Loneliness is a public-health emergency. Friends? is a tool.

• 1 in 5 American adults report having no close friends. (Pew Research Center, 2024.)
• Isolation carries mortality risk comparable to smoking 15 cigarettes per day. (U.S. Surgeon General, Our Epidemic of Loneliness and Isolation, 2023 advisory.)
• 79% of Gen Z adults report feeling lonely — the loneliest generation on record. (Cigna, 2023.)
• Americans spend less time with friends than at any point since measurement began. Time spent with friends fell ~70% for young adults over the last two decades. (American Time Use Survey / BLS.)
• Loneliness is associated with 29% higher heart disease risk and 32% higher stroke risk. (U.S. Surgeon General 2023, citing Valtorta et al., Heart.)

The tools most people have — Bumble BFF, Tinder, Hinge — are designed to maximize swipes. Instagram and TikTok are designed to maximize dwell time. Meetup.com rewards RSVP counts. None of these are measured against the thing we actually need: real friendships, formed off-platform, that persist.

The research is consistent that quality of social tie, not quantity, predicts wellbeing. You don't need 500 followers. You need 3–5 close friends you see regularly. That's the number Friends? is optimizing for.

Friends?'s north-star metric is off-platform meetings per active user per week. Not matches. Not swipes. Not DAUs. Actual face-to-face time generated by the product. Every feature is subordinate to that one number.

The matching model uses the Big Five personality traits via the IPIP-NEO-120 instrument. The AI companion plans the first hang. The DMs are end-to-end encrypted. All three exist because they collectively increase the odds of a real meetup happening.

The headline claim from the 2023 U.S. Surgeon General advisory — that social isolation increases mortality risk comparable to smoking up to 15 cigarettes a day — is one of the most-cited and most-misunderstood claims in modern public health. Here is where it actually comes from.

The number traces to a 2010 meta-analysis by Holt-Lunstad, Smith, and Layton, published in PLOS Medicine. They pooled 148 studies covering 308,849 individuals and found that people with stronger social relationships had a 50% greater likelihood of survival over the follow-up periods, controlling for age, baseline health, and other standard mortality covariates.

A 50% survival advantage corresponds to a hazard ratio of roughly 1.50 for socially isolated vs. well-connected individuals. The hazard ratio for moderate smoking (10-20 cigarettes/day) is in the 1.4-1.6 range across the major cohort studies (Doll et al., 2004; Pirie et al., 2013). Same ballpark. That’s the math behind the cigarette equivalence.

The 2015 follow-up meta-analysis by the same group extended the analysis to 70 prospective studies and found:

• Social isolation increases mortality risk by 26%.
• Loneliness (subjective experience) increases it by 26%.
• Living alone increases it by 32%.

These three measures are correlated but not identical. Living alone is structural; social isolation is observable behavior; loneliness is felt. All three independently predict mortality. The full citation breakdown is in the journal.

The mortality effect isn’t metaphysical. It runs through measurable physiological and behavioral channels. Three of the most-replicated mechanisms:

• Chronic stress activation. Lonely individuals show elevated cortisol, inflammatory cytokines, and blood pressure on average. Over years, this elevated baseline damages cardiovascular tissue. Cole et al. (2007, Genome Biology) found that loneliness shifts the genome-wide transcription pattern of leukocytes toward an inflammatory profile — a measurable, mechanistic effect, not a metaphor.
• Behavioral risk amplification. Lonely people exercise less, eat worse, sleep worse, drink more, and skip preventive medical care more often. None of these individually is dramatic; collectively they compound. The mortality math substantially flows through these mediators.
• Cognitive decline. Multiple longitudinal studies (notably the English Longitudinal Study of Ageing, ELSA) find that loneliness predicts faster cognitive decline and higher dementia risk in older adults, independent of depression. The mechanism appears to involve reduced cognitive engagement plus chronic stress effects on the hippocampus.

The takeaway: loneliness isn’t just an unpleasant subjective experience. It produces measurable physiological changes that manifest as elevated all-cause mortality 5-15 years downstream. Which is why public-health researchers now treat it as a tier-one risk factor alongside smoking, obesity, and physical inactivity.

The loneliness epidemic isn’t a moral failing of an individual generation. It tracks measurable changes in the structural infrastructure of friendship over the past 50 years. The most-cited drivers:

• The decline of third places. Sociologist Ray Oldenburg coined the term third place in 1989 — places that aren’t home or work but where casual social life happens. Bars, churches, civic clubs, barbershops, libraries, public parks. Robert Putnam’s Bowling Alone (2000) documented sharp declines in third-place participation across nearly every category in the U.S. since the 1970s. The 2020s data extends the trend.
• Geographic mobility. Roughly one-quarter of U.S. adults moved to a new metro in the past 5 years. Each move severs the existing friend graph. Re-formation takes 18-36 months and many people don’t complete it before moving again.
• Remote work + delivery. The ambient social contact that used to come free with going to an office, a restaurant, a grocery store, or a gym is now optional in a way it wasn’t pre-2020. Cigna’s 2023 index found full-remote workers report higher loneliness than hybrid or on-site workers, controlling for age and household composition.
• Two-income households + longer commutes. The disposable-time budget for adult friendship shrank dramatically in the late 20th century. Most adults under 40 now report <5 hours per week of unstructured social time, down from ~10 hours in the 1980s.
• Social media substitution. Parasocial connection without reciprocity. Time spent feeling vaguely connected to many strangers displaces time spent actually connected to a few real friends. The substitution effect is well-documented (Twenge et al., 2018, Emotion).

None of these are individually fixable by any one person. But they are diagnostic — they point at the structural reasons adult friendship is harder than it used to be, which makes systematic interventions (recurring activity participation, intentional planning, tools that close the “hey” gap) the right kind of fix.

Most loneliness advice is either platitudes (“put yourself out there”) or therapy referrals. The intervention research actually points at three specific things that work, ordered by evidence strength:

1. Join a recurring small-group activity. The strongest single behavioral predictor of friendship-graph rebuild is participation in a recurring weekly group of 4-12 people with a shared activity (sports, religious, hobby, volunteer). Three months of consistent attendance is the threshold where new friendships durably form.
2. Initiate the second meeting. Adult friendships die at the gap between first meeting and second meeting (the second-ask gap). Adults who reliably send a specific second-meeting invitation within 72 hours of a first meeting build durable friend graphs. Adults who wait for the other person to do it mostly don’t.
3. Cognitive-behavioral therapy for loneliness. The single most-effective clinical intervention is CBT specifically targeting maladaptive social cognition (Masi et al., 2011, meta-analysis). The mechanism is correcting the negative-attribution patterns lonely people develop (“they didn’t reply because they don’t actually like me,” etc.) before those patterns sabotage further attempts at connection.

Are We Friends? is built around the first two of these. Personality matching + AI-planned first hangs collapse the activation energy of finding a recurring activity AND closing the second-ask gap into one product. The third (CBT) is out of scope for an app, but if loneliness is severe and persistent, getting clinical help is genuinely worth it.

• Office of the U.S. Surgeon General (2023). Our Epidemic of Loneliness and Isolation. Advisory on the healing effects of social connection and community. HHS.gov source PDF.
• Pew Research Center (2024). Surveys on friendship and social connection in the U.S. adult population.
• Cigna (2023). Loneliness in America. Employer and generational loneliness study.
• Valtorta, N. K. et al. (2016). "Loneliness and social isolation as risk factors for coronary heart disease and stroke." Heart.
• Holt-Lunstad, J., Smith, T. B., & Layton, J. B. (2010). "Social relationships and mortality risk: a meta-analytic review." PLOS Medicine.