DONORS to charities rarely make the sort of cost-benefit calculations investors, for example, would think obligatory. So charities attract donations with pictures of smiling gap-toothed children, rather than spreadsheets showing how they actually spend their money. Tugging at the heartstrings, however, does little to allay the doubts of economists sceptical about the efficacy of charity. Who is to say whether donating to a homeless shelter is a better use of money than donating to a school?

Yet advances in social science, particularly in development economics, mean donors can now have a reasonably good idea of how far each dollar will go. Empirically minded do-gooders, members of the nascent “effective altruism” movement, argue that it is at last possible to put into practice a “fundamental axiom” of utilitarianism, first invoked in 1776 by Jeremy Bentham, a British philosopher: “It is the greatest happiness of the greatest number that is the measure of right and wrong.”

The vast majority of charitable contributions come not from big foundations, but from individuals. Data from the Giving USA Foundation, a non-profit, show that of the $390bn Americans gave to charity in 2016, $280bn came from individual donors. Of this, around $120bn went to religious organisations and $60bn to educational institutions (mostly universities).

Not all of this money was given with the intention of maximising human welfare. Take, for instance, the Make-A-Wish Foundation, which helps children stricken with life-threatening illnesses, by granting “wishes”, such as meeting celebrities or visiting theme parks. The typical wish costs the foundation around $10,000 to fulfil—heartwarming for the recipient but of little help in improving health generally. Yet some charities, notably those active in poor countries, can produce big public benefits for relatively small amounts of money. One estimate finds that surgery that prevents blindness induced by trachoma, an infectious disease, costs a charity just $100 per operation.

William MacAskill, a philosopher at Oxford University, argues that promoting inefficient charities might actually do more harm than good. Competition for donations is acute. Research by the Centre for Effective Altruism, a think-tank he co-founded, finds that every dollar raised by one charity means 50 cents less for others. Mr MacAskill also worries about “moral licensing”. One study found that people tend to treat giving to charity like buying a medieval indulgence—they may believe they have the right to act immorally if they have done something they deem altruistic.

Measuring a charity’s efficiency is not straightforward, however. Effective altruism’s most-cited evaluator is GiveWell, a non-profit group based in San Francisco founded in 2007 by Holden Karnofsky and Elie Hassenfeld, two former hedge-fund analysts. Traditionally, charities used to be rated according to their overheads. GiveWell instead calculates standardised returns on investment across charities, as measured by factors such as cost per life saved (see chart). The charities it rates most highly are not all household names.

Toby Ord, another philosopher at Oxford, argues that people from rich countries who are interested in maximising human welfare should focus their charity abroad. A donor who wants to improve educational outcomes, for instance, would do better to donate not to American schools but to charities trying to improve the diets of children in poorer countries. A rough meta-analysis by GiveWell finds that ensuring children in a poor country have enough iodine in their diets can lead to a four-point increase in average IQ.

One of GiveWell’s highest-rated charities is the Against Malaria Foundation (AMF), which distributes medically treated bed nets in poor countries. Malaria still kills some 400,000 a year, mostly in sub-Saharan Africa. There is still no cure for the mosquito-borne disease. But it is relatively easy to prevent its spread. The AMF estimates that it costs $4 to buy and distribute a treated bed net. According to GiveWell’s analysis, the health benefits from this in sub-Saharan Africa are equivalent to a child’s life saved for every $2,000 spent.

GiveWell’s approach to evaluation has its limitations. It is hard to make like-for-like comparisons of the efficacy of different charities with different goals. An alternative approach is simply to give money to poor people. A proliferation of mobile-payment apps has made this easier than ever before. GiveDirectly, a charity founded by a group of development economists in 2008, facilitates direct transfers to people in Kenya and Uganda. Mr Hassenfeld likens the organisation to an index—it serves as a baseline against which other charities can be judged. GiveWell reckons that in order for a charity to be more cost-efficient than GiveDirectly, it would have to provide goods or services that people cannot readily purchase by themselves.

Inevitably, even effective altruists have to accept a degree of uncertainty about the impact of their donation. The question is how much? GiveWell is relatively conservative when it comes to recommending charities, listing just nine organisations under its list of “top charities”. The Open Philanthropy Project, a research group spun out of GiveWell, is more willing to back ventures with only a small chance of success provided the potential benefits are big enough. An extreme example is its recommendation that donors finance research on the safe use of artificial intelligence (AI). The increasing economic importance of AI, and the fact that it is so poorly understood, have led many altruists to believe it may soon become one of the biggest threats to society.

It is hard to gauge quite how big the effective-altruism movement has become. But it does have some serious backers. Good Ventures, a non-profit group founded by Dustin Moskovitz, a co-founder of Facebook, and his wife, Cari Tuna, makes donations based almost exclusively on the Open Philanthropy Project’s recommendations. It does not accept outside donations, but is dedicated to spending Mr Moskovitz’s and Ms Tuna’s wealth, which Forbes reckons to be $15bn. Last year, Good Ventures gave out over $300m in grants.

Effective altruists fret that their movement might, in fact, have very limited appeal. Utility-maximising automatons might see the sense in buying mosquito nets over the internet for distant strangers. Human beings might find, say, volunteering at a local soup kitchen more satisfying emotionally. Ari Kagan, a researcher at the Centre for Advanced Hindsight, a think-tank at Duke University, points out that many people find the idea of applying quantitative reasoning to altruism repugnant—like charging family members for a meal. Surveys show that while the effective-altruism movement has grown quickly, it has mainly done so within a limited group of people—ie, young white men with degrees in science and philosophy.

Effective altruism can be a hard sell, even for the rationally minded. Silicon Valley-types have been keener to embrace the philosophy than those working on Wall Street, for instance. Mr Hassenfeld reckons that this is partly because programmers who get rich tend to do so at a young age, and are hence more open-minded about charity. Bankers, in contrast, start to make real money only in their 40s, by which time they may already have formed their charitable habits. With many potential donors, Mr Hassenfeld says, “it’s easy to get intellectual agreement, but harder to get action.” As utilitarians have long found, and Bentham himself lamented, “the rarest of all human qualities is consistency.”

Correction (June 7th 2018): This article originally cited an estimate of $5 to buy and distribute a treated bed net. The AMF informs us that figure should be $4. Sorry.