On Loneliness

“It is tough. Honestly, I’m gonna be very vulnerable. One of the tattoos I have on me is misspelled. There’s an extra S right there.”

Said to a room of classmates during a leadership workshop, September 21, 2026, three weeks after moving to Chicago.

That is a small confession, and it was made to people who had known each other for about twenty days. It is also a working example of the thing this page is about. Everyone in that room arrived alone, is spending a good deal of energy appearing not to be, and is quietly running a forecast about everyone else. The forecast is wrong in a predictable direction, and the fastest way to correct it is to do what that sentence does: hand the room something unpolished and see what comes back.

Loneliness gets described as the absence of connection. The research reads better as the reverse. Loneliness is the signal that connection is needed, it is nearly universal in a room like this one, and it comes with a design flaw that makes the obvious remedy feel dangerous exactly when it is most available. Being vulnerable is how you get around the flaw. What follows is the mechanism, in four parts, with the Chicago research that happens to have measured it a few miles from the Harper Center.

The base rate matters because the whole loop below runs on a bad estimate of it. In Harvard’s 2021 national survey, 36% of adults said they had felt lonely frequently or almost all the time in the previous four weeks. Among people aged 18 to 25 the figure was 61%.^[1] Two years later the U.S. Surgeon General opened an advisory on the subject with the estimate that about one in two adults reports measurable loneliness.^[2] None of these is a survey of business schools, and an MBA section is older than the youngest band; the honest reading is that the number for a first-year cohort sits between the two rates, and that placement is an inference.

The reason the room matters more than the national number: nobody in it believes the national number applies to the people they can see. This is the same finding, from the inside, as the one Booth’s own research turns up below. People systematically underestimate how much others are interested in them, how much others like them, and how much others are carrying. The room looks finished from every seat, which is how sixty people can be lonely in the same building at the same time and each conclude they are the only one.

John Cacioppo spent most of his career at the University of Chicago arguing that loneliness is not a mood but a signal, in the same category as hunger, thirst, and pain. Each is an aversive state whose job is to change behavior: hunger sends you toward food, loneliness sends you toward people.^[3] The reason the signal is so unpleasant is the reason it exists. In the environment it evolved for, being cut off from the group was a survival problem, and a body that could feel that problem early did better than one that could not.

Ignoring the signal has a measured cost. A 2015 meta-analysis of seventy studies found loneliness associated with a 26% higher likelihood of mortality over the follow-up periods, social isolation with 29%, and living alone with 32%, after the usual confounds were controlled.^[8] Those numbers are here for one reason: the feeling that arrives in a new city, three weeks in, after the family member who helped with the move has flown home, is doing its job, and it is the accurate part of the system.

“The point of that truth is mostly just to say that I value vulnerability because it leads to human connection.”

Voice memo, June 18, 2026.

That sentence was recorded three months before the move, and it reads now like a plan. The signal says connection is needed. The next part is where the plan meets the design flaw.

Here is the part that took decades of research to pin down, and that anyone who has been lonely already knows from the inside. The lonely brain does not become more open. It becomes more vigilant. Cacioppo and Hawkley documented an implicit hypervigilance for social threat: lonely people attend faster to signs of rejection, remember negative social information better, and read ambiguous cues as cold.^[3] In one eye-tracking study, lonely young adults’ gaze was captured sooner, and held longer, by faces showing rejection than by faces showing acceptance.^[5] The model that grew out of this work calls the whole sequence the reaffiliation motive: the signal fires, the person withdraws to appraise, the scan for threat begins, and the scan finds what it is looking for.^[4]

The loop explains the cruelest feature of a new place. The people who most need to approach someone are the ones whose hardware has just been tuned to make every approach look like a bad bet. A section that has already formed its lunch groups, a classmate who answers a message the next morning, a study team that seems to have an inside joke: each is exactly the kind of neutral cue the scanner files as rejection. From inside the loop, going first feels like the worst possible strategy. From the data, it is the only one that returns new information.

The best evidence that the scanner is miscalibrated comes from this campus. In 2022, Michael Kardas, Amit Kumar, and Nicholas Epley published twelve experiments in which strangers were handed deep questions and asked, before and after, how the conversation would go. The questions were not small talk: for what in your life do you feel most grateful; if a crystal ball could tell you the truth about yourself, what would you want to know; can you describe a time you cried in front of another person.^[6] Participants expected these to be awkward, expected to feel little connection, and expected their partner to be only mildly interested. They were wrong on all three, and wrong in the same direction every time.

The mechanism the paper identifies is the one the loop predicts. The miscalibration was carried by a single belief: that the other person would not care. Participants underestimated their partner’s interest before the conversation and, more strikingly, after it, when the partner had already reported being interested. Mediation analyses showed that this error in reading the other person explained most of the error in predicting awkwardness and connection.^[6] The estimate of the room, and only the estimate, is what runs cold.

One strange detail is worth preserving because a cleaner essay would cut it. In the second experiment, people who had the deep conversation reported feeling slightly more lonely afterward than people who had the shallow one, 1.82 against 1.65 on the scale used, even though they also felt more connected to the person across from them. The authors did not predict this and suspect the cried-in-front-of-someone question surfaced memories of past isolation.^[6] That is the whole argument in one data point. The conversation that connects you is the same one that names what you were missing. Connection and loneliness arrive together, because the second one is how you know the first one landed.

Epley had already run the cruder version of this experiment on the Metra. In 2014 he and Juliana Schroeder sent commuters out of the Homewood station with instructions to talk to a stranger, sit in silence, or do whatever they usually did. Riders predicted the conversation condition would be the least pleasant. It was the most pleasant, by a wide margin. In a follow-up, commuters guessed that fewer than half of the people around them would be willing to talk; of those who tried, not a single person reported being turned down.^[9] Same city, same error, same direction.

The forecast error has a shape over time, and the shape is reassuring if you know it and corrosive if you do not. In 2018 a team at Yale followed 102 first-year students assigned to share dorm suites, and asked them five times across the year how much they liked each suitemate and how much they believed each suitemate liked them. The gap between those two numbers, which the authors named the liking gap, was present in September, October, December, and February. In May it was gone.^[7]

Two things about that May. The authors offer two explanations, and both are about disclosure. Either the students finally knew each other well enough for the estimate to catch up with the truth, or the deadline for choosing next year’s housing forced conversations that revealed liking directly.^[7] On the second reading, the gap did not close because time passed. It closed because something made people say out loud what they had been leaving to inference. A first-year MBA program has the same calendar and no forced May. The conversations that close the gap have to be chosen.

The other relevant result is older and about belonging rather than liking. Walton and Cohen’s belonging intervention, the one that halved an achievement gap over three years, consisted of showing first-year students survey results indicating that most students worry about whether they belong, and that the worry fades.^[10] The intervention worked by changing the interpretation of a normal feeling, not the feeling. The problem was never the worry itself but the habit of reading it as evidence about oneself.

Put the four parts together. The signal is accurate. The room is full of people receiving the same signal. The loop makes each of them read the others as uninterested. The only measurement that corrects the loop is an actual disclosure and the response to it, and the research says the response is warmer than anyone predicts, by a margin large enough to be measured in a dozen experiments and to persist for eight months of a dorm year. Vulnerability, on this reading, works less like a personality trait than like an instrument, the one that gets the loop a better reading, and the person who goes first is the one who collects the connection everyone else in the room was waiting for.

“I’ve been really intentional about being as open as I can. My mission here is to not hold back.”
“Because I’m really grateful how much people open up to me when I do say the unpolished.”

Said to classmates during a public-speaking workshop, September 22, 2026. Two fragments from consecutive turns.

The second fragment is the data coming back. It is the Booth study, run informally, in a room where the participants have to see each other again on Monday. The unpolished thing goes out, and the partner’s interest turns out to be higher than forecast. Repeated enough times, the loop stops being the operating system and becomes one input among several.

“You never get this vulnerability from professionals out in the real world, so it’s nice to feel like the people I’m talking to actually, for as oppressive and scary as someone from IB has always been to me, now I’m like they’re just nerds too.”

Recorded during recruiting preparation, September 22, 2026.

That is the scanner being corrected by a data point. A category that read as threat from a distance, once approached, turns out to contain people who are also new, also guessing, and also relieved when someone goes first. The line from an older page on this site describes the alternative and its price:

“I would rather be known than be right. It took me a long time to actually mean that. Being right is armor, and armor works: it keeps you safe and it keeps you alone.”

From On Leaving Ego at the Door, already published on this site.

A C. S. Lewis line about friendship sits on the Jennisms page, adopted there in July:

“Friendship is born at the moment when one person says to another, ‘What! You too? I thought I was the only one.’”

C. S. Lewis, The Four Loves

The research on this page is a long footnote to that sentence. The “you too” requires a first person to say the thing, and the loneliness that makes the saying feel risky is the same loneliness on the other side of the table, waiting to be recognized. In a conversation about that line in August, the summary was three words:

“We’re just too lonely. That’s my anxiety.”

In conversation, August 7, 2026.

The anxiety is correct about the room and wrong about what to do in it. The corrected version, for a first year at Booth, is short. The feeling is the signal working. Most of the section has it. Every estimate of how the others will respond runs low by a measurable margin. The disclosure that feels like exposure is the only instrument that fixes the estimate, and the study that proved it was run here. Under the data, going first is less the brave option than the accurate one.

Related on this site: On Care, on the prosociality paradox; On Mattering, on being witnessed; On Leaving Ego at the Door, on the beautiful mess effect; Outlast, on ostracism and the four threatened needs.

Facts verified September 23, 2026. Jenn’s lines are quoted verbatim from her own recordings, a voice memo, and pages already published here, and are dated where they appear; the connective text is exposition about those lines and the research, not new first-person writing. Classmates are not named.