with a specific shock or structural change. The first slowdown even has a name. Economists call it the productivity puzzle of 1973. Not the productivity puzzle of always.

The island

The timing test is not a modern invention. It clarifies problems that are centuries old.

In 1845, the potato blight arrived in Ireland. Phytophthora infestans, a water mold from the Americas, destroyed roughly a third of the potato crop that year and more than three-quarters the next. Over the following seven years, approximately one million people starved. Another two million emigrated. Ireland's population fell from 8.2 million in 1841 to 6.5 million in 1851. In 2024, the population of the island stands at roughly 7.2 million. It has never recovered. One hundred and eighty years later, the country still has fewer people than it did before the blight arrived.¹⁹

The conventional explanation is the blight. The potato crop failed and the people starved. But the blight was a trigger. The cause was a structure that made the trigger lethal.

Over three million Irish people were entirely dependent on the potato for food. A typical family consumed roughly eight pounds per person per day. Irish potato consumption per person dwarfed any comparable European population. This dependency was not an accident of culture or preference. The Penal Laws, enacted between 1691 and 1793, made it illegal for Catholics to purchase land, inherit it normally, or hold leases longer than 31 years. By 1778, Catholic ownership of Irish land had fallen to 5 percent. By the eve of the famine, roughly ten thousand absentee landlords, most of them living in England, owned the vast majority of the island. By 1870, seven hundred and fifty families would own half of it.²⁰

While a million people starved, Ireland exported food. In 1847, the worst year, roughly four thousand ships carried grain, cattle, butter, and whiskey from Irish ports to England. That year, exports included nearly ten thousand calves, a 33 percent increase over the previous year. The country exported more live animals between 1846 and 1850 than the number of people who emigrated.²¹

The timing test asks a specific question. The same blight hit Belgium and the Netherlands in 1845. Belgium lost nearly 90 percent of its potato harvest that year, a worse initial loss than Ireland's. But Belgium's death toll was roughly 44,000. The Netherlands lost a similar proportion. Their death toll was roughly 53,000. Ireland lost a million.

The blight did not choose Ireland. The blight hit northern Europe. It produced a catastrophe in only one place. The variable was not the blight. The variable was the structure the blight revealed. The land system, the crop dependency, the export economy, the government response. Each one a decision, traceable in time, built by identifiable people over identifiable centuries. The famine had a trigger that arrived in 1845. But the cause had start dates in 1650, in 1691, in 1793. The blight was the match. The powder had been laid for generations.

Where the explanation fits

The timing test does not always overturn the conventional story. Sometimes it confirms it.

The Sahara Desert has constrained sub-Saharan African trade routes for millennia. The geographic barrier existed before any metric anyone recorded and persists today. No policy created it. No decision in a specific decade produced it. The ancient explanation is the explanation. Corruption in southern Italian institutions traces to at least the early nineteenth century, with roots arguably centuries older. The pattern has been documented across every decade. Not a recent development with an ancient excuse. Genuinely structural. Genuinely old.

This is what makes the method honest. It is a diagnostic, not a verdict. It asks when the number moved. If the number moved recently, the ancient explanation is wrong and the recent cause is findable. If the number never moved, if the pattern is as old as the explanation claims, the timing test confirms it. The method does not arrive with a conclusion. It arrives with a question. And sometimes the answer is that the problem really was always there.

The start date

None of these problems were always there. Each one has a start date. A CEO's bet in 2013. A tax exemption in 1942. An amortization rule in 2006. A commission structure in 1975. A land confiscation in the 1650s. The start date points at a cause. The cause points at a decision. The decision was made by someone, somewhere, in a year that can be checked. And in every case, the explanation that preceded the check was older, more comfortable, and wrong.

The sentence "it was always like this" does a specific kind of work. It turns a decision into a condition. It turns a policy into a fact of nature. It locates the cause in the permanent past, where no one is responsible and nothing can be changed. The timing test reverses the operation. It locates the cause in the recent past, where the decisions were made by people who can be named, in rooms that still exist.

The problem was always there. Except it was not. It had a start date. The start date pointed at a cause. The cause was recent. And the ancient explanation, the comfortable one, the one that relieves everyone in the room of responsibility, was wrong.