You remember Hiroshima. August 6, 1945. The first atomic bomb. You probably remember Nagasaki too. August 9, 1945. The second.
But Kyoto almost came first.
In the closing months of World War II, Kyoto sat at the top of America’s target list for the atomic bomb. It was a perfect target: a cultural and intellectual center, a city of 1 million people, untouched by conventional bombing. Military planners believed its educated population would understand the weapon’s significance and force Japan’s surrender.
Then Henry Stimson, the U.S. Secretary of War, struck it from the list.
Not for strategic reasons. Stimson had visited Kyoto in 1926 for a honeymoon and remembered the temples, the gardens, the cultural weight of Japan’s ancient capital. He told General Leslie Groves, head of the Manhattan Project: “I don’t want Kyoto bombed.” He held that line against repeated pressure, insisting the city’s destruction would breed lasting hostility toward America in postwar Japan.
Hiroshima burned instead. Then, three days later, Nagasaki.
But Nagasaki wasn’t even supposed to be the second target. That was Kokura, home to Japan’s largest arsenal. On the morning of August 9, the B-29 bomber Bockscar arrived over Kokura at 10:44 AM. Clouds and smoke from the previous night’s firebombing obscured the target. The crew circled three times. Forty minutes. Fifty minutes. Japanese anti-aircraft fire intensified. Fuel ran dangerously low because of a malfunctioning pump.
At 10:45 AM, they abandoned Kokura and flew 100 miles southwest to the secondary target.
Nagasaki was cloudy too. The crew prepared to dump the bomb in the ocean. Then, at 11:01 AM, the clouds parted for less than a minute. The bombardier saw the factory district. At 11:02 AM, Fat Man fell.
Seventy thousand people died because of smoke over Kokura and a break in the clouds over Nagasaki.
A single memory of temples. Weather. A fuel pump. These redirected the course of history for millions of people.
The Tyranny of Precision
It’s tempting to think that history bends under grand forces: economics, geopolitics, ideology. But the hinge is often smaller. A lunch meeting. A personality. A coin toss. A misplaced assumption.
And yet, every year, the world’s most respected institutions release precise five-year forecasts of what the global economy will look like. The IMF. The World Bank. The OECD. Consultancies whose names inspire boardroom confidence.
They’re not wrong because they’re careless. They’re wrong because Kyoto happens every day. The unmodelled, the arbitrary, the contingent.
How the Numbers Drift
I was curious how this unfolds across the mountain of IMF and World Bank data. So I looked at it.
Across decades of the IMF’s World Economic Outlook, the five-year-ahead error averages 1.6 percentage points in world GDP growth. Emerging economies miss by over 2 percentage points. Some regions by 3 to 4 percentage points.
If that sounds tolerable, recall that global growth itself averages only 3%.
A 1.6 percentage point deviation means forecasts miss by more than half the thing they’re predicting. Compounded over five years, that creates a 9% shortfall in total output. Enough to rewrite debt trajectories. Enough to rewrite fiscal histories.
The math of error follows a quiet physical law.
The Physics of Diffusion
Forecast uncertainty behaves like diffusion. The same law that governs how ink spreads in water. Think of dropping a single drop of ink into a glass of water.
At first it forms a tight swirl, easy to trace.
Give it a minute, and it begins to blur.
Wait longer, and the ink spreads until it’s everywhere and nowhere — you can’t say where the original drop went.
That’s what happens to economic forecasts.
The longer into the future you look, the more the original signal dissolves into noise.
By year five, what you’re seeing isn’t the future. It’s the fog your prediction created as it spread.
The variance of error grows with the square root of time:
σₜ = σ₁√t
If a one-year-ahead forecast has a 1 percentage point standard error, the five-year-ahead error isn’t five times larger. It’s roughly √5 ≈ 2.2 × 1 percentage point = 2.2 percentage points.
Uncertainty doesn’t expand linearly. It diffuses, slowly at first, then unstoppably.
By year five, the confidence band becomes so wide that the midpoint (say, 3% growth) loses meaning. We’re not measuring the future anymore. We’re measuring our hope.
The Mirage of Control
Every major forecast misses for the same reason earthquakes surprise geologists and revolutions surprise governments: the future isn’t linear.
Economies don’t drift smoothly. They jump.
The forces that actually drive change are technological ruptures, climate events, wars, policy shocks. They’re nonlinear, interdependent, and rare. They dominate outcomes precisely because they’re not frequent enough to fit the model.
The economist’s tragedy is statistical: what happens once in twenty years shapes the next twenty.
Why We Still Pretend to Know
Forecasting is a social performance as much as a mathematical one.
A minister can’t tell parliament, “We don’t know what GDP will be.” A central bank can’t release a statement saying, “Everything depends on luck.” So institutions produce precision. Not because they believe it, but because someone has to draw the line in the fog.
That’s not deceit. It’s civilization’s coping mechanism. Certainty is comforting, even when it’s decorative.
The Lesson of Kyoto
The Kyoto anecdote reminds us that chance isn’t noise. It’s a fundamental force shaping life and events.
The general’s honeymoon wasn’t a rounding error in the grand narrative of war. It was a causal node with irreversible consequences.
Forecasting errors tell the same story in mathematical form. The world doesn’t diverge from prediction because the models are bad. It diverges because the world is built on small contingencies that cascade.
Embracing Chance
Black swans aren’t exceptions. They’re the architecture of reality.
The economy doesn’t run on averages. It runs on outliers. Every model’s error term hides the true author of history: accident.
We should stop pretending that uncertainty is a flaw to be corrected. It is the baseline condition of human systems.
To forecast, then, is not to predict the future. It is to acknowledge that Kyoto is always waiting somewhere in the model, unseen, ready to tilt the axis of the world.