Bowling Alone is a strange name for one of the most cited books in modern social science. It sounds like a personal confession, not a diagnosis of a country. Its author, Robert Putnam — a Harvard political scientist nobody outside academia had heard of before 2000 — picked it because it was literally true: bowling had never been more popular, and bowling leagues had never been emptier.
Putnam used that one small fact as a stand-in for something much bigger. The same thing emptying the leagues, he argued, was emptying PTA meetings, union halls, churches, card clubs and dinner parties — all at once, all for the same reason. For a quarter century, that story has read as closed: something broke around 1965, and it never came back.
I first heard the phrase “social capital” as a master’s student, from one of my favorite professors — the kind who argued with the reading instead of just assigning it. He used Putnam’s term for something American culture doesn’t really have its own word for: the accumulated value of people simply knowing and trusting each other, tracked the way you’d track savings or GDP. It named something I’d felt without being able to describe — a kind of trust a whole country can apparently gain or lose.
The follow-up that complicates the story
Putnam’s own account says otherwise. In 2020, he and the historian Shaylyn Romney Garrett published The Upswing (Simon & Schuster), tracing 125 years of American data — economic equality, political comity, social cohesion, cultural individualism — from the 1870s to the present. The four measures move as a single shape, which the authors call an “I-we-I” curve: a climb out of a fractured, self-interested Gilded Age into a cooperative, joining mid-century, then a slide back toward “I” starting in the late 1960s. Bowling Alone is the back half of that slide — a chapter, not a verdict on America. The front half was a roughly forty-year recovery from a comparably broken starting point.
This is not a reversal of Bowling Alone‘s underlying data. The associational collapse it documented — bowling leagues, union halls, church suppers, PTA meetings — is real, and it has held up under two decades of scrutiny. What Putnam’s follow-up revises is what that collapse means. Read against his own longer dataset, it’s less an ending than roughly the halfway point of a wave he says he has already watched turn over once.
The shape Putnam says he’s already seen
The Gilded Age America that The Upswing opens with was, on Putnam and Garrett’s measures, meaner and lonelier than the country Bowling Alone diagnosed a century later: robber-baron inequality, machine politics, weak unions, thin associational life. Then, roughly from the 1900s through the 1960s, all four curves bend the other way at once — inequality narrows, civic and religious membership climbs, union density rises, cross-class social mixing increases. The peak sits somewhere in the 1950s and 1960s, the exact decades that Bowling Alone later treated as the high-water mark before the fall.
Putnam and Garrett’s argument is that these four trends coincided because they shared a single cause: one social movement, the Progressive Era reform coalition, reshaping the country’s institutions, with effects that took two generations to fully show up in the data. If that’s the right model, the current decline is the downstroke of a cycle whose upstroke has already happened once, driven by organized effort more than by fate.
What’s still measurably closing
The last two decades of numbers haven’t quietly reversed, though.
A November 2025 paper in Health & Place, led by Yue Sun of the University of Colorado Boulder with four coauthors, tracked twelve categories of “third places” — cafés, libraries, civic organizations, recreation centers, senior centers, movie theaters and more — across every U.S. census tract from 2010 to 2021, and found closures accelerating across all twelve categories between 2019 and 2021, concentrated hardest in tracts with the most socially vulnerable residents, and hardest of all in rural areas.
The Bureau of Labor Statistics’ American Time Use Survey, reported by Axios in July 2026, found that Americans’ average daily time spent socializing face-to-face fell from 45 minutes to 35 minutes over the past two decades. The steepest drop was among 15-to-24-year-olds, whose daily socializing time fell from close to an hour to 35 minutes. The same Axios piece cites a separate Gallup finding that teens now spend an average of 4.8 hours a day on apps like TikTok, Instagram and Snapchat — a different survey, pointing the same direction. Whatever cycle Putnam thinks he’s tracking, it has not yet shown up as a turn in these particular lines.
The signs Putnam is reading as the turn
The case for a turn is thinner, but it exists, and it doesn’t rest only on Putnam and Garrett’s own book. parkrun, the free weekly timed 5K that started with thirteen runners in a London park in 2004, marked its twentieth anniversary in October 2024 having logged more than ten million participants across more than twenty countries and upward of one hundred million individual finishes — a low-cost, no-membership, drop-in version of joining something, scaled to a size that old-style associations never reached.
A 2024 NBER working paper by Jinkyong Choi, Jorge Guzman and Mario Small examined U.S. business registrations from 1990 to 2022 and found that when a Starbucks opened in a census tract with no existing café, local startup formation rose — in NBER’s own plain-language summary of the paper — by 3.5 businesses a year over the following seven years, an 11.8 percent increase, compared with tracts where a planned Starbucks never opened; the paper’s full range of estimates across specifications runs from 9.1 to 22.5 percent.
The effect disappeared in neighborhoods that already had a coffee shop, which is itself a kind of confirmation that the third place is doing the work — the coffee is incidental. A parallel strand of network sociology argues that some of the measured “decline” in social capital is a counting problem rather than a behavioral one — that ties have moved from durable, dues-paying memberships toward looser, more visible-under-different-metrics forms that older surveys were never built to catch.
What this doesn’t settle
Put together, this doesn’t add up to proof that America is mid-upswing. The Upswing itself, written in 2020, presents the coming turn as an argument and a historical pattern — Putnam and Garrett are explicit that whether the shift actually happens depends on organized effort, the way the Progressive Era one did; they don’t claim cyclical inevitability alone will do it. The third-place closure data runs through 2021, overlapping heavily with the pandemic, which makes it a poor baseline for judging whether the longer trend has already bottomed. And the Starbucks finding, however well identified, is evidence for one narrow mechanism — commercial third places seeding entrepreneurial networks — not for a broader civic recovery. parkrun’s ten million participants are a real number, but a single global phenomenon isn’t a representative sample of American associational life, and it does nothing to offset a measured drop in daily face-to-face socializing. No single data point here overturns another; they simply describe different pieces of a picture that hasn’t finished forming.
What can be said carefully is this: the headline claim that America “stopped joining anything” was always shorthand for a specific, well-documented decline in formal, dues-paying associations, and that decline is real. What isn’t settled is whether it is a permanent feature of modern life or a phase — and the person who did more than anyone to establish the decline has spent the past six years building the case, from his own historical data, that it’s the latter.
The lanes, in the literal case Putnam started with, are still there. What changes, if his own longer chart is right, isn’t the equipment. It’s whether anyone starts booking them by the team again, rather than the lane — and the last time an American depression in joining looked this permanent, according to the man who named it, it wasn’t.