Most of the world's active volcanoes are distant from dense human settlement, but a surprising number lie within easy reach of towns and cities. Italy has Etna and Vesuvius, Mexico has Popocatepetl, Japan has Sakurajima, the Philippines has Mayon, and the entire Indonesian archipelago is studded with peaks that erupt on a timescale measured in decades rather than millennia. Taken together, around 800 million people live close enough to an active volcano that a major eruption would affect them directly. Why these communities remain, and how they cope, is a question that brings together geology, economics and social history.

The first answer is soil. Volcanic ash contains minerals that, once weathered, release nutrients important to plant growth. Farmers on the slopes of Mount Merapi in central Java report yields that are substantially higher than those on nearby non-volcanic land, and similar patterns have been documented around Etna in Sicily. A 2018 analysis by the Indonesian Volcanological Survey found that the agricultural productivity of land within fifteen kilometres of Merapi was roughly twice that of surrounding lowlands, sustaining a population density that would otherwise be economically unviable. The price is obvious: the occasional eruption that disrupts precisely what the soil supports.

The second answer is water. Volcanic landscapes often contain springs and aquifers that, because of the rock's porosity and mineral content, supply reliable water even in dry years. The town of Catania in Sicily depends heavily on groundwater drawn from beneath Etna. Similar arrangements exist in Central America and in parts of East Africa. Access to stable water is a powerful inducement to remain in a hazardous place.

The third answer is harder to quantify: attachment. Families who have lived on a volcano for generations have cemeteries, fields, houses and reputations there. They rarely wish to leave, even when the government urges them to do so, and government attempts to relocate such communities have frequently failed. Social scientist Dr. Akiko Hoshino has studied relocations around Sakurajima, in southern Japan, and found that even families who accepted new housing elsewhere often returned to the slopes of the volcano within a few years, citing loss of community, poor schools or an inability to find similar farmland in the new location. Any policy of relocation, she concludes, must address far more than physical safety.

Monitoring has improved dramatically in the last fifty years. Seismometers placed around active volcanoes record the small earthquakes that often precede eruptions, and gas analysers detect rising levels of sulphur dioxide and carbon dioxide. Satellite measurements track ground swelling caused by magma moving underground. At Vesuvius, the monitoring network is dense enough that scientists expect to give the population several days' warning of any major event. At remote volcanoes in parts of Papua New Guinea, where equipment is sparser, warning times are far shorter.

Warnings, however, have to be heeded, and here too social factors matter. Communities that have experienced frequent small eruptions often develop a confident relationship with their mountain, evacuating quickly when told to. Communities that have not seen an eruption for decades can be harder to persuade. Volcanologist Dr. Simon Osei describes this as the 'long memory problem': the very rarity that makes a particular volcano seem safe is often the feature that makes its eventual eruption more destructive, because the accumulated pressure beneath the surface produces a larger event.

Insurance is limited in most of the regions most affected. Commercial insurance against volcanic damage is usually unavailable or prohibitively expensive, and governments fund recovery from emergency budgets whose scale is rarely matched to the potential cost. A slow shift is under way, with organisations such as the Asian Development Bank offering pooled risk arrangements to several Pacific countries, but the programmes are small relative to the exposure.

Life near a volcano, in other words, is a trade-off that whole societies enter into, often unconsciously, over long periods. The rich soil and the reliable water come with a bill that most years goes unpaid. When it does arrive, no amount of modern monitoring can entirely cancel it. What good policy can do is reduce its size, spread its cost, and, at moments when seismic and social conditions come together, convince people who know exactly what they are leaving to leave anyway.