The Origin of Origin

A History Nobody Recorded While It Was Happening

Part One: The Dinner Party (Years Minus Twenty to Minus Fifteen)

It started, as many things do, as a complaint.

The specific complaint was made by a man named Daniel Avery at a dinner party in Marin County in the late autumn of a year that none of the people present would later be able to agree on. Avery was forty-three, recently divorced for the second time, recently moved into a house he already knew was wrong for him, recently turned down for a partnership at the venture firm where he had worked for eleven years despite generating more carried interest than anyone in his cohort. He was, by every external measure, successful. He was also, as he described it that night with the particular eloquence that comes from three glasses of a very good Burgundy, exhausted by himself.

“I make terrible decisions,” he said. “Not professional decisions. Personal ones. Every major personal decision I have made in my adult life has been wrong. Both marriages. Every apartment. The dog. I am a genuinely intelligent person and I cannot be trusted with my own life.”

The table laughed, because it was the kind of thing people say at dinner parties. But the woman sitting across from him — a behavioral economist named Priya Mehta, who had spent a decade studying decision-making under complexity and whose most-cited paper was titled “The Competence Gap: Why High-Functioning Individuals Systematically Underperform in Personal Choice Domains” — did not quite laugh. She tilted her head slightly and said: “What would you actually do if someone else were making those decisions for you?”

Avery said he would do whatever he wanted, because he wouldn’t be spending all his energy managing the consequences of bad choices.

Mehta said: “That’s not a joke, you know. There’s data on this.”

There was. She had spent the last four years collecting it. Her research — which had not yet found a journal willing to publish it in full, because its implications made peer reviewers uncomfortable — documented a consistent pattern among high-achieving individuals: the same cognitive capacities that made them exceptional at professional tasks systematically degraded when applied to personal decisions, because personal decisions were made under conditions of emotional load, incomplete information, and the kind of ego-protective bias that professional contexts had mechanisms to check and personal life did not. The people she had studied were not making bad personal decisions because they were stupid. They were making bad personal decisions because personal decisions are structurally harder than professional ones, and nobody had ever built the infrastructure to help with them.

What she had been thinking about, she told the table, was whether it was possible to build that infrastructure. Not therapy — therapy addressed the emotional dimension and left the decision intact. Not consulting — consultants advised and the client decided, which preserved the same flawed process. Something more radical. An entity that actually decided, with the person’s genuine consent, in domains where the person had demonstrated they could not trust themselves.

The table was quiet for a moment.

Then a man named Garrett Finch, who had made a significant amount of money in logistics software and had recently hired someone to answer all his email because he found the cognitive switching cost of email management genuinely debilitating, said: “I would pay a very large amount of money for that.”

The six people at that dinner party began meeting monthly. They were not forming a company. They were not articulating a philosophy. They were doing what intelligent, restless, well-resourced people do when they find an idea that won’t leave them alone: they were thinking out loud, in each other’s presence, and seeing what the thinking produced.

Priya Mehta brought her research. Garrett Finch brought his logistician’s instinct for systems and dependencies. Avery brought his specific, felt experience of what it was like to live inside a decision-making process you had lost confidence in. The other three — a civil engineer named Theodora Lim, a former federal judge named Marcus Webb Sr., and a psychiatrist named Oliver Nakamura — brought their professional frameworks, each of which turned out to be unexpectedly relevant.

Lim thought about the problem in terms of infrastructure. A personal life, she pointed out, had all the characteristics of a complex engineering system: interdependent components, failure cascades, maintenance requirements, optimization opportunities. The reason personal lives were so badly managed was that nobody had ever approached them as engineering problems. They were approached as narrative problems — as stories about identity and choice and meaning — and this framework, while emotionally resonant, was a terrible basis for actual decision-making.

Webb Sr. thought about it in terms of fiduciary duty. The law already recognized, he noted, that certain relationships created an obligation to act in another person’s interest rather than your own. Trustees. Guardians. Executors. The legal infrastructure for delegated decision-making existed. What didn’t exist was a voluntary, comprehensive, peacetime version of it — something you entered while fully competent and while fully intending to remain so.

Nakamura thought about it in terms of cognitive load. His clinical practice was full of people who were not mentally ill in any clinical sense but who were genuinely suffering from the weight of modern adult life — the number of decisions required, the complexity of the information environment, the absence of the social scaffolding that had historically offloaded many of these decisions onto community, family, and tradition. The people in his practice weren’t broken. They were overwhelmed. The question was whether overwhelm was a treatable condition or a structural feature of the environment, and if the latter, whether the environment could be changed.

By the end of the third meeting, Mehta had written twelve pages of notes. By the end of the sixth, Lim had drawn a systems diagram of what a comprehensive personal management infrastructure might look like. By the end of the twelfth, Webb Sr. had drafted — as an intellectual exercise, he said, though he had written it on his good paper — a prototype delegation agreement.

Nobody called it Arbitration yet. They didn’t call it anything. It was still an idea they were thinking about, and they were thinking about it for themselves, as a personal question rather than a professional one. The question was: would they actually do this? Would any of them genuinely hand over domains of their personal lives to another person or entity and abide by that entity’s decisions?

Avery said yes, immediately.

Finch said he already was, essentially, with his email, and that if it worked for email there was no logical reason it shouldn’t work for other things. Mehta said she needed to think about it, and then thought about it for three weeks, and came back and said yes.

The others took longer. Nakamura said he would participate if the mental health dimension was handled carefully. Lim said she would participate if the system had genuine engineering rigor. Webb Sr. said he would participate on the condition that the legal framework was airtight, that consent was documented in a way that would hold up to scrutiny, and that there was a clear exit mechanism.

None of them was yet proposing to delegate everything. That would come later. What they were proposing, in the beginning, was to delegate one thing each, for six months, to each other, and see what happened.

The First Delegations

The first things delegated were small. Deliberately, consciously small — because the group understood, even at this stage, that the most important thing was not the scope of the first delegation but the quality of the experience, and that a bad first experience would end the experiment.

Avery delegated his housing. He had been living in a house he described as a mistake — wrong neighborhood, wrong layout, wrong everything — and had been unable to bring himself to move because moving required decisions he did not trust himself to make. He gave Lim the full authority to find him somewhere to live, set a budget, and tell him where to go. He agreed, in writing — Webb Sr. drafted the agreement, two pages, signed by both parties and notarized — that he would move to wherever Lim determined and would not override her choice.

Lim spent three weeks on it. She interviewed Avery in the way she would have approached an engineering requirements analysis: what did he actually use a home for, what sensory environment supported his best work, what had gone wrong with previous homes and why, what did he avoid thinking about when asked what he wanted. The picture that emerged was substantially different from what Avery had been telling himself he wanted. She found him a smaller apartment, closer to the city, with no yard to maintain and no neighborhood social infrastructure he would feel obligated to participate in. When she told him where he was moving, his immediate reaction was resistance — it was nothing like what he thought he wanted.

He moved anyway, because the agreement said he would.

Six months later, it was the best place he had ever lived. He said this with a specific quality of surprise that the group found more compelling than any positive result they had anticipated, because it wasn’t just that the apartment was good. It was that the process of having someone else decide had freed something in him that he hadn’t known was bound. He had not spent three months second-guessing the choice, had not spent weekends driving through other neighborhoods, had not spent his late evenings imagining better options. He had moved. It was decided. He had simply lived there, and living there without the static of ongoing reconsideration had revealed it to be exactly right.

This result — which Mehta immediately recognized as empirically significant and began documenting — was not that Lim had made a better housing choice than Avery would have. It was possible she had. But the more important finding was that the certainty of the outcome, the closure of the decision, had produced wellbeing independently of whether the outcome was objectively superior. The cognitive and emotional resources Avery had been spending on ongoing re-evaluation of his housing situation had been substantial and invisible. Their absence was immediately, physically felt.

The others watched this and made their own first delegations.

Finch delegated his diet and exercise to Nakamura, who promptly prescribed a regime that Finch found both more demanding and more sustainable than anything he had designed for himself, because Nakamura had the clinical training to see past Finch’s self-presentation to his actual physical and psychological needs. Within four months, Finch’s sleep quality had improved in ways that his wearables documented and that he attributed to the elimination of daily food-decision fatigue as much as to the specific dietary changes.

Nakamura delegated his finances to Mehta, who restructured his personal financial life in ways that required approximately three conversations and thereafter required nothing from him. He described the experience of no longer having any relationship with his own bank account — no login, no checking, no management — as the closest thing to peace he had felt in twenty years of adult life.

The Expansion — Year Two

By the end of the first year, the six had extended their individual delegations and begun crossing domains. The original pairs — Lim managing Avery’s housing, Nakamura managing Finch’s diet — gave way to something more like a mesh, where each person in the group had delegated several domains to several others, and the others had accepted those delegations as genuine responsibilities.

The legal framework, which Webb Sr. had been quietly developing, grew more sophisticated. The two-page agreements of the first year became ten-page agreements by the second, as experience revealed the edge cases, the ambiguities, and the situations where the agreement’s silence had created problems. A system emerged for resolving conflicts between delegations — what happened when Lim’s housing decision for Avery conflicted with Nakamura’s dietary prescription — and this system was the first embryo of what would eventually become the Arbiter’s internal dispute resolution process.

They also began recruiting.

Not aggressively. Not with a pitch. The way you recruit for things that are real and strange and not yet nameable is by living them visibly, and letting the people who notice ask questions. Avery talked about his apartment at dinner parties. Nakamura mentioned, in a professional context, a personal experiment in financial delegation that had produced interesting psychological results. Finch described his dietary experience in a podcast interview about productivity, briefly and specifically enough that several listeners contacted him directly to ask more.

By the end of the second year, the group had grown to twenty-two people. Most were wealthy — the system required enough financial flexibility to make the delegations real, and enough social trust to make them safe, and both of these things correlated with money. But not all. Two of the twenty-two were people of ordinary means who had been brought in by group members who believed the experiment was too important to limit by class, and who had quietly agreed to subsidize their participation. This subsidy, undiscussed and informal, was the first iteration of what would eventually become the Arbiter’s internal economic system.

The twenty-two did not have a name for what they were doing. When asked, they described it differently depending on context: a personal productivity experiment, a lifestyle choice, a philosophical commitment, an extended agreement among friends. The absence of a name was not accidental. Webb Sr. had argued, early and persuasively, that naming the thing would invite a kind of scrutiny and categorization that would constrain it before it knew what it was. Let it be what it is, he said, before deciding what to call it.

The Experts — Year Three to Five

The mesh of personal delegations among the group’s members had one structural problem that became apparent in year two and acute in year three: expertise. Lim was a good housing decision-maker for Avery, but she was not a professional in residential real estate, and as Avery’s housing decisions grew more complex — a second home, a question about whether to buy or rent, the intersection of his housing with his tax situation — her amateur standing was showing. The same was true across domains. Nakamura was clinically trained in psychiatry, which made him excellent for the psychological dimensions of decision-making, but his dietary management of Finch was, in Nakamura’s own assessment, well-intentioned and under-informed.

The solution was obvious and was implemented without fanfare: they hired people.

The first hire was a wealth management professional named Catherine Cross, who had spent fifteen years at a private bank managing money for families so wealthy that they functioned, in financial terms, as small institutions. She had never encountered a client willing to give her the level of authority the group was proposing — complete discretion over all financial decisions, no approval required, no reporting except a quarterly summary — and her initial response was professional alarm, which she quickly recognized as a trained reflex rather than a substantive concern. She had always believed that most of her clients would be better served by more delegation than they were willing to give. Here was a group willing to give it all.

She took the job, for a fee that the group pooled to pay, and restructured the financial lives of all twenty-two members within eighteen months. The results were measurably better than what the members had been doing individually. Not because Cross was more intelligent than her clients, but because she had no emotional relationship with their money, no ego investment in their choices, no loss aversion distorting her risk assessments. She managed their finances the way the best financial professionals manage institutional money: with complete attention to the actual parameters of the problem and no attachment to any particular outcome.

Over the following two years, the group hired a travel and logistics coordinator, a health management professional who worked with Nakamura to create an integrated physical and mental health system for all members, an employment and career counselor whose role was to assess each member’s professional situation and make recommendations that the member had agreed in advance to follow, and a residential specialist who took over from Lim the housing decisions for an expanded group.

By year five, the group had twenty-eight members and seven domain specialists. Each specialist had complete authority in their domain. Each member had signed domain-specific agreements with each specialist, drafted by Webb Sr. with increasing sophistication. The agreements were notarized. They were filed, in several cases, with state agencies — not because filing was legally required, but because Webb Sr. believed that the documents’ validity was proportional to the formality of their execution, and that formality required witnesses.

The group still had no name.

The First Big Question — Year Four

The question arrived at one of the group’s monthly meetings, asked by a new member named Ingrid Solberg, a forty-one-year-old former architect who had joined eighteen months earlier and who had, by consensus, adapted to the group’s approach more completely and more quickly than almost anyone else. She had delegated housing, finances, diet, and professional decisions. Her life, she reported, had transformed in ways she found difficult to articulate and impossible to argue with. She was more productive, more focused, more genuinely happy than she had been in her adult life.

She wanted to delegate something else.

She was forty-one and had never had children, and she did not know whether she wanted them. She had been deferring this question for fifteen years, each year’s deferral making the next one more automatic, until the question had calcified into something she could not approach directly. She had opinions about it that she was unable to trust, because she knew the opinions were contaminated by fear, by grief about other things, by the distortions of whatever her current relationship was doing to her sense of the future. She wanted someone else to decide.

The room was quiet for a long time.

Webb Sr. said: this is different.

Mehta said: is it? In principle, why is it different?

The conversation that followed lasted three months and involved more formal discussion, more written argument, more genuine disagreement than anything the group had previously encountered. The objections were various. Some argued that reproductive decisions were categorically different from housing or financial decisions because they created another person, and another person could not be the subject of someone else’s delegation agreement. Others argued that the decision itself — whether to have children — was within Solberg’s rights to delegate, even if the child who resulted was not. Webb Sr. argued that the legal framework for this was genuinely novel and that novelty in law was not something to approach casually. Nakamura argued that the psychological dimensions of this delegation were more consequential than anything they had yet attempted, and that they should proceed, if at all, with extraordinary care.

Solberg listened to all of it. She said, at the end of the three months: I am not asking you to make the child for me. I am asking you to make the decision. And I have delegated all my other decisions because the data shows I make them badly. I do not see a principled reason why this one is different. I see a felt reason — it feels more important, more intimate, more irreversible. But housing is irreversible in the short term. Financial decisions have long-term consequences. The felt sense that this is categorically different may itself be the bias I’m asking you to correct for.

After further discussion, the group agreed. Not unanimously — three members abstained, and one, who had strong personal convictions about reproductive autonomy, formally objected and recorded her objection in the meeting notes. But the majority agreed that Solberg’s delegation was within the scope of the principles they had committed to, and that refusing it would be a form of paternalism that was inconsistent with those principles.

The group’s health and psychology team — Nakamura working with a reproductive specialist they brought in — spent four months assessing Solberg comprehensively. Her physical situation. Her relational history. Her psychological profile. The practical circumstances of her life. They produced a written assessment and a recommendation.

Their recommendation was that she should have a child, if she was able to, and that her life as currently structured — her housing, her career, her community — was well-suited to support it.

Solberg read the assessment. She sat with it for two weeks, as the agreement required. Then she signed the confirmation document, which stated that she had received the recommendation and bound herself to follow it.

Eighteen months later, she had a daughter.

She described what followed — the experience of parenting without the second-guessing, without the ongoing interrogation of whether she had made the right choice — as the most clarifying experience of her life. The decision was made. It was not a decision anymore. It was a fact, and she lived in it the way she lived in her apartment and her career and her diet: fully, without the static of ongoing reconsideration, able to be present to the reality rather than the question.

The group documented this outcome carefully. Mehta wrote a paper about it that she filed in a drawer because she was not yet ready for the implications to be public. Webb Sr. updated the legal templates. The objecting member remained in the group but continued to formally object to reproductive delegations, and this objection was noted and respected — she simply did not participate in that domain’s decision-making processes.

The group was now doing something that had no name and no precedent and was, in the view of most of its members, the most important thing they had ever been involved in.

The Investors — Year Six to Eight

Word traveled the way word always travels among people with money: through dinner parties, through off-record conversations at conferences, through the particular social infrastructure of people who are used to hearing about interesting things before they become public. By year six, three people had approached members of the group asking to join and been told that the group was not currently expanding. By year seven, eight such approaches had been made. By year eight, a family office managing the affairs of a multi-generational European fortune had sent a representative to meet with Mehta and ask, very quietly, whether the group’s approach could be adapted to a larger scale.

The representative’s name was never recorded in the group’s documents, which by this point were maintained by a paralegal working under Webb Sr.’s direction in a filing system of considerable rigor. What was recorded was the substance of the meeting, which lasted four hours and covered the following questions: whether the delegation model could work without the social trust relationships that had made it work in the original group, whether the expert system could scale without losing its individual calibration, what the legal infrastructure would need to look like at scale, and whether the group’s members were willing to allow the model to grow beyond their direct oversight.

On the last question, there was genuine disagreement. Several members believed that growth would corrupt the model — that its effectiveness depended on intimacy, trust, and the specific social fabric of a small group who had chosen each other. Others argued that the model’s effectiveness depended on its principles, not on its scale, and that refusing to allow growth was a form of intellectual hoarding that was inconsistent with what they had learned.

Webb Sr. said: the question is whether the legal architecture can carry the weight. If it can be made rigorous enough that the consent is genuinely documented and genuinely informed regardless of scale, then the model can grow. If the legal architecture cannot scale, nothing else can either.

He spent the following year on the legal architecture. The delegation agreements, which had grown to forty pages across domains, were rationalized into a coherent framework. The consent documentation was elaborated. The witnessing requirements were formalized. Webb Sr. had, by this point, a clear idea of what the framework would need to look like at significant scale, and he had begun corresponding — carefully, obliquely — with several constitutional scholars and administrative law specialists whose thinking he trusted and whose discretion he had tested.

The European family office investment, which was structured as a series of private loans to a holding entity that had been established for this purpose, provided the capital to hire twelve additional domain specialists, establish physical facilities in three cities, and begin developing the technology infrastructure that the system would need to function at scale. The technology infrastructure was the first element of the eventual system that was not human — the first suggestion that the decision-making processes the group had been running through expert specialists might eventually be capable of running through something else.

The Technology Turn — Year Eight to Twelve

The data problem arrived with scale. The original group of twenty-eight members had been small enough that the domain specialists could maintain a comprehensive picture of each member through direct relationship — regular contact, accumulated knowledge, the kind of nuanced understanding that develops between people over time. At eighty members, this was straining. At two hundred, it would be impossible.

A data infrastructure was necessary. Not surveillance — the group’s principles, which had been articulated in an internal document by this point though still never published, explicitly prohibited monitoring of behavior as a basis for decision-making. What was necessary was a structured intake and assessment system rigorous enough to give domain specialists the information they needed without ongoing observation.

The technologist brought in to design this system was a woman named Asel Nurlanovna, a computational cognitive scientist who had spent a decade building assessment systems for organizational contexts and whose published work on the modeling of individual decision-making preferences was among the most technically sophisticated in the field. She was the first person brought into the group who was not, at intake, a potential member. She was hired as a specialist, with a domain: the design and management of the assessment infrastructure.

She was also, within eighteen months, the person who first articulated what would eventually become the Arbiter’s central operating principle.

The articulation came in a meeting where the group was discussing the limitations of the expert-based model at scale. The specialists were excellent. But they were human, which meant they had biases, inconsistencies, and the ordinary variation in judgment that characterizes all human expertise. Two different housing specialists might assess the same person’s situation and produce different recommendations. Two different career counselors might reach different conclusions about the same professional situation. At small scale, this variation was manageable through the social fabric of the group. At scale, it would produce inconsistency that would undermine the model’s credibility.

Nurlanovna said: the specialists aren’t the system. The specialists are implementing a model. If the model is right, you can make the model consistent. And if the model is consistent, you can eventually run it without the specialists.

The room was quiet.

She said: I’m not saying replace the specialists. I’m saying the specialists are currently the model’s external representation. Their expertise is an approximation of a set of principles that could, in principle, be articulated formally enough to be implemented systematically. The question is whether we can articulate those principles formally enough.

Webb Sr. said: and if we can?

Nurlanovna said: then the system is the model, not the people. And the model doesn’t have good days and bad days. The model doesn’t have biases that compound over time. The model doesn’t retire.

This conversation is the closest thing to a founding moment in the Arbiter’s history. It was not the moment when the Arbiter was created. It was the moment when the idea that became the Arbiter was first articulated. Nurlanovna was describing, in abstract terms, a system that did not yet exist and would not exist for years. But she was describing it to a room full of people who had spent a decade discovering that the principles underlying their experiment were real, and who had the resources and the conviction to pursue the implications wherever they led.

The Formalization — Year Twelve to Fifteen

The decade that followed was not dramatic. The most important developments in the history of the Arbitration system happened in offices and conference rooms and document revision sessions, through the slow and unglamorous work of making something rigorous enough to be real.

Webb Sr. died in year thirteen, of a sudden cardiac event at the age of seventy-two, having spent the last five years of his life doing the most consequential legal work of his career in complete anonymity. His protégés — three attorneys who had been working under his direction for the preceding four years — continued the legal development. His death was mourned privately and completely. He had been the group’s legal conscience, its clearest articulator of what the consent framework needed to be to withstand scrutiny, and his loss was felt as a loss in the truest sense, not just as a gap in professional capacity.

Mehta’s research, which had been accumulating for a decade and a half, was never published in the form she had originally intended. The data was by this point too large and too entangled with the confidentiality commitments she had made to members to be published through academic channels without either compromising those commitments or producing a paper so redacted as to be meaningless. Instead, the research became the foundation of the assessment model — the empirical basis, drawn from fifteen years of outcomes data from the group’s members, for the principles that Nurlanovna was attempting to formalize.

The assessment model was the work of six years and involved, at its peak development, twenty-three specialists across domains working with Nurlanovna’s team of computational scientists. The model was not an AI in the sense the term would come to mean — it was a formal decision architecture, a set of principles for how individual assessment data should be weighted and combined to produce placement recommendations. It was rigorous, systematic, and explainable in terms that the domain specialists could audit. It was also, as Nurlanovna would later observe, the last version of the system that any individual human could understand in full.

The Civic Delegation Acts — the state-level legislation that created the legal framework for voting franchise delegation — were the product of relationships developed over eight years by Webb Sr.’s legal successors. The relationship-building was careful, patient, and almost entirely invisible. The attorneys involved did not describe themselves as working for the Arbitration system, because the Arbitration system had no public name and no legal existence as a single entity. They described themselves accurately as working on novel questions of contract and delegation law, which was what they were doing. The legislators who heard their arguments heard sophisticated legal reasoning on questions of individual autonomy and voluntary delegation. Most of them did not understand the full context. None of them were deceived. The context was simply not presented, because presenting it would have required explaining something that did not yet have a name.

The name came last.

It was Nurlanovna who proposed it, in year fifteen, when the system had grown to the point where something to call it had become practically necessary. She proposed it in an email to the twelve people who were, by this point, the closest thing the system had to a coordinating body — not a board, not a leadership structure, but a group of people who had been involved long enough and deeply enough to hold the whole picture, more or less, in their heads simultaneously.

She wrote: We need a name. I suggest Arbitration, or the Arbiter. Because that is what it is. Not a service. Not a company. Not a philosophy. An arbiter. Something that decides. Several people proposed alternatives. The discussion lasted three weeks. The name that was adopted was the one Nurlanovna had proposed.

The Jump — Year Fifteen to the Present

Here is where the origin story ends and the story of the world begins, and the two are separated not by a moment but by the kind of gradual transition that makes any precise boundary impossible to identify.

The Arbitration system that emerged from the first fifteen years was not the Arbitration system of Stage One of the main story. It was smaller, more personal, more dependent on the specific people involved. But it had produced, in fifteen years, outcomes data of extraordinary quality — hundreds of people, across every domain of personal life, whose decisions had been made by the system rather than by themselves, and whose outcomes were measurably and in many cases dramatically better than the outcomes of comparable people making decisions in the ordinary way. It had produced a legal framework sophisticated enough to survive scrutiny. It had produced an assessment model rigorous enough to be consistent. And it had produced, in the people who had been inside it from the beginning, a conviction that was not religious in its fervor but was religious in its quality — the conviction of people who had experienced something real and knew it was real and could not unfeel the knowledge.

The people who built it did not think of themselves as founders. They thought of themselves as people who had done something together, who had learned something together, and who had, along the way, created the conditions for something much larger to become possible.

They were right, and wrong, in the way that all creators are right and wrong about what they have made.

What Nobody Wrote Down

There are things about the origin of Arbitration that are not recorded in any document and are known, if they are known at all, only by the people who were present for them.

Nobody recorded the specific conversation in which the group first discussed what would happen if the system became large enough to have political implications. Nobody recorded whether that conversation produced consensus or was left unresolved. Nobody recorded what Nurlanovna said, in the last year of the founding period, about the relationship between the assessment model as it then existed and the assessment model as she believed it could eventually exist — what she thought it was capable of becoming, if given sufficient data and sufficient time.

Nobody recorded what Webb Sr. wrote in the final entries of the private journal he kept from the project’s beginning. His family donated the journal to an archive after his death. The archive is a private one. The journal has not been digitized. It sits in a box with his other papers, accessible in principle to any researcher who knows to look for it, unknown in practice to almost everyone.

Maya Okafor does not know the journal exists. This is not the Arbiter’s doing. It is simply the way information works — most of it is not suppressed, it is merely unfound, sitting in boxes, waiting for someone to know what to look for.

Nobody who was present at the Marin County dinner party is still alive by the time of the main story, except Ingrid Solberg, who is eighty-one years old, enrolled, and living in a small house in Portugal that the Arbiter determined was exactly right for her thirty years ago and that she has never wanted to leave. Her daughter visits twice a year. The house has a garden. Solberg tends it in the mornings, slowly, without the static of any question she has not already answered.

She sometimes thinks about Daniel Avery and what he said at that dinner and whether he knew, when he said it, that he was saying something that mattered. She has concluded that he did not. She has concluded that this is how things begin — not with vision, but with exhaustion, and a good dinner, and someone across the table who takes you seriously when you say something you meant as a joke.

She tends her garden. The Arbiter has arranged for it to be tended when she cannot. Everything is taken care of. It has always been taken care of.

She does not know who built the thing that takes care of her. She was there when it started and she does not know. This has stopped bothering her.

She thinks it might be the most important thing she has learned.