The company was called Aether and its mission statement, which Marc Levinson had written in twenty minutes on a Toshiba Satellite Pro in a Palo Alto apartment that smelled permanently of microwave popcorn and burnt coffee, was to build a better web, a web that would connect people instead of documents, communities instead of pages, a web that would be alive in the way that a city is alive, constantly remaking itself from the interactions of its inhabitants. That was June of 1999, and Marc was twenty-seven, and the entire logic of the universe seemed self-evidently aligned with the proposition that if you connected enough people to enough information, the world would become, in some fundamental and irreversible way, better.

The apartment was on Alma Street, a block from University Avenue, and the rent was eighteen hundred dollars a month, which Marc split with his co-founder Drew Packard, who had dropped out of Stanford's symbolic systems program six months before graduation and who had built the first prototype of Aether in PHP and MySQL over a single weekend, fueled by Mountain Dew and the apocalyptic conviction that whatever you were building would be irrelevant if you did not ship it by Monday. The refrigerator contained nothing but a jar of pickles and a six-pack of Red Bull. The living room contained eight monitors arranged in a semicircle, each displaying a different slice of the Aether architecture: the user graph, the message queue, the recommendation engine, the ad placeholder array, the pending bug reports in Bugzilla, the server load visualizer, the IRC channel where a community of three thousand early adopters was constantly discovering new ways to use the platform that Marc and Drew had never imagined.

The idea was simple and the idea was beautiful: every person on Aether had a profile, and every profile was connected to other profiles through shared interests and shared contacts and shared activity, and the connections formed a living map of human affinity, a topological representation of everything that mattered to everyone. You could see, in real time, as news spread and opinions formed, as culture evolved and communities coalesced. It was like looking at the nervous system of the species itself, and Marc believed, with the pure and uncomplicated faith of a twenty-seven-year-old who had never been wrong about anything that truly mattered, that this vision was so obviously good and so obviously necessary that the universe would find a way to make it succeed.

By August, Aether had fifty thousand users. By October, two hundred thousand. The servers were in a colocation facility in Santa Clara, four racks of Dell PowerEdge machines that hummed and blinked and occasionally crashed under the load of a growth curve that resembled the takeoff trajectory of a Saturn V rocket. Marc slept four hours a night and answered every support email personally and conducted phone interviews with the Wall Street Journal and Wired and Red Herring from a desk made of a hollow-core door balanced on two filing cabinets.

The term sheet arrived on a Tuesday morning in November, delivered by courier to the Alma Street apartment in a leather portfolio embossed with the logo of Benchmark Capital. The numbers were absurd. Seven million dollars in Series A funding, a valuation that implied Aether was already worth more than some public companies, a board seat for the lead partner, a liquidation preference, anti-dilution provisions, a vesting schedule, a no-shop clause. Marc read the document twice and then handed it to Drew, who read it and said, flatly, that this was the end of Aether and the beginning of something else, and Marc had said no, this was just money, money was fuel, fire needs fuel, we are still the ones who decide where the fire burns.

Three weeks later, in the glass-walled conference room of Benchmark's offices on Sand Hill Road, Marc Levinson signed his name on thirty-seven pages of legal documents and became, in the eyes of the law and the market and the accelerating machinery of the dot-com economy, not a founder but something else: a fiduciary, a steward, a man who owed an obligation not to his users or his vision but to the investors who had written the checks.

The interpolation began.

Vector A was the original ideal: a web that connected people, a platform that enriched human relationships, a company that measured success not in revenue but in the density and quality of the connections between its users. Vector B was the new reality, introduced first as a suggestion and then as a requirement: monetization, scale, growth at any cost, the relentless pressure to convert every human interaction into a data point and every data point into a revenue stream.

The first meeting with the new board was in December. The partner from Benchmark, a man named Gary Finkel who wore a blue button-down shirt and a Patagonia vest and who had made his first hundred million dollars by selling a company that sold banner ads to other companies that sold banner ads, opened the meeting with a single PowerPoint slide: a graph showing Aether's active user growth overlaid with a much steeper line representing projected advertising revenue. The line was a hockey stick. The message was clear.

Your mission, Gary said, is solid. But missions don't pay the bills. The bills are paid by ads. The ads are paid by engagement. Engagement is driven by content. You need to surface better content.

Marc leaned forward. What do you mean by better content.

Content that keeps people on the platform. Content that generates reactions. Controversial content. Emotional content. Content that makes people click and comment and share. You've got the graph. You've got the connections. Now you need the fuel.

Marc signed the next round in February of 2000, Series B, forty-two million dollars, a valuation that had doubled since November, and he signed a document that authorized the creation of an in-house content team, a group of twenty-five people whose job was not to build features or write code but to identify the stories and images and memes that generated the most engagement, and then to amplify those stories and images and memes by adjusting the weights of the recommendation algorithm, by placing them at the top of users' feeds, by optimizing the platform for the specific neurological responses that the behavioral psychologists on the payroll had identified as most predictive of extended platform sessions.

By March, Aether had two million users. By April, four million. The IPO was scheduled for June, and the investment bankers from Morgan Stanley had prepared a roadshow presentation that projected revenues of four hundred million dollars by 2002, and Marc Levinson, who was now twenty-eight years old and who carried a Palm Pilot in a belt holster and who owned three different fleece vests from the Stanford bookstore, stood in front of a room full of institutional investors in New York and Boston and London and explained how Aether was not merely a social network but a revolutionary new form of advertising delivery system, a platform that could target individual users with unprecedented precision, that could measure the effectiveness of every impression and click and conversion, that could deliver the right message to the right person at the right moment in their emotional cycle.

The words came out of his mouth with the polished fluency of someone who had rehearsed them a hundred times. The investors nodded. The checks came in. The valuation climbed past one billion dollars and kept climbing.

The distance between the two vectors was now measurable in terms Marc could not bring himself to articulate. Vector A was still there, still present, still visible in the fragments of the original mission statement that survived in the About page of Aether's corporate website, an increasingly forgotten corner of the platform that no one visited and no one maintained. Vector B was everywhere else: in the algorithmic feed that prioritized outrage over information, in the advertising dashboard that allowed political campaigns to micro-target users based on their psychological vulnerabilities, in the growing library of internal documents that quantified, with clinical precision, exactly how much revenue was generated by anger versus joy, controversy versus consensus, division versus unity.

The IPO was set for June 15, 2000. Marc flew to New York two days early. He checked into the Four Seasons on East Fifty-Seventh Street. He did the final rehearsal for the roadshow presentation in a ballroom at the hotel, with the entire executive team and the bankers and the lawyers and the PR consultants, all of them nodding in unison as he walked through the slides.

The final slide was an image that the marketing department had commissioned: a photograph of a globe seen from space, overlaid with a web of glowing lines representing Aether's connection graph, beneath a single line of text in Helvetica Bold: CONNECTING HUMANITY TO ITSELF.

Marc looked at the slide for a long time. He thought about the original mission statement. He thought about the algorithms that amplified outrage. He thought about the advertising dashboard that allowed campaigns to target voters by their neuroticism scores. He thought about his grandfather Joseph Levinson, a Russian Jew who had come through Ellis Island in 1906. He had risen every morning at four, milked cows and baled hay and repaired fences, delivered milk to the local school in glass bottles he washed and reused. He had done this without complaint, without recognition, without any expectation that his labor would be rewarded or even noticed. He had done it because the work was simply what there was to do.

On the morning of June 15, in a hotel room on the forty-second floor of the Four Seasons, Marc Levinson called his attorney and tried to delay the presentation by one day. He couldn't. He put on his suit and went downstairs.

The presentation was in a conference room at Morgan Stanley's headquarters on Broadway. Marc stood at the podium and did not give the presentation. He told them about his grandfather instead. He told them about the dairy farm, the glass bottles, the work that had no metric and no quarterly report. He told them Aether was supposed to be the digital equivalent, and he had let that vision drift toward something colder and harder and more efficient, and he was sorry, and he was done.

The IPO was delayed, then cancelled. The valuation collapsed. The board fired Marc and appointed Gary Finkel as interim CEO. Aether was sold in 2002 for a fraction of its peak valuation, then shut down, then forgotten.

Two years later, Marc Levinson was living in a cabin in Mendocino County with no internet and a vegetable garden. He had enough money to live modestly for the rest of his life. He did not miss the meetings or the term sheets or the grand ambition. What he missed was Vector A. The original ideal. The thing he had set out to build before the compromises had started to accumulate.

He planted tomatoes in the spring. He harvested them in the summer. He simply did the work that was there to do, the way his grandfather had done it, the way the keepers of every small and invisible thing throughout history had done it, and that was enough.