You know the shape of the evening. A room hired for the occasion, name tags on a trestle table, forty or fifty people holding drinks at chest height. You came to meet people. Everyone came to meet people. And somewhere around the ninety-minute mark you realise you’ve had six conversations about what everyone does for work, you can’t remember a single name, and the most comfortable stretch of the night was the twenty minutes you spent talking to the one person you already knew.
It’s tempting to read that as a personal failing: not outgoing enough, no good at working a room. It isn’t. The mingling format itself stalls, in ways researchers have measured, and the alternative is smaller and quieter than most social advice admits: one person, across a table, for about an hour.
What actually happens in a room full of strangers
Two Columbia Business School professors, Paul Ingram and Michael Morris, once put the mixer format under a microscope. For their 2007 study “Do People Mix at Mixers?” in Administrative Science Quarterly, they fitted around a hundred businesspeople at an evening mixer with electronic name tags that logged, second by second, who stood facing whom. Beforehand, guests overwhelmingly said the same thing: they were there to make new contacts. On average each guest already knew about a third of the room, which left two-thirds of it (dozens of genuinely new people) on offer.
The tags told a different story. Guests were far more likely to spend the evening with the people they already knew. Given a room engineered for meeting strangers, most of us drift back to the familiar. It isn’t laziness: breaking into a circle of strangers mid-anecdote is genuinely hard, and the room offers a socially acceptable way to avoid ever doing it.
Conversations have a ceiling, and it’s low
Suppose you do break in. Now you meet the second problem: conversation itself has a maximum size. The evolutionary psychologist Robin Dunbar (of “Dunbar’s number” fame) watched freely forming conversations in cafeterias, bars and receptions for a 1995 study in Human Nature with Neil Duncan and Daniel Nettle, and found an upper limit of about four people on spontaneous conversation. Add a fifth and the group doesn’t become one bigger conversation. It splits into two smaller ones.
A 2018 paper in Evolution and Human Behavior by Jaimie Krems, Dunbar and Steven Neuberg calls this the “dinner party problem”: the real reason a table of eight always ends up as two or three conversations rather than one. Their explanation is cognitive. Following a conversation means keeping a running model of what each speaker knows, means and wants, and most of us can track about three other minds at once. Four people is a ceiling set not by politeness but by working memory.
Even below that ceiling, group talk changes character as it grows. Psychologists Nicolas Fay, Simon Garrod and Jean Carletta compared five-person and ten-person discussions in a 2000 study in Psychological Science. Five-person groups still behaved like dialogue: people responded to each other and were most influenced by the person they’d actually been talking with. Ten-person groups drifted into what the researchers called serial monologue: one dominant speaker holding forth while everyone else rotates through the role of audience. If you’ve ever stood in an event circle listening to the loudest person tell a third consecutive story, you’ve seen the finding live.
Friendship runs on disclosure, and disclosure needs a pair
None of this would matter if friendship could form in an audience. It can’t, because the raw material of closeness is self-disclosure, the gradual trade of real information about who you are. And disclosure behaves very differently in front of a crowd.
The evidence here is unusually solid. In 1994, Nancy Collins and Lynn Carol Miller published a meta-analysis of 94 studies in Psychological Bulletin and found the effect runs in three directions at once. We like people more after they open up to us. We open up more to people we like. And, strangest of all, we like people more because we’ve opened up to them. Disclosure isn’t a symptom of closeness. It’s the mechanism.
It also has to be traded, not performed. Susan Sprecher and colleagues showed in a 2013 experiment in the Journal of Experimental Social Psychology that strangers who took turns disclosing (back and forth, a little at a time) came away liking each other more and feeling closer than pairs where one person disclosed at length before the roles swapped. Reciprocity works at close range and in small increments. A pair can do that for a whole hour. A circle of six cannot; by the time the turn comes back around, the moment has moved on.
The cleanest demonstration of what a pair can do remains Arthur Aron’s 1997 “fast friends” study in Personality and Social Psychology Bulletin, the one the famous “36 questions” come from. Aron and his colleagues sat pairs of strangers together for 45 minutes with questions that escalated gently from small talk towards things that matter. Pairs who did the escalating version reported far more closeness than pairs who made ordinary small talk for the same 45 minutes: on average, closeness in the range people report for their real, established relationships. Two strangers. One conversation. Under an hour. The study has been cited thousands of times, and its quiet precondition is easy to miss: the whole design is two people, taking turns. It only works as a pair.
One at a time is not the slow way
Put the pieces together and the arithmetic of the mixer looks different. A night of mingling gives you fragments: six half-conversations inside groups too large for anyone to say anything real, most of them ending when someone drifts off to refill a glass. An hour across a table gives you the one format where the research says closeness actually compounds: two people, turn-taking, no audience, nowhere for the conversation to split.
This isn’t an argument against ever going to things. Group events are decent at one job: surfacing candidates. The run club, the book club, the industry night can show you who’s out there. But the conversion, the moment an interesting stranger starts becoming an actual friend, almost always happens later, in the follow-up coffee, the walk, the pair peeling away from the group. Most of us never get around to asking. The event feels like the social effort, so we let it stand in for the part that works.
That conversion step is the entire premise of Flat White. No room full of name tags, no circle to break into: just one person, matched to you, one café, one hour. It skips the mingling and starts where the research says friendship starts: two chairs, facing each other.
Meeting one person at a time sounds slow. It’s the opposite. It’s every minute spent inside the only conversation size that was ever going to work.
Meet one person. Over coffee. On purpose.
Australia’s opening city-by-city. Join the waitlist at joinflatwhite.com →