Daniel Park was thirty-one years old in the summer of 1999, and he believed with the force of revelation that the internet could make people less lonely. This belief lived inside him like an auxiliary heartbeat, something he had carried since his undergraduate years at Stanford, where he had watched brilliant classmates surrounded by brilliant classmates and yet each one seemed to exist inside a separate jar of glass, visible to everyone but touchable by no one. He had coded the first version of Near during the winter of 1997 in the garage of a rental house on Emerson Street in Palo Alto, drinking Mountain Dew at three in the morning while a Netscape Navigator window glowed phosphorescent on his Apple PowerBook, and the code had been simple then, almost naive: a web form where people could describe the kind of connection they were looking for, a database that matched keywords and geographic proximity, an email notification system that said someone nearby also liked the novels of Haruki Murakami or had also lost a parent to cancer or was also walking through the same species of loneliness at the same hour of the same night. That was all it was. That was everything it needed to be.

Position zero point zero on the interpolation: pure intention, unmediated by capital, unexamined by the machinery of the market. Daniel had forty-three dollars in his Wells Fargo checking account and a Visa card that was about to be declined and he did not care, because the code was beautiful and the idea was true and people were writing to him — actual strangers, actual humans, sending emails to the contact address at the bottom of the page — saying things like "I met someone on your site who understands what it is like to lose a brother" and "I have been lonely for seven years and last night I had coffee with a person who made me laugh for the first time since my divorce." These emails were the only compensation Daniel needed. He printed them out and taped them to the wall above his monitor, and when the code refused to compile or the server crashed or he wondered whether the entire enterprise was a foolish waste of a Stanford degree in computer science, he looked at the emails and remembered why he had started.

Position zero point fifteen: the first angel investor. Her name was Margot Kessler, and she was fifty-three years old, and she had made her first fortune in enterprise database software in the late 1980s and was now looking for internet plays because everyone on Sand Hill Road was looking for internet plays. She sat across from Daniel at a coffee shop on University Avenue — Caffe del Doge, which had opened the previous year and was already a Palo Alto institution — and she listened to his pitch for eighteen minutes without interrupting, and then she said, "You have something here, Daniel, but you need to think about scale. You need to think about how this becomes a hundred thousand users, a million users, ten million users. You need to think about the business."

Daniel nodded because he did not know what else to do. Scale, in his understanding, meant more people finding each other, which was exactly what he wanted, so scale was good, scale was aligned with the mission, scale was the natural extension of everything he believed. He did not yet understand that the word "scale" in the mouth of an investor was a different word — a word that meant growth curves and revenue models and eventual exits — and that this word, once introduced into the conversation, would begin to colonize every other word in his vocabulary.

Margot Kessler wrote a check for two hundred and fifty thousand dollars. Daniel used it to rent an office — a converted warehouse near the Caltrain tracks on California Avenue, where the trains rumbled past every seventeen minutes and the vibration felt like progress — and to hire his first three employees: a designer named Rachel who had worked at Apple and understood that interfaces were emotional experiences before they were functional ones, a backend developer named Marcus who had dropped out of the Stanford PhD program to join a startup that had already failed, and a community manager named Lila whose job was to read every message posted on Near and make sure nobody was being harmed. They were a family. They ordered pizza at midnight and argued about typefaces and believed, all four of them, that they were building something that would change the way humans connected to each other.

Position zero point thirty: the redesign. Rachel spent six weeks rebuilding the Near interface from scratch. She chose warm colors — amber and terracotta and a particular shade of blue that she called "trust blue" — and generous whitespace that made the site feel like a room with high ceilings, and a registration flow that asked new users "What kind of connection are you looking for?" instead of asking for their email address first. She added a feature called "Letters" — a space where users could write long-form messages to no one in particular, messages about grief and joy and the ordinary texture of daily life, and other users could respond, not with comments but with their own letters, creating threads of correspondence that stretched across weeks and months and sometimes years. The Letters feature became the soul of Near. People wrote things on Near that they had never told anyone — about the miscarriage they couldn't discuss with their husband, about the job they hated and the fear that they had wasted their life, about the specific way the light fell through the window of their childhood bedroom and how they missed it with an ache that felt like homesickness for a place that no longer existed. And other people responded with tenderness and recognition, and the connections that formed were real, were human, were exactly what Daniel had imagined when he wrote the first lines of code in the garage on Emerson Street.

Position zero point fifty: Series A. The users had grown to forty thousand, not through advertising but through word of mouth, through people telling other people about this strange and gentle corner of the internet where you could be honest and be heard. The venture capitalists on Sand Hill Road had noticed — not the honesty, not the tenderness, but the growth curve, the engagement metrics, the fact that users spent an average of forty-seven minutes per session on Near, which was extraordinary for a text-based platform in the age of flashy multimedia portals like Yahoo and GeoCities. The lead investor for the Series A was a man named Gerald Fletcher, a partner at one of the top Valley firms, a man who wore Patagonia fleece vests and had funded three of the biggest IPOs of the decade and who leaned back in his chair in a conference room with floor-to-ceiling windows overlooking the 280 freeway and said, "The product is a love letter to human connection, Daniel. Genuinely. I have never seen engagement like this. But what I need to understand is: what is the business model?"

Daniel said, "We were thinking subscription. Five dollars a month for premium features. Unlimited Letters, maybe, or the ability to form private groups."

Gerald Fletcher smiled the smile of a man who had heard this exact sentence from two hundred founders and had watched one hundred and ninety-seven of them fail. "Subscription caps your growth. The internet rewards free. The internet rewards network effects. You need to be everywhere, and you need to be free, and you need to monetize the data."

The data. The word landed in Daniel's stomach like a stone dropped into still water. The data meant the patterns of lonely people searching for connection. The data meant knowing who was desperate enough to click on what kind of advertisement. The data meant mapping the geography of human vulnerability and selling the map to the highest bidder. But Gerald Fletcher was offering eight million dollars at a valuation that would make the early employees rich and would fund the company for two years and would allow Near to reach the millions of users that Daniel genuinely believed needed it, and the alternative was running out of money in seven weeks and laying off Rachel and Marcus and Lila and everyone else who had believed in the vision, and so Daniel said yes, and the interpolation advanced, and he told himself that the data could be anonymized, that the advertising could be tasteful, that the mission could survive contact with the market, and he almost believed it.

Position zero point sixty-five: the growth imperative. Near now had two hundred and eighty thousand users, and the office had moved to a larger space on Page Mill Road with exposed brick walls and an espresso machine that cost more than Daniel's first car, and the team had grown to thirty-four people, and the board meetings had become a different species of conversation. Nobody asked about the Letters feature anymore. Nobody asked about the emails from users who had found best friends or support groups or the courage to leave abusive relationships. The questions were about monthly active users and daily active users and the DAU-to-MAU ratio and whether that ratio was sticky enough to justify the Series B valuation. The questions were about churn and cohort analysis and lifetime value. The questions were about monetization — banner ads versus sponsored content versus a freemium model that would nudge users toward a paid tier by subtly degrading the free experience. Daniel sat in these meetings in a Herman Miller chair that cost two thousand dollars and felt himself becoming translucent, felt his own intentions and convictions being replaced pixel by pixel by the intentions and convictions of the board deck, the growth projections, the term sheet that Gerald Fletcher had emailed that morning with the subject line "Final — please sign."

He stopped sleeping well. He would lie awake in his apartment — a nice apartment now, thirty-two hundred dollars a month, overlooking the Baylands preserve — and try to remember what the code had felt like at three in the morning in the garage, the specific quality of the silence, the way the Mountain Dew tasted after it had gone flat, the certainty he had felt that he was building something that mattered for reasons that could not be captured in a spreadsheet. The memory was still there but it was distant, like a photograph of a photograph, degraded with each act of reproduction.

Position zero point eighty: the feature that undid him. The product team, now eighteen people and led by a former Amazon product manager named Todd who used the word "optimize" the way other people used punctuation, had been running A/B tests on the Letters feature, and they had discovered something that should not have surprised anyone who understood human nature: engagement spiked when the algorithm surfaced content that made users feel strong emotions, and the strongest emotion, the stickiest emotion, the emotion that kept people scrolling and clicking and refreshing, was anger. Arguments in the Letters threads generated thirty-four percent more page views than supportive exchanges. Heated debates about politics and religion kept users online for an average of eleven additional minutes per session. The algorithm, left to its own devices and optimized purely for engagement, was learning to amplify conflict because conflict was profitable, because conflict drove the metrics that drove the valuation that drove the Series B that Gerald Fletcher was already lining up.

Daniel stood in the all-hands meeting on a Thursday afternoon in September while Todd presented these findings on a slide deck titled "Engagement Optimization Q3 Results," and he watched his team nod along, taking notes on their Palm Pilots, asking smart questions about implementation timelines and statistical significance and whether the algorithm should prioritize anger over other emotions or whether a portfolio approach would be more robust. Nobody asked what happened to the users on the receiving end of the anger. Nobody asked whether the platform was still serving the purpose for which it had been built. Nobody asked whether "engagement" was the right metric or merely the most measurable one. Daniel stood at the back of the room with his arms crossed and his mouth closed and felt himself dissolving, felt the last particles of his original intention being scattered by the wind of market logic, and he said nothing, because the company was now too large and too well-funded and too committed to its trajectory for one person's doubt to alter its course, even if that person was the founder, even if that person had written the first line of code with nothing but idealism and Mountain Dew.

He went home that night and opened his laptop and navigated to Near, to the Letters section, and he read for three hours. The letters were different now. The tenderness was still there, buried under layers of algorithmic recommendation, but the prominent letters — the ones the system pushed to the top of the feed — were angry letters, polemical letters, letters designed to provoke rather than to connect. A woman named Susan had written a gentle letter about her mother's dementia, and it had received seven responses. A man who went by the username FreeThinker88 had written a letter about why feminism was destroying Western civilization, and it had received two hundred and forty-seven responses, most of them furious, all of them driving engagement metrics upward and to the right.

Daniel read Susan's letter three times. It was beautiful. It was exactly the kind of letter that had made Near worth building. And the algorithm had buried it because beauty did not optimize, tenderness did not scale, human connection could not be A/B tested into existence.

Position zero point ninety-five: the acquisition offer. A major media conglomerate — one of the old-line companies that was desperately acquiring internet properties to prove to its shareholders that it understood the digital future — offered to buy Near for two hundred and forty million dollars. The board was ecstatic. Gerald Fletcher called Daniel on his cell phone, a Nokia that was slightly smaller than a brick, and said, "This is it, Daniel. This is what we were building toward. This is the exit." And Daniel, sitting in his apartment with the lights off and the Baylands dark beyond the window, understood with a clarity that felt like physical impact that he had become exactly the thing he had set out to fix. He had built a platform for human connection and had, through a thousand reasonable decisions, each one small and defensible and inevitable, transformed it into a machine for monetizing loneliness and amplifying conflict, and now he was selling that machine to a conglomerate that would strip it for parts and fold the users into a portfolio of properties and lay off everyone who had ever believed in the mission. And he would be rich. And nobody would ever know his name except as a line item in a press release.

Two weeks before the acquisition closed, Daniel drove to Emerson Street. The rental house had been repainted in a color that the real estate agent would call "sand" and a family with young children lived there now, and the garage — his garage, the birthplace of Near — had been converted into a playroom with foam mats on the floor and finger paintings taped to the walls. He parked across the street in his new car, a silver BMW that he had bought with the Series A money and had never particularly wanted, and he sat in the driver's seat with the engine off and tried to feel something. The garage was right there, forty feet away, but it belonged to someone else now, and the code he had written inside it had been rewritten so many times that not a single line of his original work remained on the production servers, and the users were angry and the board was happy and Daniel Park, the person, the human being who had once believed that the internet could make people less lonely, had become invisible inside his own creation.

He was the company. He had no identity apart from the entity. When people said "Near," they meant the platform, the stock price, the user experience, the product roadmap, the exit strategy — they did not mean Daniel. The young man who had coded through the night with idealism and flat soda had been interpolated away, replaced frame by frame by the version of himself that the market had demanded, and the process had been so gradual, so incremental, that he could not identify the moment when the substitution had become complete.

The acquisition closed on a Thursday in November. Daniel's net worth increased by forty-four million dollars. He went home to his apartment, sat on the couch that had cost more than the first year of Near's server expenses, and opened his laptop. He went to the Letters section of Near — for the last time, he knew, because the conglomerate would be replacing it with a content management system designed for engagement and monetization — and he read through the old letters, the ones from the early days, the ones the algorithm had long since buried. There was a letter from a woman named Margaret, posted in March of 1998, that he had read before and had never forgotten. "I met my best friend on this site," Margaret had written. "We have never met in person. She lives in Oregon and I live in Florida. But she knows things about me that my husband does not know, and I know things about her that her mother does not know, and I think this is what the internet was supposed to be. Thank you for building this."

Daniel read Margaret's letter and then he opened a new document, and he began to write a reply — not as the CEO, not as the founder, but as himself, as the person he had been before the interpolation began. He wrote about the garage on Emerson Street and the first emails from users and the moment when he had said yes to Gerald Fletcher and all the moments after. He wrote about how each decision had seemed small and reasonable and inevitable and how the sum of reasonable decisions had produced something unreasonable and unrecognizable. He wrote about Rachel and Marcus and Lila and the pizza at midnight and the arguments about typefaces, and about the trust-blue buttons on the registration page, and about Susan's letter about her mother's dementia that seven people had read while two hundred and forty-seven people had argued about feminism.

He wrote until the sun came up over the Bay, the light spreading across the water in bands of pink and gold, and then he closed the laptop without posting any of it. The letter he had written was addressed to a version of Near that no longer existed, meant for a version of himself that had been written over so many times that the original was unrecoverable. The interpolation was complete. There was no way to go backward through the latent space, no way to reverse the gradient that had pulled him from idealism to exit, no way to return from position one point zero to position zero. He was the endpoint now. He had become the distance between two truths — the truth of human connection and the truth of market logic — and he lived there, suspended in the space between what he had wanted to build and what he had actually sold, invisible to everyone including himself.

He did not take the severance package or the advisory role that the conglomerate offered. He did not start another company. He did not invest in startups or join a venture firm or do any of the things that successful founders were supposed to do after an exit. He moved to a small town in coastal Oregon, to a house that looked out over the Pacific, and he lived there quietly for a long time, and the only trace of Near that remained in his life was a single printed email taped to the wall above his desk: Margaret's letter, from March of 1998, the proof that once, for a brief period, before the interpolation began, he had built something that actually connected two human beings across the distance between Florida and Oregon and loneliness and recognition, and this was not nothing, even if it had not lasted, even if the market had consumed it, even if the distance between what we build and what we sell is always, in the end, unbridgeable.