The Ice Age We Cancelled

Here is a fact that sounds like it was written by a committee of nihilists but is, regrettably, just physics: the planet Earth was in the middle of an ice age when we arrived, and we cancelled it.

Not postponed. Not mitigated. Cancelled. The way you cancel a dentist appointment you never intended to keep. Except the appointment was the next glacial period, it was pencilled in for about 50,000 years from now, and we blew past the cancellation window sometime around 1965.

The western snowpack collapsing outside your window right now — the record lows, the bare mountains, the fire seasons that now start in March — is not a blip. It is not a cycle. It is the visible surface of the largest intervention in the Earth system since an asteroid killed the dinosaurs. Except the asteroid didn’t know what it was doing. We do.

This is that story. It involves orbital mechanics, ice cores, a very patient ocean, and the uncomfortable mathematics of what happens when you dig up 500 million years of dead plants and set them on fire in 200.

Part I: We Were Supposed to Be Cooling

The Quaternary Ice Age began 2.58 million years ago and, despite the name, it has not ended. You are living in it right now. We are in an interglacial period called the Holocene — a relatively warm intermission between glacial advances — that started about 11,700 years ago when ice sheets retreated from Chicago.

The timing of these intermissions is not random. It is governed by three orbital wobbles that a Serbian mathematician named Milutin Milanković worked out in 1920, by hand, while a prisoner of war:

• Eccentricity: Earth’s orbit stretches from nearly circular to slightly elliptical every ~100,000 years. Currently 0.0167 and decreasing — nearly circular.
• Obliquity: Earth’s axial tilt wobbles between 22.1° and 24.5° every ~41,000 years. Currently 23.44° and decreasing.
• Precession: Earth’s axis traces a slow circle every ~21,000 years, like a spinning top winding down.

When these three cycles align to reduce summer sunlight at northern latitudes, winter snow survives into the next year. Snow reflects 80–90% of sunlight. More snow means more reflection. More reflection means more cooling. More cooling means more snow. A positive feedback loop kicks in, and within a few thousand years you’ve got an ice sheet two miles thick sitting where Toronto is.

The key insight — the one that should worry you — is that ice ages are not triggered by cold winters. They are triggered by cool summers. If summer isn’t warm enough to melt last winter’s snow, the process begins. One surviving snowfield becomes two, becomes a glacier, becomes a continental ice sheet.

By every orbital metric, Earth should be cooling right now. Summer sunlight at 65°N has been declining for 6,000 years. The Holocene Climatic Optimum — the warmest natural point of our interglacial — was roughly 8,000 years ago. Since then, a slow neoglacial cooling trend has been nudging us back toward glaciation.

As NASA puts it: “Our planet should be cooling, not warming.”

The temperature record shows this reversal in brutal clarity. 131 years of western summer temperatures, drifting upward — accelerating since the 1980s:

We will get to why it is not cooling in a moment. First, let us discuss a number.

Part II: The Number

280.

For roughly 10,000 years — the entire span of human civilisation, from the first wheat fields of the Fertile Crescent to the moon landing — atmospheric CO₂ sat at about 280 parts per million. It wiggled up and down by maybe 10 ppm. It was, for all practical purposes, stable. Human beings invented writing, built the pyramids, discovered calculus, and composed Bohemian Rhapsody, all at 280 ppm.

As of May 2026, that number is 429 ppm.

A 53% increase. In 270 years. On a planet where the previous rate of change — during the fastest natural CO₂ swings at the end of ice ages — was about 0.01 ppm per year.

We are currently adding CO₂ at 3.75 ppm per year. That is roughly 200 times faster than the fastest natural rate recorded in 800,000 years of ice core data.

If that comparison doesn’t land, try this one. During the last deglaciation, it took nature roughly 6,000 years to raise CO₂ by 100 ppm. We did it in under 120 years. Then kept going.

The last time CO₂ was at today’s level was 14 million years ago, during the mid-Miocene. There were no ice sheets in Greenland. Trees grew at the edge of Antarctica. Sea levels were 15–25 metres higher than today. Earth at 425 ppm is not the Earth you were born on. We just haven’t finished finding out what it looks like yet — the ice sheets are still catching up.

Part III: The Long Tail (or, Carbon Is Forever)

Here is where it gets uncomfortable. Not “we should probably recycle more” uncomfortable. “We have altered the trajectory of the next 100,000 years” uncomfortable.

David Archer, a geophysicist at the University of Chicago, has spent his career modelling what happens to CO₂ after we emit it. His finding, published across a series of papers culminating in his book The Long Thaw, is this:

“CO₂ sticks around for hundreds of years, plus 25% that sticks around forever.”

About 75% of emitted CO₂ is absorbed by the ocean over a few centuries. But 20–25% remains in the atmosphere for tens of thousands of years. The mean atmospheric lifetime of the anthropogenic CO₂ pulse, weighted by this long tail, is approximately 30,000 years. For reference, we usually associate that kind of longevity with nuclear waste.

In 2016, Andrey Ganopolski and colleagues at the Potsdam Institute published a landmark paper in Nature titled “Critical insolation–CO₂ relation for diagnosing past and future glacial inception.” They modelled the precise relationship between orbital parameters and CO₂ concentration required to trigger the next ice age. Their conclusion:

• Without human emissions, the next glacial period would begin in approximately 50,000 years.
• With current and projected emissions, it is delayed to approximately 100,000 years.
• We have extended the interglacial by roughly 50,000 years.

Ganopolski put it plainly: “Anthropogenic climate change will last not hundreds, but hundreds of thousands of years.”

I want to sit with that sentence for a moment. Every civilisation that has ever existed — Sumer, Egypt, Rome, the Tang Dynasty, the British Empire, whatever we are doing now — fits inside the last 5,000 years. We have altered the climate system for a duration ten times longer than the entire history of civilisation.

And we did it mostly since 1950.

Part IV: What the Snow Is Trying to Tell You

Let us bring this down from orbital mechanics to the mountains you can see from your kitchen window.

On April 1, 2026, 1,012 of 1,570 SNOTEL stations across the western United States — 64% — recorded their all-time lowest snowpack. Eight states set records simultaneously. As of May 15, Washington, Oregon, and Idaho report 0.0 inches of snow water equivalent. The snowpack didn’t melt late. It barely formed at all.

This is not a drought in the traditional sense. Precipitation was close to normal in many areas. But temperatures were high enough that the precipitation fell as rain instead of snow. The water arrived, ran off immediately, and was gone by March. Hydrologists call this a warm snow drought — and it is becoming the dominant mode.

Every line in this chart is one year of real snowpack data. The red line at the bottom is 2026:

At Mt. Rainier, the meadows of Paradise have lost 98.7% of healthy vegetation in four summers. The snowpack that sustained them is gone.

A 2024 study by Zhuang and colleagues in Science Advances found that since 2000, rising temperature and evaporative demand have driven 62% of drought severity and 66% of drought coverage across the western US. Before 2000, temperature contributed only 31%. The drought regime itself has changed. It is no longer about whether it rains. It is about whether it is too hot for the rain to matter.

How much of this is us? The science on attribution has converged:

• Pierce & Barnett (2008): ~50% of observed snowpack decline from 1950–1999 is attributable to anthropogenic greenhouse gases.
• Barnett et al. (2008), Science: Up to 60% of climate-related trends in river flow, temperature, and snowpack are human-induced.
• Williams et al. (2020), Science: ~42% of soil moisture deficit since 2000 is human-caused. The 2000–2021 period is the driest 22 years since at least 800 CE.
• Gottlieb & Mankin (2024), Nature: “Virtually certain” (>99% probability) that human emissions have contributed to observed snowpack losses. They identified a “snow-loss cliff”: above -8°C average winter temperature, losses accelerate nonlinearly.

That last finding is the one that keeps hydrologists up at night. Eighty percent of Northern Hemisphere snow mass currently sits in regions cold enough to be below the cliff. The other 20% — concentrated near the population centres that depend on snowmelt for drinking water — has already begun falling off it.

Part V: The Feedback Loop (or, Everything Is on Fire and That Makes It Worse)

Less snow means earlier melt. Earlier melt means drier forests. Drier forests means more fire. More fire means more CO₂. More CO₂ means higher temperatures. Higher temperatures means less snow.

This is not a metaphor. It is a measured feedback loop, and the numbers are not small.

In 2020, California’s wildfires released approximately 127 million metric tons of CO₂ — roughly twice the state’s total emission reductions from 2003–2019. Seventeen years of climate policy, undone in one fire season.

Globally, forest fires released over 8 billion tonnes of CO₂ between March 2024 and February 2025 — roughly equal to the combined annual fossil fuel emissions of the United States and European Union.

Abatzoglou & Williams (2016) calculated that human-caused climate change has doubled the forest area burned in the western US since 1984, adding 4.2 million hectares of fire that would not have occurred without anthropogenic warming. Fire season has lengthened by 84 days since the 1970s. Average fire duration has increased from 7.5 days to 37.1 days.

The U.S. Forest Service is being dismantled in real time — 6,000 employees lost, 57 research facilities closing — just as the crisis accelerates. Lake Powell is receiving 13% of normal inflows. Hydropower could halt by September.

Meanwhile, the Arctic has warmed nearly four times faster than the global average since 1979 — significantly exceeding model predictions. This weakens the temperature gradient that drives the jet stream, causing it to develop larger, slower meanders. When these meanders stall, you get blocking patterns: persistent ridges that park high pressure over the western US for weeks, creating the heat domes that cook forests and melt snowpack simultaneously.

The system is not breaking down. It is working exactly as physics says it should. We changed the inputs. The outputs changed accordingly.

Part VI: The Denier’s Handbook (Annotated)

If you have read this far, you have likely encountered some of these arguments. Let us address them with the respect they deserve, which is: precisely the amount warranted by the evidence.

Carl Sagan had a phrase for the posture of denial in the face of evidence: “One of the saddest lessons of history is this: if we’ve been bamboozled long enough, we tend to reject any evidence of the bamboozle.”

The evidence is not ambiguous. It is not a matter of perspective. 8,000 SNOTEL stations, 800,000 years of ice cores, 131 years of temperature records, 49,072 mapped fire perimeters, and the orbital mechanics of the planet itself all tell the same story. You can argue with a person. You cannot argue with a planet.

Part VII: The Review

John Green, in The Anthropocene Reviewed, rates aspects of the human experience on a five-star scale. It is a generous framework, because it forces you to find the humanity even in the things that are destroying us. In that spirit:

Part VIII: What Now

The science is clear on what happens next if nothing changes. It is also clear on what happens if something does.

If all emissions went to zero tomorrow, temperature would stabilise near current levels within a decade. The planet would not continue warming dramatically. The CO₂ already in the atmosphere would not vanish — 25% of it will still be there in 10,000 years — but the warming would stop accelerating. This is the ZECMIP finding, and it is, in its way, the most hopeful number in climate science: the warming stops when the emissions stop.

But the emissions have not stopped. They are accelerating. And the things already set in motion — ice sheet dynamics, sea level rise, ecosystem transitions, the shift from cold droughts to warm droughts — are largely irreversible on any timescale that matters to a human life.

So what do we do?

We do what the data says. We track the snowpack. We monitor the fires. We measure the drought. We publish it all, for free, so that the people who live in the path of these changes — the 80 million people in the western fire corridor — have access to the same information that the agencies and insurers who make decisions about their lives do.

We count what counts. And we do not look away.

That is what this platform is for.

Further Reading

This article is the hub. Each of these goes deeper into a specific piece of the crisis:

Sources