Research note

The 1960 Doomsday Equation: Reproducing the 2025 Forecast and Its Sensitivity

The Null Institute · Independent Research Note · 19 November 2026
By Ilya Nikitin, PhD · Recalculation of a published equation against an independent UN estimate · Not peer-reviewed.

In 1960, Heinz von Foerster, Patricia Mora and Lawrence Amiot fitted a population-growth equation to 24 historical world-population estimates. Its point estimate placed a mathematical singularity at 2026.87, popularly rendered as Friday, 13 November 2026. The original paper did not observe a future catastrophe; it extrapolated a fitted mathematical relationship. We reproduced the 2025 extrapolation and checked how sensitive the result is to the time coordinate and the paper’s reported uncertainty in the singularity date.

The independently reproducible result

The original paper gives equation (11): N(t) = 1.79 × 1011 / (2026.87 − t)0.99, with t measured in years and N in people. Substituting t = 2025 returns 96.323 billion. The United Nations’ 2025 midyear world-population estimate is 8.231613 billion. The nominal-year model value is therefore 11.70 times the UN estimate, or 88.091 billion higher.

That is a transparent recalculation of the original model, not a newly fitted population forecast. The UN figure is an estimate, not a direct global census. The model’s t = 2025 value and a midyear 2025 estimate are also not perfectly aligned in time, which matters enormously near a mathematical singularity.

Calculation Result
Original equation evaluated at t = 2025.0 96.323 billion
Original equation evaluated at t = 2025.5 131.069 billion
UN 2025 midyear population estimate 8.231613 billion
t = 2025.0 model / UN estimate 11.70×
t = 2025.5 model / UN estimate 15.92×

Why the precise date gives false precision

The original authors reported t0 = 2026.87 ± 5.50 years, alongside uncertainties for the coefficient and exponent. Holding the other parameters fixed and moving only t0 five and a half years later, to 2032.37, lowers the calculated t = 2025.0 population from 96.323 billion to 24.778 billion. Moving it five and a half years earlier, to 2021.37, puts 2025 beyond the model’s singularity, where this real-valued expression is no longer usable.

This is a one-parameter sensitivity illustration, not a confidence interval. The reported parameter errors are not an uncertainty distribution for the 2025 forecast, and the fitted parameters may covary. Nonetheless, the calculation shows why a calendar-day headline conveys far more certainty than the historical fit could support. Even choosing the exact coordinate of 2025 within its calendar year moves the predicted population by about 36%.

Reproduce every number

import math
K, t0, k = 1.79e11, 2026.87, 0.99
un_2025 = 8231613070  # UN 2025 midyear estimate

def forecast(t, singularity=t0):
    return K / (singularity - t)**k

for year in (2025.0, 2025.5):
    n = forecast(year)
    print(year, f"{n / 1e9:.3f} billion",
          f"{n / un_2025:.2f}x UN")

print("Nominal-year excess people:", round(forecast(2025) - un_2025))
print("Later singularity:", f"{forecast(2025, t0 + 5.5) / 1e9:.3f} billion")

No curve was refitted and no observations were excluded or selected to improve the comparison. The inputs are the paper’s published point coefficients and the UN’s 2025 midyear estimate. For a differently dated population estimate, change un_2025 and report the new reference date.

What this check can and cannot establish

The reproduction decisively falsifies the unchanged extrapolation as a description of 2025 world population. It does not show that the original least-squares fit to pre-1958 estimates was numerically incorrect, nor does it by itself identify the causes of subsequent demographic change. Declining fertility and demographic transitions are plausible mechanisms supported by separate demographic evidence, not mechanisms measured by this small calculation.

The original authors reported approximately 7% root-mean-square deviation across their 24 historical population estimates. That in-sample statistic is not an out-of-sample forecast guarantee. A model can closely describe its calibration data while failing when the system’s governing mechanisms change.

Sources and method boundary

1. von Foerster, H., Mora, P. M., and Amiot, L. W. (1960). Doomsday: Friday, 13 November, A.D. 2026. Science 132(3436):1291–1295. Full original article (PDF); equation (11) and parameter estimates on p. 1293.

2. United Nations Statistics Division, Statistical Pocketbook 2025, World population 2025: 8,231,613 thousand (midyear estimate). Independent corroboration: UNFPA country dashboard reports 8.232 billion (rounded).

Scope: direct arithmetic reanalysis of a 1960 population model; sensitivity to fixed parameters and time coordinate. No causal analysis, new demographic dataset, refitting, or statistical forecast interval. Calculations checked on 19 November 2026.