Was Anybody
Else There?

Part I

The American day, as far as the federal government is concerned, begins at four in the morning. Not at midnight, when the calendar turns, and not at six or seven, when the alarms go off, but in the dead hour between, when almost everyone is asleep and almost nothing needs accounting for. Since January 2003 the Census Bureau, working for the Bureau of Labor Statistics, has been telephoning Americans and asking them to walk back through the twenty-four hours that began at four o’clock the previous morning. What were you doing? For how long? Where were you? And, after almost every activity except sleeping and washing, a question that sounds less like a statistician than a suspicious spouse: who was with you?

Each respondent is assigned a day in advance, a Tuesday say, and called the morning after. If nobody answers, the bureau tries the following Tuesday, and the one after that, for up to eight weeks. Someone who reports doing two things at once is asked which was the main one; if they can’t decide, whichever they mentioned first wins. Stretches people can’t remember, or would rather not describe, go into a residual category alongside the activities nobody could classify. The result is the American Time Use Survey, the nearest thing the United States has to a national diary: not a record of what people believe, or how they feel, but of what they did.

It is a peculiar document. An hour cleaning the kitchen counts as an hour of housework, whether it was penance or meditation. A conversation with a neighbour over the fence counts as socialising only if talking was the point of being at the fence. Millions of hours have been logged this way, sleeping and commuting and shopping and gardening and minding children and doing nothing in particular, and laid end to end they make a portrait of ordinary life more patient than any novelist’s. They also contain a puzzle. Set the American day of 2003 beside the American day of 2025 and something has gone missing. It isn’t the thing everyone assumes.

Part II

Consider what has happened to inconvenience since 2003. That year there was no Facebook, no YouTube and no iPhone, and Netflix was a company that posted DVDs. Paying a bill, booking a flight, finding an address or buying a pair of shoes usually meant a trip, a queue, a phone call or a map that never folded back the way it came. Two decades of invention have gone into abolishing those small tolls, with remarkable success. Dinner arrives without a conversation. Money moves without a bank. Nearly anything can be bought from the sofa, and the pitch for almost all of it has been the same: this will give you time back.

Yet the defining complaint of modern life is that there isn’t any. The week is a blur, the weekend evaporates, everyone is flat out. So I expected the diaries to show the squeeze: leisure quietly eaten by work, commuting and admin, the promise of the machines broken on the clock.

They show nothing of the kind. In 2003 Americans aged fifteen and over recorded an average of five hours and seven minutes of leisure and sport a day. In 2025 the figure was five hours and ten minutes. Across twenty-two years of smartphones and streaming, a financial crisis, a pandemic and the reinvention of the office, the nation’s free time moved by three minutes.

An average hides a great deal, and this one hides more than most; we will come to that. But as a national total it is hard to argue with. Whatever happened to American life after 2003, it was not a theft of hours. The hours are still there. What changed is how they line up.

Part III

That American social life has been thinning is not news. The Surgeon General declared an epidemic of loneliness and isolation in 2023, and “the anti-social century” made the cover of The Atlantic in 2025. Most of that coverage measured how much social time had been lost. The diaries also say something about how it was lost, and that is the more interesting half of the story.

The survey publishes two numbers for each activity that sound like the same thing and aren’t. One is how much time the average person spent on it. The other is how many people did it at all on a given day. The first is a quantity, the second a frequency, and most reporting on the survey sticks to the first.

Take a hundred American diary days from 2003, chosen at random, and look for any stretch in which socialising was the main thing going on: a visit, a party, a long talk on the porch. (Phone calls, texts and email are counted separately, and a chat while cooking dinner goes down as cooking.) About forty of the hundred will have one. Do the same with a hundred days from 2025 and you will find about thirty. Ten social days have gone from every hundred. More precisely, the share of days with socialising as a main activity fell from 40.1 per cent to 30.4 per cent, a relative decline of almost a quarter.

Now look only at the people who did report socialising, and ask how much of the day it took. In 2003 the answer was about an hour and fifty-six minutes. In 2025 it was about an hour and fifty-five. The survey can’t say whether that was one long dinner or three short conversations, but the total barely moved. Social days have not been trimmed to fit busier lives. There are simply fewer of them. Run the arithmetic on the published, rounded figures and around 96 per cent of the fall in average socialising time comes from fewer days containing any socialising, not from less of it on the days that did.

The limits of the evidence belong here, once. Each respondent contributes a single day and is never called again, so the survey can’t say whether the people still socialising are the same people as before, or whether some Americans now go weeks without a visit while others carry on as normal. The sample is much smaller than it was, about 21,000 people in 2003 and about 6,100 in 2025, so recent estimates carry wider margins. And the diary can’t say whether anybody minds. A quiet night in after a crowded week and a quiet night in waiting for a message that never came look identical on the page. What it can say is that, across the whole population, an ordinary American day is about a quarter less likely to contain the central event of social life than it was in 2003.

Part IV

If American social life had a home address for most of the twentieth century, it was Friday night: the hinge between the working week and whatever came after, the evening of going out, staying out and turning up at someone’s place with no plan beyond the company. Not everyone lived it, but nearly everyone understood it. In 1978 the sociologist Murray Melbin described the night itself as a frontier that Americans, especially in cities, were steadily settling.

In August, Neal Caren, a sociologist at the University of North Carolina at Chapel Hill, published a short paper on what has become of that frontier. His question was not how much social time had been lost, which is well documented, but when. Caren used a different part of the diary from the one above: not what people said they were doing, but who they said was with them, which lets you count every minute spent in the physical company of friends, whether the activity was dinner, a game or a film. Mapped across the hours of the week, the early-2000s pattern looked like a heartbeat, with small pulses each weekday evening building to sharp peaks on Friday and Saturday nights. At the Friday peak, about nine per cent of all adult person-time was spent with friends. By 2022–24 it was about four per cent. Friday had come to look like Thursday. Saturday held up better, but slid too. Across the whole week, time with friends fell by more than half.

Two details in Caren’s work matter here. The first is timing. The weekly rhythm held for the survey’s first decade, began to come apart in the mid-2010s, collapsed in 2020 and never recovered; the pandemic was an accelerant, not the cause. The second is composition. The mix of things people did with friends barely changed. Americans had not swapped the bar for the dinner party, or the dinner party for the gym. The whole repertoire had faded at once, like a print left in the sun.

Set beside the government’s own headline tables, the picture is less tidy than I would like. Those tables show weekends losing social days at about the same rate as weekdays, if anything slightly slower: between 2003 and 2025 the share of weekend and holiday diary days with socialising as a main activity fell from 45.5 to 35.2 per cent, and the share of weekdays from 37.9 to 28.4. The two measures count different things. Caren counts any time spent in the company of friends, whatever people were doing; the government’s figure counts socialising as the main activity, with anyone at all. Somewhere in the gap between them is a question this essay can’t settle: whether Friday night lost its friends while the rest of social life thinned more evenly, or whether the two measures are seeing the same retreat from different angles.

Part V

The person with five hours and ten minutes of leisure a day doesn’t exist. The average blends the retired with the newly employed, teenagers with the parents of toddlers, Saturdays with Tuesdays, and the groups it blends live in different worlds. Americans aged thirty-five to forty-four recorded three hours and fifty-three minutes of leisure a day in 2025; those aged seventy-five and over recorded seven hours and twenty-three. Employed adults living with a child under six recorded three hours and eleven minutes, and employed adults with no children at home recorded four hours and thirty-two, a gap of eighty-one minutes (the groups differ in age and in other ways, so not all of it is the children). Men’s lead over women has widened, from about thirty-five minutes a day in 2003 to about forty-nine. For a great many people the shortage of time is not a feeling. It is arithmetic, and a flat national average is no comfort to them.

But read what the same tables say about socialising and the arithmetic stops behaving. Americans over seventy-five, with nearly twice the leisure of thirty-five-to-forty-four-year-olds, spend almost exactly the same time socialising: about thirty-four minutes a day against thirty-five. Employed parents of children under six, who have less free time than almost anyone in the survey, spend more time socialising than employed adults with no children at home: about forty-four minutes a day against thirty-one. The people with the most hours do not have the most company. The people with the fewest hours have more of it.

The survey doesn’t say why, and the comparison has limits. A child under six might be a newborn, and socialising here could mean a visit from the grandparents as easily as an afternoon at the playground. But one possibility is worth taking seriously. Small children come with a timetable other people share: the school gate at three, the birthday party on Saturday at ten, the playground everyone with a two-year-old drifts towards at the same hour. Parents of young children are short of time, but the time they have may line up with other people’s. Retirees have time in abundance, but on a Wednesday afternoon most of the people they might spend it with are at work.

In 2014 two sociologists, Cristobal Young and Chaeyoon Lim, offered a useful way to think about it. Time, they argued, is a network good: like a telephone or a language, its value depends on how many other people have it at the same moment. Their evidence was the weekend. Workers are happier on Saturdays than on Mondays, which surprises nobody. But the unemployed are happier on Saturdays too, by a similar margin, even though for them a Saturday is no freer than a Monday. What changes at the weekend isn’t how much time they have. It is how many other people are free alongside them. A day off alone and a day off everyone shares are different goods.

Part VI

This reframes the obvious suspect. It is hard to write about the decline of company without blaming the screen, and the diary does record screens. Americans spent about two hours and thirty-four minutes a day watching television in 2003 and about two hours and thirty-seven minutes in 2025. The total barely moved, but its shape did: fewer days included any television at all (78.7 per cent, down to 74.3), and on the days that did, viewing ran longer, from about three hours and sixteen minutes to three hours and thirty-one. Television did not spread out to fill the missing social evenings. It consolidated.

The phone is harder to see, and that is no accident. The diary records one main activity at a time and, apart from looking after young children, collects nothing people did simultaneously. A text answered over dinner, a feed scrolled through a film, a group chat murmuring all evening: none of it appears. The category for phone calls, mail and email has since been widened to include texting and video calls, and it now turns up on fewer diary days than it did in 2003, which says less about Americans than about the instrument. Messages live in the gaps between activities, and the diary only has room for activities. The survey can’t convict the smartphone. What it can do is describe the shape of the change, and the shape fits some explanations better than others.

None of the obvious explanations fits cleanly. If Americans had grown impatient with company, the time given to it on social days would have fallen. It didn’t. If they had become too busy, leisure would have fallen. It didn’t. What remains is a population with roughly the same free time, giving roughly the same time to socialising on the days it happens, and having far fewer of those days. It looks, at least in part, like a problem of coordination. The time is still there. The open question is whether it falls at the same hour for the people who might want to spend it together.

There are plausible reasons, and several are versions of one thing: the slow dissolution of shared timetables. Television used to be an appointment; now it waits for you. Shops and banks used to close; now they are a tap away at midnight. The working week used to end at roughly the same hour for most people in the same town; now, for many office workers, Friday is a day at home, while for many service workers it is the busiest shift of the week. Even plans have lost their fixity. When everyone can be reached at any moment, a Friday dinner can be floated, postponed and quietly dropped in the group chat without anyone having to cancel, and the default, which once ran towards turning up, now runs towards seeing how the week goes. Each of these is a convenience. Together they may have dismantled the common calendar that once made getting together the path of least resistance. The diaries can’t prove any of this. At most, they show that it fits.

Part VII

For twenty years we have measured our inventions in time saved, and by that measure they worked: the hours are all still there, three minutes to the good. What nobody priced was the schedule they replaced. The shop that shut at five, the show everyone watched at eight, the night nobody had to organise because everybody already knew: these were inconveniences, and they were also the scaffolding of a shared life. A free evening is not a social one. It becomes social only when someone else’s evening happens to be free at the same time and both of you know it, and a society that has optimised every individual’s hours has made that coincidence rarer.

The next wave of inventions is being sold with the old pitch. Artificial intelligence, its makers say, will give us our time back. Perhaps it will. But the diaries suggest that time was never what we were short of, and they leave a sharper question for the machines arriving now. Will they help two people find the same free evening, spotting the Friday that four calendars share and booking the table? Or will they offer each of us something else: someone to talk to when everyone else is busy, who is always free, never cancels, and in the government’s diary would count as nobody at all?

At four tomorrow morning another diary day will begin. The morning after, somewhere, a phone will ring, and a stranger from the Census Bureau will ask a person to walk back through yesterday. Where were you? What were you doing? And then, after almost every answer, the question that turns out to matter most.