Catastrophe is more certain than calm
The quiet we were born into
The calm is not the ordinary state of things, occasionally interrupted. It is the interruption. Everything anyone has ever called normal happened inside one still stretch of about twelve thousand years, and against the record the planet keeps of itself, that stretch is the anomaly.
One
What came before the quiet.
The Greenland ice does not hold the last ice age as a long cold plateau. It holds a series of lurches: temperatures over Greenland moving between 5 and 16 degrees within a few decades, over and over, with the North Atlantic and east Asia shifting together inside five to ten years of each other.
Sixteen degrees in a few decades is the difference between Hobart and Cairns, arriving inside one person's life and then leaving again.
The lurches are called Dansgaard-Oeschger events, and their larger cousins Heinrich events. They run through the last glacial period at roughly fifteen hundred year spacing. The Holocene has none of them, and that flatness is the thing wanting an explanation, not the shaking. The glacial rhythm underneath is orbital, and even there the strongest beat in the record sits where the orbital forcing is weakest, which the field calls the 100,000-year problem.
Two
A child of the lull.
People were anatomically modern long before the Holocene: toolmaking, painting caves, sailing to Australia. What they did not do anywhere, across tens of thousands of years, was farm.
The standard account in the literature is not that nobody thought of it. A field is a bet on next season resembling this one, and in a climate that can move ten degrees inside a working life, that bet loses. Richerson, Boyd and Bettinger put the case in 2001 under the title Was agriculture impossible during the Pleistocene but mandatory during the Holocene?: once the climate held still, farming appeared independently in several parts of the world, and once the neighbours farm, the choice is gone.
Which puts it plainly
Every crop, village, road, written word, law and machine sits inside the one stretch calm enough to allow it. Civilisation did not arrive and then get good weather. The weather held, and civilisation is what grew in it.
Three
Catastrophe is the ordinary case.
The sentence carrying most of the weight against all of this is that catastrophe is an extraordinary claim. Exoplanet astronomy has made that the wrong category. Neither table below asks anything about the Earth's past.
| The event | Where it stands elsewhere |
|---|---|
| A planet thrown out of its system | Free-floating planets are found by microlensing and directly by JWST, and both simulation and observation suggest they may dramatically outnumber bound planets below Earth mass. Ejection by scattering, by a passing star, or by a binary companion is a standard outcome. |
| Endless day on one face, endless night on the other | Tidal locking is the expected condition for terrestrial planets around M dwarfs, and M dwarfs are the commonest star in the galaxy. Permanent day and permanent night is arguably the normal planetary condition, and Earth's daily rotation the odd case. |
| A star flaring hard enough to rework a planet's air | Flares from an active M dwarf drive ozone up by a factor of about twenty against a quiet star, and raise night-side middle-atmosphere winds by tens of metres per second. Work on young-star flares and early exo-Earth atmospheres was still being published in 2026. |
| Two planets hitting each other | Observed. The afterglow of a giant planet collision has been caught, with further collision evidence reported in 2026. Per system the largest impacts are rare; across a few hundred billion systems, rare is a busy schedule. |
Liu Cixin's game inside The Three-Body Problem kills its civilisation a different way each time. Read as a list of failure modes rather than as a story, the entries have since been photographed happening to something real.
| How the world ends in the game | Where it stands in the journals |
|---|---|
| Orbits nobody can predict, stable spells that end without warning | The three-body problem is chaotic with no general closed-form solution, which is a mathematical result rather than a plot device. Planets are found in double and triple star systems. |
| The planet falls into a sun | Caught in the act. A survey picked up the optical flash of a star swallowing its own planet, with dust thrown off as the planet spiralled into the star's outer layers. |
| Everything torn up and swallowed at the end | Between a quarter and a half of all white dwarfs carry iron, calcium and magnesium in their atmospheres that can only be eaten planetary material. X-rays have been seen from that debris heated to a million degrees as it lands, and around one star, fragments have been watched breaking up in transit. |
| Burned by the sun coming too close | Observed. Planets are found evaporating under their star, losing mass, with the vapour recondensing into dust trails behind them. |
| Frozen, and thrown out of the system entirely | Free-floating worlds, ejected from their systems, found by microlensing and directly by telescope. |
| Space itself collapsed from three dimensions to two | Fiction, and this site will say so rather than blur it. No observational analogue. The holographic principle is a separate and heavily worked line in theoretical physics, and a different claim. |
Which moves the burden
Every entry but the last has a paper behind it, and the last is marked because marking it is what makes the rest usable. What is unusual is not the catastrophe. It is our own uneventful stretch, which is a fact about the length of the record.
Four
A default with a losing record.
Geology spent the nineteenth century choosing between sudden violent events and present rates applied over long enough. The second won, and won completely enough that catastrophe became a marker of not being a serious person.
| The claim | Its reception | Where it stands |
|---|---|---|
| The scablands were cut by a catastrophic flood, J Harlen Bretz, 1923 | Attacked for decades. The evidence was good and the mechanism was missing. | Accepted. Glacial Lake Missoula, releasing on the order of 2,000 cubic kilometres in days, about four thousand Sydney Harbours. |
| The continents move, Alfred Wegener, 1912 | Dismissed for roughly fifty years, again largely for offering no mechanism. | Accepted. Plate tectonics is the organising frame of the science now. |
| An asteroid impact ended the Cretaceous, Alvarez et al., 1980 | Resisted through most of the eighties in favour of gradual volcanic and sea-level accounts. | Accepted, with the crater found and the debris layer traced worldwide. |
Which is narrow, and holds
None of that makes any particular catastrophic claim true, and Bretz was right because he had coulees, giant ripple marks and erratics, not because he was bold. What it establishes is that a field's default setting was overturned three times in a century, on its biggest questions, by the same kind of claim each time. The prior is not neutral, and the burden of proof being applied is asymmetric.
Five
And the room is not sealed.
The Solar System gets pictured as a clock in a room, keeping its own time while the weather outside changes. Published work has taken that picture apart.
About two million years ago the Solar System crossed a cold dense interstellar cloud, put at around 3,000 particles per cubic centimetre against roughly 0.2 in the Sun's present neighbourhood. That pushed the heliosphere inside the Earth's orbit and left the planet descreened, sitting in interstellar plasma rather than in the solar wind. Both halves of the number carry weight: the ratio is enormous and changes the physics, while the absolute density is still a better vacuum than any laboratory on Earth can make. A change of medium, rather than a wall of gas arriving.
That medium is a plasma, so it carries current, and the heliosphere is a measured electrical filter: interstellar grains below about 0.03 micrometres are turned away entirely. Collapse the heliosphere and the filter is gone.
A visitor is on the record too: about seventy thousand years ago a small red star, Scholz's, passed within about eight tenths of a light year, five times nearer than the closest star sits now, and went through the outer edge of the Solar System's comet shell. It was unknown until 2015.
Where this stops
Those effects are gravitational and play out over millions of years rather than centuries. What the work establishes is narrower and worth holding on its own: the assumption that the Solar System sits still while its surroundings change was an assumption, it was load-bearing, and it turned out to be wrong.
Six
What arrives, and how often.
No row here needs a micronova, a cycle, or anything contested. Each is a mainstream hazard with a published return period, and each one on its own would take the grid down or the harvest with it.
| What | Roughly how often | Last one |
|---|---|---|
| Carrington-class solar storm | Every 100 to 200 years on one reading, though the auroral record holds at least four since 1770, a higher rate than that | 1859, and again in 1872, and 1921. The May 2024 storm, the first severe one since 2003 and the largest satellite manoeuvre event on record, was under half their size. |
| Miyake-class particle event, roughly ten times bigger again | About ten across the last 15,000 years, so near one per 1,500 years | 993 CE, over a thousand years ago |
| Superflare on a star like ours | About once a century, per star | Not known for the Sun |
| VEI-7 eruption, the Tambora scale | One or two per thousand years, and the record probably undercounts | Tambora, 1815, the year without a summer |
| VEI-8 eruption, the Toba scale | Tens of thousands of years, and irregular | Toba, ~74,000 years ago |
| Geomagnetic excursion, on the accepted count alone | About 13 across the last 780,000 years | Laschamp, ~41,000 years ago, field down as much as 90% |
| Abrupt climate lurch of 5 to 16 degrees inside decades | Roughly every 1,500 years, through the last ice age | None in the Holocene, which is the anomaly |
Seven
We have not been safe. We have been lucky.
Storms of the kind that would take a modern grid apart are not rare in the Sun's terms. They turn up roughly every solar maximum. What has saved us so far is not scarcity. It is aim.
A large eruption leaves from wherever the active region happens to sit. The published arrival work finds that the ones reaching Earth come from a band of about eighty degrees of solar longitude, and that roughly three in five launched from inside that band arrive. Eighty degrees out of three hundred and sixty is about a fifth of the Sun, so a large eruption has something like a one in eight chance of being pointed at us. The Sun turns once a month, so a long-lived active region sweeps that band again.
| Over | Chance of at least one direct hit |
|---|---|
| A decade | 10 to 12 per cent |
| A lifetime | 41 to 47 per cent |
| A century | 65 to 72 per cent |
| Two centuries | 88 to 92 per cent |
On the mainstream published rate, a Carrington-class storm reaching the grid inside a hundred years is more likely than not. Taking that one row and treating arrivals as random, the chance of a civilisation meeting none of it across three centuries comes out between three and four per cent.
Held honestly
The eighty-degree window is longitude only. Earth sits within about seven degrees of the solar equator and active regions drift equatorward through a cycle, so one in eight is an order of magnitude rather than a precision figure. The per-decade rate has been argued over, with published estimates running from well under one per cent up to twelve depending on method. The table takes the high end. At the low end a century still comes out around one in twenty.
Eight
The register, and what follows.
Twelve thousand years of calm is a windfall, not a countdown. It produced farming, cities, writing, music and medicine.
What follows is the ordinary conclusion about a house in a place that floods. It has happened before, it will happen again, nobody knows which year, and floors get built high in places like that.
What is worth building, sorted by response rather than by cause →