Generated by Codex with GPT 5.6 Sol XHigh

A forecast built from one number

Most predictions of human extinction begin with a threat: nuclear war, an asteroid, ecological collapse or some other physical mechanism. The “doomsday argument” begins somewhere stranger. It uses probability and a single estimate—that about 117 billion humans have been born—to suggest that humanity may be much nearer the end of its history than the beginning.

The intuition comes from a simple lottery. Suppose one drum holds tickets numbered from 1 to 100 and another holds tickets numbered from 1 to one billion. Drawing ticket 14 strongly favors the smaller drum, because such a low result would be extraordinarily unusual in the larger one. The doomsday argument treats a person’s place in human history in much the same way. Being the 117 billionth human would be unremarkable if the species ultimately produced a few hundred billion people, but astonishingly early if a future interstellar civilization eventually produced trillions.

The calculation assumes that any present-day person can be treated as a random sample from the complete sequence of humans who will ever live. Half of that sequence must fall between the 25th and 75th percentiles. If 117 billion previous births place the current generation somewhere in that middle half, then the final human population would contain roughly 156 billion to 468 billion people. At today’s birth rate of about 132 million per year, that translates into a 50 percent probability that the last human will be born between 295 and 2,659 years from now. A wider interval gives an 80 percent probability of the final birth occurring within 98 to 7,977 years.

Those dates depend heavily on the assumed birth rate. The article holds it constant; exponential population growth would consume the remaining birth slots even faster. Either way, the forecast is startlingly short compared with the roughly 300,000 years that modern humans have already existed, and it leaves little room for a vast, long-lived galactic civilization.

Why the idea is hard to dismiss—and hard to accept

The method has produced some conspicuous successes. In 1969 astrophysicist J. Richard Gott III visited the eight-year-old Berlin Wall and assumed that his visit was not occurring at a special moment in its lifespan. Treating that moment as somewhere in the wall’s middle 50 percent led him to predict that it would fall between 2.67 and 24 years later. It fell 20 years later. Gott later applied similar reasoning to 44 Broadway shows; all 37 that had closed by the publication of his 2001 book had ended within his predicted intervals.

These examples illustrate the Copernican principle behind the argument: absent contrary evidence, an observer should not assume a privileged place in space or time. Earth is not the center of the universe, and the present generation may likewise be neither humanity’s opening act nor an exceptionally early audience for an immense future. Yet the examples do not prove that the same sampling logic applies to the lifespan of the species. The argument’s critics challenge that bridge from ordinary duration estimates to human extinction in several ways.

The first problem is the reference class. Why should the sample include exactly all humans rather than Neanderthals, intelligent extraterrestrials or future descendants altered into cyborgs? Expanding or narrowing the population can move the forecast dramatically, making an apparently objective prediction depend on a disputed boundary.

The caveman objection tests the method against the past. An early human applying the same logic would have badly underestimated the species’ eventual longevity. A rule that would repeatedly have predicted premature doom throughout most of human history needs an explanation for why it should become reliable now.

The self-indication assumption points in the opposite direction. If one possible universe contains hundreds of billions of humans and another contains hundreds of trillions, a randomly situated conscious observer has far more opportunities to exist in the larger universe. Accounting for that selection effect may counteract the pessimism produced by present birth rank.

Finally, birth order has no causal power. Nuclear weapons and asteroids could kill people; a person’s numerical position among all births cannot. Critics therefore question whether birth rank provides genuine evidence about extinction at all, while supporters answer that probability can carry information even without supplying a mechanism.

A puzzle about reasoning, not a countdown clock

Every objection has inspired a counterargument, and both the doomsday argument and its rebuttals now exist in multiple technical forms. The unresolved dispute is therefore more useful as a stress test for probabilistic reasoning than as a practical forecast. It exposes how much depends on assumptions about who counts as an observer, whether a person can be treated as a random sample, and when statistical typicality is informative.

The argument does not identify a danger, measure any physical risk or show that extinction is imminent. Its force comes from making a modest-sounding rule—do not assume the present is special—produce an extraordinary conclusion. Whether that is profound evidence or a warning about misapplied probability remains unsettled. The lasting lesson is that predictions drawn from existence alone can never be separated from the hidden choices that define the sample.