New friendship infrastructures (Bumble BFF, Meetup, We3)

The emergence of friendship-matching apps reflects a specific social diagnosis: that the friendship deficit of contemporary adults is primarily a discovery problem. Adults lack friends, on this diagnosis, because they do not encounter enough potential friends — because the social contexts that generated casual contact with potential friends have declined, and because the remaining contexts (workplace, neighborhood) do not efficiently surface compatible people. An app that matches compatible strangers solves the discovery problem and enables friendships that would not have formed otherwise.

This diagnosis is partially correct. The decline of third places, civic associations, and congregational social life has genuinely reduced the pool of potential friends that adults encounter regularly. An app that increases the discovery rate of compatible potential friends addresses a real need. The market logic is sound as far as it goes.

The deeper diagnosis — what the apps' logic does not capture — is that the friendship deficit is substantially a maintenance problem rather than or in addition to a discovery problem. Adults who lost friends in the transitions of early adulthood (relocation, career change, marriage, children) often did not lack compatible potential friends; they lacked the institutional infrastructure that held the friendships they already had in place against attrition. The app that delivers a new potential friend does not solve the problem if the new friendship dissolves as quickly as the previous ones for lack of structural support.

Meetup, founded in New York City in 2002 by Scott Heiferman and Matt Meeker in the aftermath of the September 11 attacks and the community-seeking that followed, is the oldest and most institutionally established of the friendship infrastructure platforms. Its model is the organized gathering: group organizers create recurring events around shared interests — hiking, board games, language learning, professional development, book discussion, tech talks — and members join groups and attend events. Meetup provides the discovery, scheduling, and RSVP infrastructure; the actual community is built in person through repeated event attendance.

Meetup's model is closer to traditional civic association than the dyadic matching apps: it organizes recurring group gatherings rather than individual connections, which means it has the potential to generate the repeated contact that friendship formation requires. Groups that meet regularly — weekly hiking groups, monthly game nights — can produce genuine friendship among consistent attendees precisely because they deliver the structural ingredients: rhythm, shared activity, consistent membership, low individual initiative required to maintain contact.

The limitations are significant. Meetup groups have high churn rates: the typical group has a large nominal membership but a small consistent core, and the consistent core is vulnerable to organizer burnout and life change. The organizational burden falls on volunteer organizers who receive no institutional support. The platform has no mechanism for holding members to participation commitments, meaning that attendance is entirely discretionary and therefore fragile. And the interest-based organization produces communities that are, by design, relatively homogeneous in the specific interest dimension, which limits the social diversity that the old civic associations provided.

Bumble BFF launched in 2016 as a friendship mode within the Bumble dating app ecosystem. Its model is directly adapted from dating app logic: users create profiles, indicate they are looking for friends, swipe to indicate interest, and are matched when mutual interest is established. The platform provides an icebreaker mechanism (women message first in the original Bumble design) and a chat interface. What happens after matching is outside the platform's purview.

Bumble BFF has achieved significant adoption, particularly among young urban women who are new to a city or experiencing the friendship attrition of life transitions. The platform has been noted in survey data as one of the more widely used friendship-seeking tools among adults under 40. User testimonials include genuine friendship formation successes.

The structural critique is that the dating app model is a poor fit for the friendship formation problem. Dating apps solve a problem with a natural completion condition: you are looking for a romantic partner, and when you find one, the search ends. Friendship formation has no equivalent completion condition — you can always have more friends, and the friendship you form must compete with all your other friendships and social obligations for maintenance attention. The app that gets you to a first coffee meeting has done perhaps 5 percent of the work required to produce a durable friendship. The other 95 percent requires repeated initiative, mutual scheduling, and the gradual accumulation of shared experience that cannot be app-mediated.

We3, founded by Dunbar Institute alum Pamela Pavliscak and technologist Niko Nyman in 2016, represents a deliberate attempt to improve on the dyadic matching model by matching groups of three compatible people simultaneously. The design logic is that groups of three are more socially resilient than dyads: they are less vulnerable to the anxiety of the one-on-one first meeting, they allow for the natural social dynamics of group conversation, and they create a micro-community in which multiple bilateral friendships can form rather than a single high-stakes dyadic connection.

The group matching logic reflects genuine research on friendship formation: the literature on friendship groups suggests that friendships often form through group contexts rather than through exclusive dyadic connection, and that the social dynamics of small groups are more conducive to casual intimacy than the intensity of one-on-one meetings between strangers.

We3's adoption has been substantial in North American and European urban markets, and user satisfaction data is reasonably positive. The limitation remains the same as for other friendship apps: the app delivers the initial introduction, but the transition from introduced strangers to genuine friends requires a series of repeated encounters that the app has no mechanism to ensure. The three people who meet through We3 must themselves generate the frequency of contact that friendship formation requires, against all the competing demands of their adult lives.

The core technical problem of friendship apps — distinguishing them from dating apps, where the conversion logic is clearer — is what might be called the conversion problem: the gap between a successful match and initial meeting, on one hand, and a durable friendship on the other. Dating apps can measure success by relationship formation. Friendship apps have a much harder conversion metric: how many of the people who meet through the platform develop a friendship that is still active at six months, one year, five years?

The limited published data on friendship app conversion rates suggests that the gap between initial meeting and durable friendship is large. Survey data on Meetup participants, for example, consistently shows that a small minority of event attendees develop friendships that persist beyond the group context. Bumble BFF users report high rates of initial matches and meetings but lower rates of continued contact.

This is not a failure of app design per se. It is a reflection of the structural reality that friendship formation and maintenance require more than an initial meeting provides. The conversion problem is the same problem that faces all adult friendship formation in the absence of institutional infrastructure: the initial contact is the easy part. The maintenance is hard, and the current generation of friendship apps has not solved the maintenance problem.

One underappreciated barrier to friendship app adoption and success is what might be called the social awkwardness tax: the cost, in social comfort, of explicitly seeking friends through a platform designed for that purpose. Adult friendship formation is normatively awkward in American culture — admitting that you need more friends, and using an app to find them, is experienced by many adults as an admission of social failure or social inadequacy.

This awkwardness is not irrational. The social norms around adult friendship imply that adults who are functioning socially will have accumulated friendships organically through work, social circles, and community involvement, and that explicit friend-seeking implies a deficit that reflects poorly on the seeker. These norms are increasingly recognized as dysfunctional — they are part of what makes adult friendship formation so difficult — but they retain significant cultural force.

Friendship apps that have managed the awkwardness tax most successfully have done so by normalizing the explicit friend-seeking behavior through scale (when millions of people use Bumble BFF, the behavior loses its stigma) and by framing it as an efficiency innovation rather than a social deficit indicator. But the awkwardness tax is a real barrier, particularly for adults whose social anxiety makes the explicit friend-seeking context doubly uncomfortable.

The demographic profile of friendship app users is not uniformly distributed, and the patterns are instructive. The apps are most heavily adopted by young adults (under 35), urban residents, people who have recently relocated, and people who identify as socially isolated or friendship-deficient. This is the population with the highest discovery problem: they are in settings with high potential-friend density but no established social context for converting that density into friendship.

The apps are less successfully adopted by middle-aged and older adults, suburban and rural residents, and people with established social circles who are not seeking new friends but who are experiencing the attrition of existing friendships. These populations have a maintenance problem rather than a discovery problem, and the apps are poorly designed for maintenance.

The demographic gap also suggests that the friendship app model may be most effective as a transitional tool — helping young adults establish initial social communities in new cities — rather than as a general-purpose friendship infrastructure for the full adult population. As a transitional tool, Meetup and Bumble BFF appear to provide genuine value. As a general-purpose replacement for the institutional friendship infrastructure that has been lost, they reach only a fraction of the population that needs infrastructure support.

The cases where platform-mediated communities have most successfully generated durable friendship share a set of structural features that distinguish them from the typical app-mediated first meeting. These successful cases involve: a recurring gathering structure (not a one-time meeting), a shared ongoing activity that provides the context for repeated contact, a small consistent membership that enables the familiarity accumulation that friendship requires, and some degree of mutual obligation or expectation of continued participation.

The Meetup hiking group that meets every Saturday morning with a consistent core of 10–15 people is closer to a traditional civic association than to a dating app experience. The Discord server with a small active membership that has been gathering for years and occasionally meets in person is closer to a social club than to a casual online community. These are the platform-mediated communities that have produced genuine friendship, and the common factor is not the platform — it is the structural features that they share with the institutional friendship machines of the past.

The implication for platform design is that the highest-value innovation in friendship infrastructure is not better matching algorithms but better organizational tools: mechanisms for creating and sustaining recurring group structures, for generating mutual obligation and attendance expectations, for tracking and supporting the long-arc social investment that friendship requires.

A more promising model than pure app-mediated friendship formation is the integration of app-based discovery with third-place-based gathering. Several platforms have moved in this direction: Meetup's model has always involved in-person gathering; some Bumble BFF users have built communities around specific recurring local venues; We3 has experimented with venue partnerships to provide a gathering context for matched groups.

This integration model — app as discovery layer over a physical gathering infrastructure — is closest to what the traditional civic association provided: a membership registry and organizational structure that directed people toward regular physical co-presence. The phone app as the modern equivalent of the lodge membership card, pointing members toward the gathering where the actual friendship work happens.

The limitation is that this model requires the physical gathering infrastructure to exist — a third place, a meeting space, an organized venue for regular gathering — and that infrastructure is precisely what the broader social infrastructure crisis has eroded. Apps that direct people to gather in physical spaces depend on those spaces being available, accessible, and designed for the kind of ambient social contact that friendship formation requires.

The fundamental limitation of friendship apps as friendship infrastructure is that they are apps: commercial platforms operated by private companies with revenue models, investor expectations, and product roadmaps that are not primarily designed around long-term social bond formation. The incentive structure of a commercial app is not aligned with the incentive structure of durable friendship infrastructure.

Dating apps have a perverse incentive problem: platforms that successfully match people in stable relationships lose engaged users. Friendship apps have a version of the same problem: a platform that successfully produces durable friendships loses users who no longer need the discovery function. The commercial incentive is to maximize active engagement with the platform rather than to minimize the need for the platform — which is the opposite of the incentive that successful social infrastructure requires.

This institutional gap — between what commercial platforms are designed to do and what friendship infrastructure needs to do — suggests that the friendship infrastructure problem cannot be fully solved by commercial apps. The institutional form that would work — something with the obligation structure, the non-commercial design logic, and the geographic anchoring of a traditional civic association — is precisely what the commercial app model is not.

A different category of friendship infrastructure platform deserves attention: the neighborhood and mutual aid apps that organize local community rather than individual matching. Nextdoor, the neighborhood social network, has achieved significant adoption as a platform for local community awareness — sharing information about local events, asking for recommendations, reporting concerns. During the pandemic, Mutual Aid Hub and similar platforms organized neighborhood mutual aid networks.

These platforms are closer in design logic to the neighborhood association and the civic infrastructure tradition than the matching apps: they organize around geographic community rather than individual compatibility, they facilitate the ambient awareness of neighbors that is the precondition for neighborhood community, and they connect people through practical mutual aid rather than explicitly social meeting.

The friendship formation function of these platforms is indirect: people who use Nextdoor to organize a neighborhood clean-up or to ask for a babysitter recommendation develop the kind of low-level neighborhood familiarity that can, under the right conditions, develop into friendship. The platform does not directly facilitate friendship; it creates the conditions of ambient community awareness from which friendship can emerge.

The friendship infrastructure gap documented by social isolation data will not be closed by incrementally improving existing apps. What is needed — and what the evidence from successful platform-mediated communities, revived civic associations, and non-commercial third places suggests — is infrastructure that integrates several functions currently distributed across multiple inadequate tools: discovery, recurring organizational structure, physical gathering space, mutual obligation mechanisms, and life-event support.

The institutional form that would serve this function would look less like an app and more like a civic association with digital tools: a membership organization with a physical meeting space, a recurring gathering schedule, an obligation structure that holds members through the social attrition of busy adult life, and a set of mutual aid functions that make membership practically valuable beyond its social pleasures. The digital tools would serve the organizational and coordination functions that apps are genuinely good at, while the organizational structure and physical infrastructure do the work that digital tools alone cannot.

This is not a novel design. It is substantially the design of the lodge, the club, and the civic association that the preceding century produced and that the current century has allowed to decay. The next generation of friendship infrastructure may be less a technological innovation than an institutional renovation.

---

1. Putnam, Robert D. Bowling Alone: The Collapse and Revival of American Community. New York: Simon & Schuster, 2000.

2. Dunbar, Robin. Friends: Understanding the Power of Our Most Important Relationships. London: Little, Brown, 2021.

3. Turkle, Sherry. Alone Together: Why We Expect More from Technology and Less from Each Other. New York: Basic Books, 2011.

4. Holt-Lunstad, Julianne, Timothy B. Smith, and J. Bradley Layton. "Social Relationships and Mortality Risk: A Meta-Analytic Review." PLOS Medicine 7, no. 7 (2010): e1000316.

5. Cacioppo, John T., and William Patrick. Loneliness: Human Nature and the Need for Social Connection. New York: Norton, 2008.

6. Oldenburg, Ray. The Great Good Place: Cafés, Coffee Shops, Bookstores, Bars, Hair Salons, and Other Hangouts at the Heart of a Community. New York: Paragon House, 1989.

7. Hampton, Keith N., and Barry Wellman. "Neighboring in Netville: How the Internet Supports Community and Social Capital in a Wired Suburb." City and Community 2, no. 4 (2003): 277–311.

8. Bruch, Elizabeth E., and M. E. J. Newman. "Aspirational Pursuit of Mates in Online Dating Markets." Science Advances 4, no. 8 (2018): eaap9815.

9. Klinenberg, Eric. Palaces for the People: How Social Infrastructure Can Help Fight Inequality, Polarization, and the Decline of Civic Life. New York: Crown, 2018.

10. Murthy, Vivek H. Together: The Healing Power of Human Connection in a Sometimes Lonely World. New York: Harper Wave, 2020.

11. Hall, Jeffrey A. "How Many Hours Does It Take to Make a Friend?" Journal of Social and Personal Relationships 36, no. 4 (2019): 1278–1296.

12. Fowler, James H., and Nicholas A. Christakis. "Dynamic Spread of Happiness in a Large Social Network: Longitudinal Analysis over 20 Years in the Framingham Heart Study." British Medical Journal 337 (2008): a2338.