IMAGINE building a small pile of wood and kindling in the smallest room in your house, and setting fire to it. You can keep the door open, to let out some smoke, but cannot switch on an extractor fan. You must tend the fire for an hour. Repeat the process three times a day.

This is how Fatou N’Dour lives. Her kitchen, separate from her home and built of mud bricks, measures roughly two metres by two. She usually cooks indoors because of the winds that whip across Lambayene, the village where she lives in central Senegal. Asked about ventilation, she points to a hole in one wall, which is about ten centimetres square. Other women in the village cook rice, couscous and meaty sauces in similar conditions, using wood from a nearby forest.

Wood and charcoal in Africa; coal in East Asia; wood and animal dung in South Asia—in much of the world, food is heated by burning primitive solid fuels. Each fire is tiny, but the International Energy Agency (IEA), a Paris-based research group, estimates that 5% of the world’s primary energy demand in 2016 was supplied by “traditional solid biomass”. Wind turbines and solar panels combined generated less than half as much energy.

The awful effects of these fires begin with their impact on human health. Household smoke is thought to be the world’s most lethal environmental problem, killing 2.6m people a year. Where wood and charcoal are burned, trees often disappear. Africa loses some 0.5% of its forests every year, a higher rate of destruction than South America’s. Soot from domestic fires also warms the planet, particularly when it settles on snow. Black carbon like that from dirty cookstoves is thought to be the third most important cause of climate change after carbon dioxide and methane.

Governments, aid agencies and charities have for decades tried to coax people towards cleaner fuels like liquefied petroleum gas (LPG) and electricity. Those who must burn wood and dung are prodded to do so in more efficient stoves.

Progress has been astoundingly slow. Since 2000 the number of people living in extreme poverty has plunged from 1.7bn to about 600m. Neonatal deaths have fallen by 49%. Yet the number of people heating their food with dirty fuels has stuck at 2.5bn-2.8bn, according to the IEA, largely because of growth in Africa (see chart). The Global Alliance for Clean Cooking, which uses a slightly different measure, estimated in 2015 that the number might even have risen. As for those improved cookstoves, researchers who hand them out in a village almost invariably find, when they return several years later, that people have gone back to cooking over handmade mud stoves or large stones.

That efforts to change how people cook have fallen so short for so long can be blamed on weak markets, unco-ordinated charity interventions and muddled priorities. It also illuminates why development is so much harder in Africa than in Asia.

Cooking over an open fire is no fun, especially if you have to do it every day. In another village in western Senegal, Felane, women complain that their kitchens are always hot and smoky. The smoke stings and irritates—one woman blames it for colouring the whites of her eyes. Firewood is becoming ever harder to find. A local man, Cheikh Diouf, who has nine children, says that wood-collecting may take four trips a week, each one of up to four hours.

Those who have a simple metal cookstove with a clay liner, known as a jambaar, say it is better than the traditional method of balancing a pot over three big stones atop a fire. The jambaar is more efficient, needing less wood. Surveys in other countries show that many poor women realise this. A jambaar can also be moved outside when the weather allows. And it just feels superior: one woman in Lambayene describes it as “civilised”.

Yet jambaar stoves are seldom on sale at the weekly markets. People seem not even to know how much they cost. Gunther Bensch and Jörg Peters, both of the Leibniz Institute for Economic Research in Germany, gave jambaar stoves to Senegalese villagers in 2009. When they checked, in 2015, almost all had worn out. Hardly any had been replaced.

The problem is not only poverty. Mr Bensch and Mr Peters have tried auctioning jambaar stoves. They found that villagers often bid more than they would pay in nearby towns. Perhaps they do not buy them in markets because shopping is seen as women’s work, and women are not allowed to spend much without consulting their husbands. Or perhaps it is too difficult to carry stoves from town to village. One urban stove vendor, Malick Niang, says he would not try to sell the stoves in villages. They are heavy and breakable, and demand there is uncertain. Another problem is that, being safe, poor and French-speaking, Senegal attracts charities and aid agencies. Some at times hand out stoves for little or nothing. That confuses people about their true value, and can wreck markets.

Even better cookstoves may not do much to improve health. The linkage between household smoke and harm seems not to be linear, says Kevin Mortimer of the Liverpool School of Tropical Medicine. Above a certain level, breathing more smoke might not make a person much sicker. Even the most efficient wood stoves expose cooks to many times the level of smoke that the World Health Organisation regards as safe. Mr Mortimer was involved in a large trial in Malawi, using a top-of-the-line stove, which found no evidence of an effect on rates of childhood pneumonia.

Perhaps it is better to pick a genuinely clean fuel (clean to cook with, not necessarily in the planet-preserving sense) and promote it hard. Brazil, Ecuador and Indonesia, among others, have all subsidised LPG. Since 2016 the Indian government has made LPG available to 34m households, giving them gas stoves and one cylinder free. The petroleum ministry says that four-fifths of the newly connected households have bought a replacement cylinder. On average, they buy four cylinders a year, which implies they get at least half of their cooking energy from wood, dung and the like. Still, this is rapid progress.

So switching fuels rather than stoves seems the more hopeful approach. “We were fooling ourselves, thinking that we could pick any old fuel off the ground and make it burn cleanly,” says Kirk Smith, an environmental scientist at the University of California, Berkeley, who is involved with India’s programme. Not only is LPG much cleaner than solid fuel. It also feels like a step up in the world and is easier to use (even men can cook with it).

Subsidies make for poor policy tools. They are snaffled by wealthy, well-connected people. They create lobbies supporting them, and become hard to cut. Particularly in small countries, subsidised goods are likely to leak over borders. Subsidies may also vary from year to year with the government’s budget. That is a particular danger in the case of cooking fuel, because cooks prize reliability. If people cannot always obtain clean fuel, they will probably revert to dirty stuff, says Radha Muthiah, the departing head of the Global Alliance for Clean Cookstoves.

These are mighty problems even for large middle-income countries with more-or-less competent governments. India, which for years frittered away money on LPG for the middle class, has managed to steer the subsidies—the world’s biggest cash-transfer programme—more accurately towards the poor, partly thanks to the Aadhaar biometric-identity scheme. But in smaller, poorer, more corrupt countries, LPG subsidies are probably out of the question. India has found a tricky, costly way of clearing the air. In sub-Saharan Africa, the smoke lingers.