One Photograph Raised Donations a Hundredfold. Five Weeks Later It Had Worn Off.

A Syrian refugee fund opened a month before Alan Kurdi drowned, which gave researchers a clean before and after. What the numbers show about us is not flattering, and it is not our fault.

On September 2, 2015, a small boy named Alan Kurdi drowned in the Mediterranean.

His family had fled Kobani, in northern Syria, and was trying to reach the Greek island of Kos from the Turkish coast. His older brother Ghalib drowned. So did their mother. Their father survived.

A photograph of the boy, lying face down at the edge of the water where the surf had left him, ran on front pages around the world.

What happened over the following weeks is unusually well measured, and the reason is an accident of timing. The Swedish Red Cross had opened a fund designated specifically for Syrian refugees on August 4, a month before the photograph existed, and kept it running until the end of November. Paul Slovic and his colleagues at the University of Oregon realized that gave them something researchers almost never get on a question like this: a clean before and after, in real money, with no experiment involved.

They published the result in the Proceedings of the National Academy of Sciences in January 2017.

What the money did

In the week before the photograph, the fund was receiving fewer than a thousand donations a day.

In the week after, it was receiving close to fourteen thousand. The mean number of daily donations rose more than a hundredfold. The amount coming in each day rose by a factor of fifty-five, an additional 1.9 million kronor every day, roughly two hundred and ten thousand dollars.

Google searches for Syria and for refugees spiked in the same window.

Two days after the photograph appeared, Germany agreed to admit thousands of refugees who had been stranded in Hungary. The British Prime Minister said publicly that anyone who saw those pictures could not help but be moved, and that as a father he had been deeply moved by the sight of that young boy on a beach in Turkey. In Canada, the country the family had been trying to reach, the photograph measurably affected a general election.

One image of one child did all of that inside a week.

Then it stopped

The elevated giving held for about five weeks.

After that, the number of daily donations declined to a level statistically indistinguishable from the week before anyone had ever seen the photograph.

Nothing had improved. The war was in its fifth year. People were still getting into the same boats on the same coast and drowning in the same water, in numbers that did not fall and would not fall for years. The only thing that changed across those five weeks is that the picture stopped being in front of anyone.

Slovic’s own summary of the period before the photo is the line I keep returning to. The world, he wrote, was basically asleep as the body count in the Syrian war rose steadily into the hundreds of thousands.

It woke up for a month because of one boy. Then it went back to sleep, and the boy was still dead, and so was everyone else.

There is one durable effect in the data and it matters more than the spike. Signups for recurring monthly donations rose roughly tenfold, and that part did not fully revert. A small number of people took the feeling they had that week and converted it into a standing instruction to their bank.

Why numbers cannot reach us

Slovic has spent a career on the underlying problem and calls the extreme form psychic numbing. The short version is that our concern does not scale with the size of a catastrophe. It attaches to individuals, and it does not aggregate.

The related and better-established finding is the identifiable victim effect: people respond more strongly, and give more, to one named person with a face and a story than to statistics describing many. That effect has held up across a large body of work and a meta-analysis, and there is brain imaging evidence that the neural signature associated with valuing a life declines as the number of lives increases. The discounting appears to happen upstream of anything we would recognize as a decision.

Some of the more dramatic laboratory versions of this have not aged as well, and I am going to say so rather than leave it out.

A 2014 study by Västfjäll, Slovic, Mayorga, and Peters found that Swedish undergraduates offered a seven-year-old girl named Rokia gave an average of 37.7 kronor, and that adding a nine-year-old boy named Moussa dropped the average to 26.3. Giving went down when the need went up. It is a startling result and it has been quoted everywhere, including by me until about an hour before I finished this.

In 2025 a pre-registered direct replication with a US sample six times larger found no support for it. A separate 2025 paper examining real-world online donation data is titled Compassion for All, and it argues the opposite.

So the lab version of compassion fade is genuinely contested, and anyone telling you the two-child finding is settled science has not checked.

None of which touches the Kurdi data, because that was not an experiment. That was a real fund, real donors, real money, and a decay curve that took five weeks. Whatever the mechanism turns out to be, the behavior is documented.

This is not a character flaw

The scolding version of this essay writes itself. People are shallow, attention is fickle, we weep over one child on a beach and step past a hundred thousand.

I do not think that is right, and I think it points us at the wrong repair.

Caring about someone you have never met is not a malfunction. It is the same faculty that lets you care about anyone outside your own house, running on the only input it was built for, which is one person you know something about. A face, a name, a story, an age. That is the format the system reads. It has never been able to read a figure with five zeroes in it, and there is no evidence it ever will.

The failure is not that we feel too much for one. It is that the equipment does not scale, and the world has been arranged in a way that requires it to.

I use this

Two weeks ago I wrote about deaths in immigration custody and opened with a man named Emmanuel Damas, fifty-six years old, who asked for care for about a week for an infected tooth, was given ibuprofen, and died of septic shock.

I opened there deliberately. The actual finding of that piece was fifty-two deaths in five hundred days at the highest mortality rate in over a decade. That number is the story, and I knew it would not land, and that a man and his tooth would.

I did the same with the aid cuts. The honest figure is somewhere near six hundred thousand people already dead, two thirds of them children. Nobody can hold that in their head. So I wrote about one child in a camp whose feeding program ended, who did not die that week but declined, and then caught something ordinary.

Every writer who covers this does it. Every charity does it, which is why the envelope contains one photograph and one name rather than a spreadsheet. It works, it is not dishonest, and it is also a standing admission that the real scale is out of reach by any method we have.

What actually works

Slovic’s conclusion, after decades, is not that we should try to feel harder. He does not believe that is available to us.

His argument is that because moral feeling does not scale, it cannot be the trigger for responding to mass suffering. What is required instead is laws, institutions, and decision procedures that commit in advance, while nobody is looking at a photograph, and then execute regardless of whether anyone is moved that week.

Standing orders instead of spikes. Treaties with automatic triggers. Agencies funded on a baseline that does not depend on whether an image went around this month.

Which explains a great deal about which things have proven easiest to dismantle over the last eighteen months. An agency that prevents deaths nobody can picture, in places no camera goes, on behalf of people the public has never been shown, is the least defended line in any budget in the world.

The tenfold rise in recurring donations is the only part of this story with a future in it. Some number of people took an emotion that was going to be gone in five weeks and converted it into a monthly transfer that happens on a date, automatically, without them.

That is the whole technology available to us. Not caring more, which we cannot do on command. Deciding once, while we still care, and then making the decision difficult to reverse.

The photograph moved two governments inside forty-eight hours.

Five weeks later the giving was exactly what it had been before anyone saw him, and the boats kept leaving, and the water was the same temperature, and none of it was in the news.