What Makes a Good Life?

In 1938, in a Harvard office, a physician called Arlie Bock announced a research project of almost comic ambition. Medicine, he complained, had spent its whole history studying sickness and had learned nothing whatever about health; he proposed to correct this by taking a group of promising young men and simply watching them — not for a year, but for the rest of their lives. He recruited 268 Harvard sophomores, healthy, male and, in the manner of the time, entirely white and entirely privileged. Among them, though nobody knew it yet, was a shy undergraduate named John Kennedy, who would occupy a rather more public office than his classmates. The men were measured, X-rayed, interviewed, their handwriting analysed and their parents questioned; then the researchers settled down to wait.

Across the same city, a second study had begun with the opposite population. Sheldon and Eleanor Glueck were following 456 boys from Boston's poorest districts — tenements without hot water, families the city had written off — trying to understand why some of them ended up in prison and others did not. For decades the two projects ran separately, each recording the small facts of ordinary lives: jobs taken, wives married, children born, drinks poured, illnesses survived. Eventually the two were merged into one, and the combined study, which came to be known as the Harvard Study of Adult Development, could do something almost no research can: compare the arc of a life that started with every advantage to one that started with none.

What makes the story remarkable is that it did not die. Studies like this almost always do — the money runs out, the young researcher who cared about it retires, the participants move house and stop replying, and attrition quietly eats the evidence. This one survived four changes of director, a world war, the deaths of nearly all of its original subjects and more than one funding crisis in which the whole archive nearly went into a skip. The men filled in questionnaires every two years and sat for interviews every fifteen; their medical records were collected, their brains eventually scanned, and when they married, their wives were added, decades late and after some argument. Eighty-five years on, the study is still running, now following the children of the original men — and the files, which fill room after room, contain something no snapshot survey can ever produce: not what people say makes a life good, but what actually happened to those who tried it.

The man who read most of those files was George Vaillant, a psychiatrist who took over the study in the early 1970s and spent more than thirty years inside other people's biographies. He had expected the answer to be about character, or discipline, or the kind of things that get printed on office walls. What the data kept saying instead was smaller and stranger. The men who aged best were not the richest, the most famous or the most academically brilliant; several of the most successful careers belonged to men who ended up isolated and unwell, and several of the poorest boys from the Boston tenements finished their lives contented and surrounded. When Vaillant was finally asked to compress three decades of research into a single sentence, he gave five words: Happiness is love. Full stop.

His successor, Robert Waldinger, the study's fourth director, put numbers on it. The single strongest predictor of how the men's later years would go was not cholesterol, not blood pressure, not income: it was how satisfied they had been with their relationships at fifty. The people who were most content in their connections at fifty were the healthiest at eighty — and the effect showed up in the body, not just the mood. Lonely people's health declined earlier, their brains declined earlier, and their memories became unreliable sooner. Chronic loneliness, on the study's evidence and on the wider research that followed it, does damage on a scale usually reserved for smoking. Waldinger's 2015 talk about all this became one of the most-watched lectures in the world, which tells you something in itself: tens of millions of people wanted to hear a scientist say it.

Two details rescue the finding from greetings-card sentimentality. The first is that quantity did not matter: the number of friends made no difference, and neither, by itself, did being married. What mattered was whether there was at least one relationship in which the man felt he could be genuinely known — someone he could actually call at three in the morning. The second is harsher. High-conflict marriages, the ones held together for the children and the mortgage, were worse for health than divorce; a bad bond is not a small good, it is a subtraction. And warmth had a long reach: men who had described close relationships in their fifties were, decades later, better protected against physical pain than men who hadn't. On days when they hurt, their mood stayed level. The lonely men's pain got into everything.

Now the honest part, because a study this famous attracts a lot of loose quotation. Its original subjects were all men, almost all white, all American and all born in the 1910s and 1920s; the wives, whose lives were arguably the more interesting experiment, were added decades too late. Much of the data is self-reported, and self-report is where human beings lie most gracefully. And the deepest problem is the one this lesson set up in the vocabulary: prediction is not proof of cause. Perhaps warm relationships protect health — or perhaps healthy, even-tempered people find it easier to keep friends, and the arrow points the other way. Serious researchers say both are happening at once. What lifts the finding above one study is that it keeps reappearing elsewhere: large reviews of hundreds of thousands of people across many countries find the same association between social connection and living longer, arriving from entirely different directions.

Which leaves the practical question, and here the study's advice is almost embarrassingly small. Waldinger's argument is that relationships require maintenance in the way bodies do — the idea of social fitness — and that most of us behave as though friendship, once made, is permanent. It is not; it drifts. The recommended interventions are not grand: replace one hour of screen time with one hour of person time; phone the friend you have been meaning to phone for a year and simply admit that it has been a year; do something alongside someone rather than merely near them, because shared activity carries conversation better than conversation carries itself. Take one relationship that has gone flat and change what you do in it. That's the entire prescription.

It is worth noticing what this collides with. The engineered feeds of the attention lesson are built to be more reliably rewarding than a friend who is slow to reply; the vanished evenings of the free-time lesson are exactly the hours in which relationships used to be maintained. Nothing in modern life is optimised for the thing the study says matters most, because nobody profits from your Thursday evening being spent on a friend. The 724 men were followed for eighty-five years so that we would not have to run the experiment ourselves — and the result, held up against most people's actual calendars, reads less like a discovery than an accusation. That is also why it is useful. The question they answered is now the question the calendar puts to you: whatever else is on it this week, is there an hour on there for a person?