Egor Potseluev

Overpopulation is a scarcity claim in disguise: too many people for the resources available. Such a claim may have been defensible in the past but not in the 21st century. Headcount in absolute terms is the wrong variable. Population growth reverses as countries develop, and output expansion outruns the rate of population increase, diminishing the claim's relevance.

The global fertility rate is 2.25 births per woman, down from 3.31 in 1990. More than half of all countries are already below the replacement rate of 2.1 and nearly a fifth are below 1.4. 63 countries holding 28% of the world population have passed their peak and are projected to shrink by 14% over the next 30 years, including Japan, Germany, Russia and China (UN, World Population Prospects, 2024). The fall has followed income, education and urbanisation in every region, in the same order, which is why the UN projects it to continue. World population is still projected to peak at 10.3bn in 2080. Can output keep up?

There are now 2.7× more people than in 1960 and real output per person has risen 3.2× over the same period, thanks to technology (World Bank, WDI). Thomas Robert Malthus expected food to bind first. Cereal production has grown 3.5× since 1961, attributable to yield per hectare rising to 4 tonnes from 1.4 rather than to land, which only increased by 16% (FAO/World Bank). Nor was this growth extracted from the bottom: the share in extreme poverty fell from 38% in 1990 to 9% in 2024. Carrying capacity is a ratio of output to people, and the numerator has outrun the denominator for 60 years. This breaks the link between headcount and scarcity that the overpopulation argument depends on.

Load per head, not headcount, remains a strong objection. A rich-country birth costs far more in resources, land and water. I accept that. But that is an argument about consumption patterns, not population increase. It suggests “consume differently”, not “there are too many people for all to be sustained”. Yes, world population still rises to 10.3bn by 2080, but the increase is concentrated where consumption per head is lowest, and plateaus within one lifetime. The real fear is demographics. Pensions and healthcare are among the claims an old cohort makes on a younger one, disproportionately increasing its load. The severity of this depends on the speed of the demographic shift.

Two things would change my mind. First, a developed country sustaining 2.1 births for over a decade would indicate that the fall was simply a price problem and growth would eventually resume. Hungary and South Korea, despite all the policy intervention, have not managed to reach the replacement rate yet. Second, if productivity stalls, since the link between headcount and scarcity is only broken while output expansion outruns population growth.

Sources: UN World Population Prospects 2024 · NBC News, Jan 2026 (China NBS release) · Xinhua / Statistics Korea, Feb 2026 · Gizińska, OSW Commentary 658, 2025

The change I am most optimistic about is in cancer treatment, specifically that cancer is being unbundled. It used to be treated as one disease per organ. Increasingly it is being treated as many diseases defined by what an individual tumour is made of, and discovered without the need for symptoms. What is most reassuring is that it is not one breakthrough but progress arriving on several independent fronts at once.

On treatment: in August 2026 Merck and Moderna’s Phase 3 trial of an individualised mRNA therapy met its primary endpoint. Both recurrence-free and distant metastasis-free survival improved against standard immunotherapy alone in 1,137 patients with resected melanoma. Every dose is manufactured from the sequence of that patient’s own tumour. So, it is individualised treatment rather than a randomised trial.

On detection: the MASAI trial, published in The Lancet in January 2026, followed over 100,000 Swedish women and found that AI-supported mammography detected 81% of cancers against 74%, with 27% fewer aggressive subtypes and no rise in false positives, while cutting radiologist reading time by 44%.

The trajectory was already apparent and positive. UK cancer death rates fell 22% between 1973 and 2023, and half of people diagnosed now survive at least ten years, against one in four in the early 1970s. Meanwhile incidence rose 47% over the same period.

However, not everything works. The NHS-Galleri trial of a multi-cancer blood test, in more than 142,000 people, missed its primary endpoint in June, although it did find over 20% fewer stage IV diagnoses in its later rounds. Melanoma is the friendliest possible test case for an individualised vaccine, and more complex cancers remain early-phase. Recurrence-free survival is not the same as overall survival. A therapy manufactured per patient will not scale the way a pill does, so cost and access will bind long before the science does.

Despite this I remain optimistic because the binding constraint has moved. The question used to be whether any of this could work at all. Increasingly it is sequencing cost, manufacturing throughput and logistics, and those are economic rather than scientific and can be addressed through scale. The small wins also compound: better detection produces more patients with operable disease, which is precisely the group in which individualised therapy has been most effective.

Sources: Moderna/Merck INTerpath-001 announcement, Aug 2026 · MASAI, The Lancet, Jan 2026 · NHS-Galleri results, QMUL, June 2026 · Cancer Research UK, 50-year analysis, 2025