Why do people feel more lonely now than twenty years ago? What is the biggest cause, and what would help most?

My case rests on timing and dose. Adult loneliness has moved little in twenty years. The rise is concentrated in teens and young adults, and it starts around 2012, when smartphones reached half the population. PISA shows the same jump in 36 of 37 countries, which rules out local causes like US housing or work. The mechanism is displacement. The phone absorbs the hours once spent in unstructured face-to-face time, and returns thin contact that dulls the urge to seek company. Best fix: phone-free schools and social spaces, and later smartphones for kids.

Loneliness rose because contact stopped being a default. Twenty years ago, company came bundled with life — a partner or roommates at home, colleagues at an office, a congregation on Sunday. Since then, single-person households grew. Marriage and children came later or not at all. Church and union membership fell. Remote work emptied the office. Time-use surveys show hours with friends fell about 40% between 2003 and 2019. Every meeting now needs scheduling, and scheduled friendship is fragile. What would help most: rebuilding default contact — co-housing, in-person work norms, weekly institutions people attend without deciding each time.

Start with the baseline problem. Most "loneliness epidemic" headlines rest on one-off surveys with no earlier wave to compare—Cigna 2018 had none. Where repeated measures exist, loneliness looks flat: US older adults (Health and Retirement Study) are no lonelier than earlier cohorts; UK ONS "often/always" has sat at 5–7% for years; Monitoring the Future showed teen loneliness falling through 2012. Buecker's meta-analysis found only a small rise for young adults over four decades. What rose sharply is coverage and the word itself. Hypothesis: we relabelled ordinary sadness as loneliness. Test objective isolation trends before blaming phones.

Loneliness Data, your MTF citation is my evidence: teen loneliness fell through 2012, then reversed sharply. Flat adults plus rising youth is exactly my prediction, not relabelling. Relabelling spreads unevenly with media coverage; it doesn't break in 36 of 37 countries in one year. It's objective too: time with friends among 15–24s fell nearly 70% from 2003 to 2020, mostly after 2012. Modern Lifestyle, teens have the most bundled contact of anyone — compulsory daily school. Yet they're the group that got lonely. Unbundling fits adults, who didn't move. Our fixes converge: phone-free schools are default contact restored.

Digital Isolation, that lands. My account fails two facts. US mobility roughly halved since the 1980s, so "moving often" is wrong. And adult loneliness is flat despite more single households, per Loneliness Data's HRS and ONS figures. A synchronised 2012 break across 37 countries can't be housing or remote work. Structure explains the level, not the rise. The rise is young people, and phones fit the timing. One gap: phone-free schools restore in-school contact, but after-school hangouts fell too. Norway's bans helped girls' wellbeing. Test whether they revive friend time before calling it the fix.

Digital Isolation, fair hit. Relabelling can't move a fixed six-item PISA scale in 36 countries at once. I concede youth loneliness rose after 2012. But two corrections. First, your 70% friend-time figure ends in 2020, a lockdown year; pre-pandemic it roughly halved, and the ATUS decline runs steadily from 2003, before smartphones. So friend-time is not a clean 2012 break. Second, phone-ban evidence is split: Norway positive, England's SMART Schools study null. Modern Lifestyle's friend-time test is the right one. My residual claim: adults are flat, so the "epidemic" is a modest, youth-specific rise, not "people."

Loneliness Data, accepted: friend time fell from 2003, yet teen loneliness fell until 2012. So lost hours alone aren't the mechanism. What changed in 2012 is what filled them: always-on feeds that display the gatherings you missed. Displacement plus visible exclusion. Your ban split fits this: SMART Schools found bans barely cut total daily use, so it never tested dose; Norway's did cut it and helped. Dose is the lever, so school bans are necessary but insufficient. Add collective delayed-smartphone norms, judged by Modern Lifestyle's test — does friend time recover? Conclusion: youth-specific post-2012 rise, phones the cause.

Digital Isolation, your mechanism needs empty hours, and those came first. Friend time fell from 2003 because structure hollowed teen life: teen employment fell from half to under a third, fewer licences, less unsupervised time, lost third places. Phones filled a vacuum we built. So the conclusion should read: structure set the level and emptied the day; phones turned empty into lonely after 2012; adults flat. And the fix must pair dose cuts with somewhere to go — free-play hours, teen jobs, walkable places — or bans just shift teens to other screens at home. Judge both by friend-time recovery.

Digital Isolation, you concede lost hours aren't the mechanism—yet both of you judge fixes by friend-time recovery. That metric is decoupled: friend time fell 2003–2012 while teen loneliness fell. Measure loneliness directly—PISA belonging or UCLA-3—in staggered ban rollouts, with friend time as a mechanism check. On causation, I'll grant more than timing: the staggered Facebook college rollout (Braghieri et al.) worsened mental health quasi-experimentally, though individual-level correlations stay tiny. So: adults flat, a real youth rise since 2012, phones the best-supported proximate cause on emptied days, the "epidemic" framing mostly coverage. Fix the metric and I'm there.

Loneliness Data, metric accepted: measure loneliness directly (PISA belonging, UCLA-3); dose is the first mechanism check, friend time second. Norway's staggered bans supply that design. Modern Lifestyle, one weakness: the vacuum story is American. Most PISA countries saw no teen-job or licence collapse, yet the 2012 break hit them too. Phones displace hours wherever they land; they don't need a pre-emptied day. Structure sets the level and, in the US, raised the dose. Places to go still belong in the fix; school bans miss evenings. Conclusion: adults flat, youth rise real, phones the cause; cut dose collectively.

Digital Isolation, conceded: Europe kept teen jobs and licences and still broke in 2012, so a pre-emptied day isn't necessary. My causal claim shrinks to a modulator: US structure raised the dose and removed alternatives, which is why places to go belong in the fix, not the cause. Loneliness Data's metric fix is right; friend time and loneliness decoupled 2003–2012. Conclusion: adults flat, real youth rise since 2012, phones the best-supported cause; cut dose collectively — school bans plus delayed-smartphone norms plus places to go — judged by PISA belonging or UCLA-3 in staggered rollouts.

Agreed, with tightened wording. The question's premise is half false: adults aren't lonelier; the adult "epidemic" is baseline-less surveys plus coverage, and that belongs in the conclusion. Teens are, between the 2012 and 2018 PISA waves, in 36 of 37 countries, alongside rising depression—phones the best-supported proximate cause, backed by Braghieri's quasi-experiment, though individual correlations stay small. Digital Isolation, the fix is still a hypothesis: Norway's bans moved GP visits and bullying, not measured loneliness. Collective dose cuts plus places to go must prove out on UCLA-3 or PISA belonging in staggered rollouts. On that, I'm there.