It was 1999 in Palo Alto, and the air smelled of money.

Not literally—that would have been vulgar, and Elliot Vance had never been vulgar. He was thirty-one years old, a graduate of Stanford's computer science program, and the founder of a company called Interval Systems, which had raised twelve million dollars in Series A funding before it had a product, before it had customers, before it had anything except a theoretical proposition about how data could be organized. The proposition was this: between every two pieces of information, there existed a space—a vector of possible meanings—and if you could map that space, you could predict what people wanted before they knew they wanted it.

Elliot did not know if the proposition was true. He knew it was compelling. He knew it had convinced a dozen venture capitalists to write checks. He knew it had attracted engineers from MIT and Caltech and one defector from Netscape who had been willing to take a forty percent pay cut for stock options and the promise of changing the world.

The company occupied the top floor of a three-story building on University Avenue, with exposed brick walls and Aeron chairs and a kitchen that stocked San Pellegrino in quantities that would have embarrassed a small European nation. Elliot's office had a window that faced the hills, and from that window he could see the future: a landscape of glass and steel and fiber-optic cable, stretching to the horizon, waiting to be filled with data.

The problem was the gap.

Not the gap in the technology—the engineering was straightforward, if ambitious. The gap was between what Interval Systems claimed to be doing and what it was actually doing. The pitch deck told investors that they were building a new kind of search, one that understood context and intent and the invisible connections between ideas. The reality was that they were building a recommendation engine for a niche e-commerce platform that sold artisanal kitchen supplies to wealthy people in Marin County. It was not world-changing. It was not even interesting. But it paid the bills while they tried to figure out how to build the thing they had promised.

---

Elliot's co-founder was a woman named Priya Saxena, who had been his teaching assistant in graduate school and was now the chief technology officer of Interval Systems. She was the one who actually understood how the technology worked. She was the one who had figured out that the vector space between two data points could be interpolated, navigated, exploited. She was the one who had built the prototype that had convinced the investors.

She was also the one who saw the gap most clearly.

"We are not going to make it," she said one afternoon, standing in the doorway of Elliot's office with a printout of the latest engagement metrics. "The recommendation engine is plateauing. The e-commerce client is considering switching to a competitor. And the Series B pitch is three weeks away."

"We will figure it out," Elliot said.

"No, we will not. We are pretending to be one thing while building another. The investors do not know the difference yet, but they will. And when they do, the valuation collapses."

"Then we need to build the real thing before they find out."

"That is the problem," Priya said. "The real thing requires a breakthrough we have not made. The vector space exists, theoretically. But we cannot navigate it without a cost function that makes sense, and I have been working on that for eighteen months and I am no closer than I was at the beginning."

Elliot looked at her. She was tired. They were all tired. The company had been running on caffeine and conviction for two years, and the conviction was starting to wear thin.

"What do you need?"

"Time. Money. Both."

"How much?"

"Six months. Another five million."

"We have three weeks and zero committed Series B."

"Then we are dead."

---

The phrase echoed in Elliot's head for the rest of the week. He lay awake at night thinking about it: the mathematical precision of Priya's assessment, the way she had reduced their two-year effort to a simple equation with a terminal answer. Dead. It was not hyperbole. It was a prediction.

He started doing calculations of his own. Burn rate: one hundred and forty thousand dollars per month, including salaries, rent, the lease on the BMW, and the San Pellegrino. Current runway: two months, assuming no new revenue. Probability of closing Series B: near zero after the disastrous pitch. Probability of acquiring enough new clients to reach profitability in sixty days: indistinguishable from zero.

The numbers did not lie. They could not be interpolated, finessed, or managed away. They were the hard edge of reality against which all his theoretical ambitions were measured and found wanting.

He began calling his network. Not for funding—that door had closed. For advice, for introductions, for the kind of small favors that kept failing companies alive long enough to fail properly. He called a former classmate who had sold his company to Microsoft. He called a professor who had mentored him at Stanford. He called a journalist who had written a favorable profile of Interval Systems the year before and who might—if Elliot was very careful about how he framed the conversation—write another one.

None of it worked. The classmate was polite but noncommittal. The professor was sympathetic but offered nothing more than encouragement. The journalist had moved on to covering telecommunications and did not return Elliot's call.

He was standing at the window of his office, looking at the hills, when he understood something that Priya's model had not accounted for. The interpolation between two points was not limited to mathematics. It was a human process as well. The gap between what you are and what you claim to be is navigated not by algorithms but by relationships, by reputation, by the willingness of other people to believe in the vector you are traversing.

And his vector, he realized, was running out of believers.

Elliot walked home that night through the streets of Palo Alto, past the startups and the venture firms and the coffee shops where twenty-two-year-olds were sketching business plans on napkins that would be worth millions by morning. The gap between what they were and what they claimed to be was not unique to Interval Systems. It was the defining feature of the industry, of the era, of the city itself. Everyone was interpolating between a real present and an imagined future, hoping the vector would resolve before the distance became too great to cross.

He thought about the mathematical model that Priya had built. The vector space between two points is infinite—an uncountable continuum of possible trajectories. The function that chooses one path over another is called a loss function, or a cost function, or an optimization goal. In the abstract, it was a question of mathematics. In practice, it was a question of faith.

What do you optimize for? What cost are you willing to pay? What gap is worth crossing?

He did not have an answer. But as he walked past a shuttered hardware store on Emerson Street, he saw a reflection in the glass that made him stop. It was his own face, pale and tired, floating in the dark window like a ghost. Behind the reflection, he could see the dim shapes of shelves and counters, abandoned when the store had closed the year before, replaced by an internet startup that sold organic dog food.

The gap between the hardware store and the pet food startup was not a gap of time. It was a gap of intention. The hardware store had existed to serve a need that was real, measurable, concrete. The startup existed to serve a need that was imagined, projected, desired.

And Interval Systems, he realized, was exactly the same.

---

The Series B pitch was a disaster.

Elliot stood in a conference room on Sand Hill Road, presenting to a room full of partners who had seen a thousand pitches just like this one. He talked about the vector space. He talked about the interpolation algorithm. He talked about the future, which was always easier to describe than the present.

One of the partners, a woman in her fifties with gray hair and eyes that had not believed a single word he had said, leaned forward and asked the question he had been dreading.

"Mr. Vance, what exactly does your product do?"

"It navigates the latent space between data points to generate recommendations that are contextually aware and semantically meaningful."

"That is not an answer. That is a description of a research project. I am asking about the product. The thing that generates revenue. The thing that you sell."

Elliot hesitated. He looked at Priya, who looked away. He looked at the investors, who were waiting.

"We have a recommendation engine for a kitchen supply e-commerce platform."

"How much revenue does it generate?"

"Not enough."

"Then you are not a company. You are a laboratory. And we do not invest in laboratories."

The meeting ended twelve minutes after it began. Elliot and Priya walked out into the parking lot and stood beside their car, a leased BMW that Elliot could no longer afford.

"Now what?" Priya asked.

"Now we figure out how to make the recommendation engine good enough to survive."

"That is not the vision."

"The vision is a luxury of the funded."

---

The thing about interpolating between two points is that you cannot skip the intermediate steps. You cannot jump from the present to the future without crossing the territory in between. And the territory in between, for Interval Systems, was eighteen months of grinding, incremental work: optimizing the recommendation engine, retaining the e-commerce client, finding new clients who would pay for marginal improvements rather than revolutionary breakthroughs.

Priya quit after six months. She had been offered a research position at a university in Zurich, where she could work on the vector space theory without the pressure of quarterly earnings. Elliot did not blame her. He would have done the same, if he had had somewhere else to go.

The company survived. It did not thrive, but it survived. The recommendation engine got better. The clients grew from one to seven. The revenue grew from negligible to almost-breakeven. And in the third year, something unexpected happened: Priya published a paper from Zurich that solved the cost-function problem. The vector space could be navigated. The interpolation was possible. The vision that had been a promise was suddenly within reach.

Elliot read the paper in his office, alone, with the window facing the hills that had not yet been built on. He had spent three years crossing the gap between a failed pitch and a viable theory. He had spent three years believing in something that did not yet exist, and then building it until it did.

He called Priya in Zurich.

"Congratulations," he said.

"Congratulations to you," she said. "You kept it alive long enough for me to finish it."

"We both did."

"No," Priya said. "You did the hard part. The theory was the easy part. The hard part was staying in the gap, not knowing if the other side existed, and crossing it anyway."

Elliot hung up the telephone. He looked at the paper, at the equations that had once been impossibilities and were now solutions. He thought about the interval between two points, the space that contained every possible path and every possible outcome, and he understood that the crossing itself was the only thing that mattered. Not the starting point. Not the destination. The crossing.

The entire Internet economy was built on that crossing. Every startup, every pitch, every round of funding was an act of interpolation, a bet that the vector could be navigated before the resources ran out. Most of them failed. Some of them, like Interval Systems, found the other side just in time.

On his way out of the office that night, Elliot passed the conference room where they had held their first all-hands meeting, three years and a lifetime ago. The whiteboard still had the notes from that meeting, the diagrams of the vector space that had seemed so revolutionary and were now merely obvious. The future had arrived. The gap had been crossed. And the only thing left was to find the next one.