How Long Could Humanity Last? — A Cosmic Inventory of Things That Could End Us

The Time We Almost Did Not Make It

Roughly seventy-four thousand years ago, in what is now Sumatra, the volcano Toba produced the largest volcanic eruption of the past two and a half million years. The Toba supereruption ejected an estimated 2,800 cubic kilometers of magma — enough material to bury Spain to a depth of five meters — and pumped sulfate aerosols into the upper atmosphere on a scale large enough to depress global average temperatures by an estimated three to five degrees Celsius for several years.

Genetic evidence first published by Stanley Ambrose in 1998 and refined in subsequent work suggests that the early human population went through a severe bottleneck around the same time. Estimates of the surviving breeding population vary, but the consensus range is somewhere between three thousand and ten thousand individuals, distributed across only a few isolated refuges, primarily in eastern and southern Africa. Every human being now alive — every population on every continent — descends from that surviving group.

The Toba bottleneck is not, in modern phrasing, a near-miss. It is the closest our species has come to extinction in our genus's three-hundred-thousand-year history, and the cause was a single geological event we did not see coming and could not have prevented if we had. The bottleneck is also, to a first approximation, the baseline: this is what cosmic and geological catastrophes can do to a global population that has no industrial infrastructure to lose. We were the population that had nothing to lose.

The Things That Could End Us, by Time Scale

The following list is not speculative. It is a catalog of phenomena that have happened, are happening, or are statistically certain to happen, ranked by the timescale on which the next instance is expected. It is drawn from the peer-reviewed literature on existential risk, planetary defense, and stellar dynamics. Estimates have wide error bars; the scales are reliable to within an order of magnitude in most cases, two in a few.

On the scale of years to centuries: extreme space weather (Carrington-class events), severe supervolcano eruptions (the historical record contains roughly one VEI-7 eruption per millennium), and the impact of an asteroid in the 50–150 meter size class. Each of these has happened within the past hundred thousand years; none would, in itself, end humanity, but each could substantially reduce industrial civilization's ability to function for periods measured in years to decades.

On the scale of thousands to tens of thousands of years: the impact of a kilometer-class asteroid. NASA's near-Earth object surveys have so far identified roughly 95 percent of the estimated population of 1-kilometer or larger near-Earth asteroids, and none of the known objects are on collision courses with Earth in the next century. The remaining 5 percent — and the unknown population of long-period comets, which surveys cannot yet inventory completely — represents the residual risk.

On the scale of hundreds of thousands to millions of years: stellar flybys. The Sun's neighborhood in the galaxy is in constant motion. Stars pass within a few light-years of each other on long timescales, and a sufficiently close passage can gravitationally perturb the outer Oort Cloud, the spherical reservoir of long-period comets that surrounds the solar system. A perturbed Oort Cloud sends a fraction of its members on trajectories that fall toward the Sun, raising the cometary impact rate at Earth for periods measured in millions of years.

On the scale of tens of millions to hundreds of millions of years: the impact of a 10-kilometer-class object. The asteroid that produced the Cretaceous–Paleogene extinction sixty-six million years ago was in this class. The fossil record records about half a dozen mass extinctions of comparable magnitude over the past five hundred million years, attributable to a mix of impacts, large igneous province volcanism, and climatic shocks the causes of which are still debated.

On the scale of a few billion years: the gradual brightening of the Sun, the loss of Earth's oceans through hydrogen escape, and the eventual loss of habitability for any complex surface life. These are subjects we have covered separately. They are not on the human timescale and we are not, in any meaningful sense, threatened by them.

Gliese 710 — The Star That Is Coming

Of the entries on the list, the one with the most precisely measured arrival time is the close passage of the K-dwarf star Gliese 710. Gliese 710 is a star roughly six tenths the mass of the Sun, currently sixty-three light-years away in the constellation Serpens, moving toward the solar system at about 12 kilometers per second.

Its trajectory was first inferred from Hipparcos astrometry in the 1990s. The trajectory was substantially refined in 2016 by Filip Berski and Piotr Dybczyński of Adam Mickiewicz University in Poznań, who combined Hipparcos data with the first major release of the European Space Agency's Gaia mission. Their paper in Astronomy & Astrophysics reported a closest-approach distance of 13,365 astronomical units — about 0.21 light-years — at a time approximately 1.35 million years from now.

That was a record-breaking close-approach calculation when published, and it has since been refined again with later Gaia data releases. The 2018 update, using Gaia DR2, tightened the calculation to a closest approach of approximately 0.06 light-years — about 4,000 astronomical units — placing Gliese 710 well inside the Sun's hypothesized inner Oort Cloud at closest approach.

The implications are not direct. Gliese 710 will not collide with the Sun, and at closest approach it will appear in our descendants' sky as one of the brightest stars visible — as bright as Mars at opposition. The danger is gravitational. A star of Gliese 710's mass passing that close to the Oort Cloud would scatter cometary orbits over a period of hundreds of thousands of years following the encounter, raising Earth's long-period comet impact rate during the perturbation episode. The expected rise is small in absolute terms but non-trivial: an additional ten to twenty cometary impactors of the same size class as Comet C/1995 O1 (Hale-Bopp) over the perturbation period, on top of the background rate.

The First Test of Whether We Could Stop Anything

On September 26, 2022, NASA's Double Asteroid Redirection Test — DART — became the first space mission in human history to deliberately alter the orbit of a celestial body. DART, a half-ton spacecraft launched in November 2021, struck the small moonlet Dimorphos as it orbited the larger asteroid Didymos. Dimorphos is roughly 160 meters across — a size class that would, if it struck Earth, produce regional damage measured in the hundreds of kilometers.

The collision was at 6.1 kilometers per second. The intended outcome was to shorten Dimorphos's twelve-hour orbital period around Didymos by a few minutes — a measurement of the kinetic deflection efficiency. The actual outcome, measured by ground-based telescopes in the weeks after the impact, was a shortening of the orbit by 32 minutes — roughly four times the predicted value, because Dimorphos turned out to be a loosely bound rubble pile rather than a solid object, and the impact ejected enough material that the recoil substantially boosted the deflection.

That was, in narrow terms, a success. In broader terms, it demonstrated that a kinetic deflector launched a few years before a known impact could in principle move a 150-meter asteroid out of an Earth-crossing orbit. It also demonstrated how much the outcome depends on details of the target's structure that we cannot easily measure from Earth — and therefore how much the system needs follow-up observations and, for any real planetary defense scenario, multiple deflectors with conservative margins.

ESA's Hera mission, launched in October 2024, will arrive at the Didymos system in late 2026 to characterize the post-impact crater, the moonlet's mass, and the structural state of the rubble pile. Hera's results will set the parameters for any future kinetic deflection mission against a real impact threat.

DART changed the orbit of a moonlet by 32 minutes. It was the first time in our species's history that we had altered, on purpose, the path of an object in space.

The Risks We Cannot Deflect

Three of the entries on the catalog are not, on current technology, deflectable. Supervolcanoes are entirely terrestrial; mitigation strategies focus on monitoring (the U.S. Geological Survey continuously instruments Yellowstone, the Phlegraean Fields outside Naples, and a handful of other major systems) and on emergency response, not on prevention. The eruption itself can no more be averted than a hurricane.

Gamma-ray bursts — the brief, intense flashes produced by certain massive stellar collapses and binary neutron star mergers — would, if one occurred close enough to Earth and pointed in our direction, deplete the ozone layer for several years, sterilize the upper layers of the ocean, and produce a sudden ultraviolet flux at the surface for which we have no defense. Some authors have suggested that a nearby gamma-ray burst was responsible for the late Ordovician extinction roughly 444 million years ago, on the basis of a small but suggestive ratio of marine extinctions in shallow versus deep water. The hypothesis is contested. The threat exists; the timescale is poorly known. The expected rate within a danger radius is one event per 100 million to one billion years.

Stellar flybys, finally — Gliese 710 is the most precisely known example, but it is not the only one. Recent analyses of Gaia DR3 by Coryn Bailer-Jones and collaborators have identified roughly thirty stars that will pass within one parsec of the Sun in the next million years and several hundred over the next ten million. Each is a potential perturber of the Oort Cloud. The cumulative effect over geological time is to maintain the long-period comet flux at Earth at roughly its current level. There is nothing to do about it. The galaxy keeps moving.

The Honest Answer

The honest answer to the question of how long humanity could last, taking the catalog seriously, has three parts. First: the dominant near-term existential risks to industrial civilization are not cosmic. They are anthropogenic — pandemics, climate disruption, nuclear conflict, and a long tail of technological accidents that the literature on existential risk has begun to inventory. The cosmic risks are real, but they are smaller per unit time than the anthropogenic ones over the next several centuries.

Second: over geological time, the dominant risks shift. On scales of millions of years, asteroid impacts and supervolcanoes set the floor. On scales of hundreds of millions of years, mass-extinction-class events become statistically inevitable. Whether humanity persists through those is not a question of whether we can dodge any particular event — it is a question of whether we have built enough redundancy, off-Earth and on-Earth, that no single event terminates the lineage.

Third: we now have, for the first time, the technology to act on at least one entry on the list. DART worked. NASA, ESA, JAXA, and CNSA are all developing follow-up missions. The asteroid catalog is roughly 95 percent complete at the kilometer scale, and the gaps are being filled by the NEO Surveyor mission, scheduled for launch in 2027. The other entries — supervolcanoes, gamma-ray bursts, stellar flybys, and the long catalog of contingent terrestrial events — remain outside our power to prevent.

Toba was a near-miss. The catalog of what we know is coming is short, well-measured, and almost entirely beyond our control. The first item we have learned to push back against is the asteroid. We have not yet learned to push back against anything else.