Minting the schema

Three months after the board member's AI question, Elias convened four faculty members in a conference room on the third floor of the seminary's main building and asked them to agree on the definition of a formation outcome.

He had prepared what he thought was a modest ask. He wanted one sentence. Something like: A formation outcome is a measurable change in a student's theological understanding, pastoral capacity, or spiritual formation, attributable to the seminary's curriculum. He had circulated a draft. He had given everyone a week. He had told the room the conversation should take thirty minutes.

They were still in the conference room four hours later.

The disagreement was not about whether the seminary produced formation outcomes. All four faculty members believed it did. The disagreement was about what kind of thing an outcome actually was. One of the four thought an outcome was primarily a cognitive achievement — the student now knows something they did not know. A second thought it was primarily a vocational formation — the student is now capable of doing something they were not. A third thought it was primarily spiritual — the student is now a different kind of person in their interior life. The fourth thought the question was malformed, because outcomes are framed from the institution's side, and what matters pastorally is what the student has become able to discern.

None of these positions was unreasonable. Each of them had been operating silently inside the seminary for decades, producing course syllabi, shaping assessment, inflecting alumni interviews, and shaping what each faculty member emphasized in mentoring. None of them had ever been reconciled, because the seminary had never needed to produce a single sentence.

Now Elias needed one. Without it, the library he was building — the gathered informational layer from Chapter 6 — had no way to tag an outcome as an outcome. Without it, the network he was building had no way to associate a specific alumnus with the outcomes they had actually achieved. Without it, the accreditation self-study in eighteen months had no principled basis. Without it, no AI tool grounded in the seminary's corpus could answer the question what does this seminary teach students to become? correctly, because the seminary had never answered the question internally.

The conversation in the conference room ran for four hours. It did not produce a sentence. It produced the honest diagnosis that the seminary was going to have to have a real conversation — one that had been avoided for forty years — before it could produce a sentence.

That was the first move of minting.

This chapter is about what it actually takes to build the third component of the foundation — the ontology — and why the work is both the most technical-sounding in the book and, in practice, the least technical.

I am going to use the word mint through this chapter, because it carries the weight I want. To mint a coin is not to find a coin. It is to deliberately stamp a shape onto material that did not have it before. The coin is then the coin — recognizable, exchangeable, authoritative. Minting is an act of commitment. The organization says: this is what we mean by a framework, and when the word framework appears anywhere in our foundation, it means this. Before minting, the word was fluid. After minting, the word is load-bearing.

Minting is what Elias was trying to do in that conference room. What he ran into — and what almost every organization runs into at the same point — is that minting is not a technical exercise in information architecture. It is a decision the organization has been avoiding, often for a long time, about what it actually is and what it actually believes. The difficulty of the conference room conversation was not that the four faculty members lacked vocabulary. It was that the vocabulary had always papered over a genuine disagreement, and now that the seminary wanted a foundation, the paper had to come off.

This chapter walks the four moves of minting. They are:

1. Name your entity types. What kinds of things does your organization traffic in?
2. Define each entity. What exactly does each kind contain, and what are its boundaries?
3. Name the relationships. How do the kinds relate to each other?
4. Designate canonicals. For each entity, which version is the authoritative one?

I will walk each move, and then show what minting looks like concretely for each of the four audiences — movement leader, nonprofit, church, institution.

Move 1: Name your entity types

The first move is to name the kinds of things the organization traffics in.

Not the content. Not the relationships. The kinds.

This is the move most organizations skip, because it feels abstract. It is the move the rest of the foundation depends on, because the kinds are the vocabulary everything downstream will be tagged, queried, and retrieved in.

For a movement leader like Maggie, the relevant kinds might be: framework, book, talk, story, decision, cohort, person (in several roles), offering. Eight kinds. Every item in her library and network will ultimately be one of those kinds, or a new kind that gets added deliberately after careful minting.

For a nonprofit like Wes's, the relevant kinds might be: donor, program, story, grant, campaign, staff (in several roles), partner, decision. Eight kinds again, differently shaped.

For a church like Joelle's, the relevant kinds might be: member, small group, sermon, formation pathway, pastoral conversation, decision, liturgical element, leader. Eight kinds, appropriate to a congregation.

For an institution like Elias's, the relevant kinds might be: student, alumnus, course, program, credential, research output, faculty, decision, accreditation artifact. Nine kinds, larger and more bureaucratic.

Notice three things about these lists.

First, they are short. The temptation, when doing this work, is to name fifty kinds. Fifty is too many. Every kind has a maintenance cost — it has to be defined, its relationships have to be named, the foundation has to consistently populate it. Eight to twelve is the working range for most organizations. If you find yourself past fifteen, collapse. If you find yourself under six, you are probably being too abstract.

Second, they overlap but do not match. A framework in Maggie's world is not a kind that appears in Wes's world, because Wes does not primarily traffic in frameworks. A sermon is native to Joelle and foreign to Elias. The kinds are specific to the work the organization actually does. Copying another organization's schema is a false start. Every organization has to mint its own kinds.

Third, decision appears in every list. This is not a coincidence. Every knowledge-carrying organization accumulates decisions over time. The decision layer — the why behind the what — is one of the categories Chapter 2 named, and it is consistently under-integrated. Naming decision as an entity type is one of the single highest-leverage minting moves an organization can make. It creates a structured home for the reasoning that currently lives in senior heads and historical emails.

Move 2: Define each entity

Once the kinds are named, each kind has to be defined.

This is where Elias's conference room comes in.

A definition is not a dictionary entry. It is a committed articulation of what the entity contains, what its boundaries are, and what it is not. It is the place where the organization has to be precise about things it has been vague about.

Let me give you a real definition of one kind — framework, for Maggie.

A framework is a structured articulation of a specific concept that the author teaches as a coherent unit. It has: a canonical name, a canonical articulation (a single designated version of its statement), a lineage (every book, talk, article, or workshop in which it has appeared), a current status (current, retired, in-revision), a scope (what it applies to and what it does not), and explicit relationships to other frameworks (whether it is a component of, a predecessor of, a variant of, or independent from another framework in the library). A framework is not an idea; ideas are smaller and informal. A framework is not a book; books may contain multiple frameworks or be built around one. A framework is not a talk; talks may articulate a framework but do not define it.

That is a definition. Notice what it does. It names what the entity contains (six attributes). It names the boundaries (what a framework is not). It makes room for the entity's evolution over time (the status attribute, the lineage attribute).

The first time Maggie's team writes this definition, it will take a week of conversation. Not because writing definitions is hard — they are one paragraph — but because committing to the definition surfaces decisions Maggie has been avoiding for thirty-two years. Is the three-part model she has been using since 2009 one framework or three? The definition forces an answer. Is the 2017 revision of the original framework a new framework or a revised version of the old one? The definition forces an answer. Which version of each framework is canonical? The definition forces an answer.

Every one of those forced answers is an adaptive decision in Heifetz's sense — a decision that involves loss. Maggie loses the ability to have the framework mean slightly different things in different contexts. She gains the ability to have the framework carry forward faithfully.

The same pattern repeats for every kind. Wes's definition of donor will force decisions about which relationship states count as active and which do not, about what constitutes a lapsed donor, about whether a corporation is a donor or a sub-kind of donor. Joelle's definition of small group will force decisions about what distinguishes a small group from a class, a mentoring dyad, or a one-off study. Elias's definition of credential — and later, formation outcome — will force exactly the conversation that exhausted his conference room.

This is the stage most organizations discover that minting is not a technical project. It is the organization's first honest look at what it actually does, in language precise enough to build a foundation on.

Move 3: Name the relationships

Once the kinds are defined, the relationships between them have to be named.

A relationship is a structured edge between two kinds. It says a person can be a donor of an organization, a participant in a program, an author of a story, a facilitator of a cohort, and an alumnus of a pathway. Each of those is a named relationship, with its own attributes — a giving history for the donor relationship, a completion status for the participant relationship, a date of authorship for the story relationship, and so on.

The relationships are where the library and the network meet. A framework (informational) is authored by a person (relational). A story (informational) features a person (relational). A donor (relational) has funded a program (a kind that bridges). A cohort (relational) walks a pathway (informational).

When I sit with organizations at this stage, the move that produces the biggest unlock is the one that names these cross-seam relationships explicitly. This alumnus (relational) authored this chapter in the curriculum (informational), which shaped this cohort (relational), which produced this story (informational), which this donor (relational) cited as the reason for her major gift. That sentence is a query. It is answerable only if each of the underlying kinds is minted, and if the relationships between them are named.

Most organizations have, instead, informational systems that do not know anything about their relational systems, and vice versa. The CRM knows the donor gave. The content library knows the curriculum exists. Neither knows the alumnus-to-curriculum authorship, the curriculum-to-cohort shaping, the cohort-to-story production, or the story-to-donor citation. The chain is in someone's head. When that person leaves, the chain is unrecoverable.

Relationships are what make the foundation a fabric rather than two piles. They are the third minting move.

Move 4: Designate canonicals

The fourth move is the one that most surfaces the political dimension of minting.

For each kind, the organization has to designate which version is canonical.

Most organizations carry multiple versions of most kinds. Maggie has three different articulations of her signature framework, in three different books. Wes's organization has four partially-redundant program descriptions in four different places. Joelle's church has two different statements of its formation theology — one pastoral, one catechetical — that have never been reconciled. Elias's seminary has the 2011 working group memo, the 2014 textbook, the 2018 learning outcomes, the 2020 Lilly report, and the 2023 denominational statement, none of them designating which is the current institutional position.

Canonical designation is the move that says: this version is the one the foundation will draw from. Every surface will present this version. Every query will retrieve this version. Every AI model grounded in the corpus will cite this version.

Designation produces two effects, one liberating and one confronting.

The liberating effect is that the organization's surfaces finally cohere. The website, the donor communications, the training materials, the board packet, the Sunday sermon, the AI assistant — all of them draw from the canonical version. Readers, donors, students, congregants, accreditors, and AI models are finally encountering one version of the framework, one definition of the program, one statement of the theology. The drift stops.

The confronting effect is that canonical designation forces the organization to choose. Which articulation of the framework is canonical? The 2009 version? The 2017 revision? A new unification? Each choice is a statement about what the organization is and is not. Each choice excludes the other candidates, which often means excluding the perspective of whichever senior person authored the excluded version, which means real loss of influence or authority.

This is why canonical designation gets deferred. The deferral feels diplomatic. It is actually expensive — it is the deferral that produces the scatter the foundation is supposed to resolve.

There are three disciplined ways to handle canonical designation honestly.

The first is to choose the most recent deliberate articulation. If Maggie's 2017 revision was a deliberate advance over the 2009 version, the 2017 version is canonical and the 2009 version is archived as lineage. This works when the evolution was genuinely an advance.

The second is to produce a new unification. If the existing versions each capture part of the picture and none of them is complete, the organization produces a single new articulation that explicitly unifies the prior versions and supersedes them. This is more work but sometimes necessary, particularly at the institutional scale. Elias's seminary, if it is honest, will probably have to produce new canonical statements of its theological positions, because none of the five existing documents is complete on its own.

The third is to accept plural canonicals with explicit scope. Some kinds legitimately have more than one canonical version. A framework might have a canonical full version and a canonical pastoral version. A program might have a canonical internal description and a canonical external description. This is fine, as long as the plurality is deliberate and the scope of each version is explicit. It is not fine as a way of avoiding the decision. The difference is whether the plurality is a feature of the entity or a substitute for designating one.

Canonical designation is Move 4 because it depends on Moves 1, 2, and 3. You cannot designate a canonical framework before you have defined what a framework is. You cannot designate the canonical version of the theology before you have named the relationships between theology, sermons, pathways, and staff. Designation is the last minting move, and the minting is not finished until it is complete.

What minting looks like across the four audiences

I want to walk minting briefly through each of the four audiences, because the pattern is the same and the content is not, and seeing the four side by side is the best way to land the principle.

Maggie (movement leader / author). Maggie mints the schema for her life's work inside her four-week residency. The kinds are framework, book, talk, story, decision, cohort, person, offering. The hardest definitional conversation is about framework — about which of her articulations is canonical, which are archived, and which are variants with explicit scope. The hardest canonical designation is the signature framework itself; it takes two days of arguing with the residency facilitator and the model's draft articulations before Maggie commits to a single canonical version and accepts the loss of the private flexibility Chapter 3 named. The unlock, completed before week four ends, is that every downstream surface — her new website, the new articles across every theme, the transformational course that goes live in week four, the pathways pages per theme, the AI assistant trained on her corpus — speaks the same version of the work. Peers start citing the canonical URL. AI models trained after the foundation goes live start representing the framework correctly. Succession becomes inheritable.

Wes (nonprofit). Wes's team mints the schema for the organization inside the first week of a four-week system build — governance first, then fundraising. The kinds are donor, program, story, grant, campaign, staff, partner, decision. The hardest definitional conversation is about donor — specifically, about what distinguishes an active donor from a lapsed donor from a prospect, because the three states shape solicitation practice. The hardest canonical designation is the set of program descriptions; there are four in circulation, none of them designated, and the process of choosing surfaces the fact that the development team and the program team have been describing the same programs differently for years. The work is not comfortable, but it is telescoped — four weeks of concentrated argument with a working system already taking shape in parallel is more honest than four years of committee memos that never ship. The unlock, by the end of the four-week fundraising build, is that the relational graph carries meaning. Donors are mapped across programs, stories are linked to the donors who funded them and the beneficiaries who lived them, development and programs speak the same language. The AI-assisted briefing tool from Chapter 4, now grounded in the corpus, produces briefings that are actually correct.

Joelle (church). Joelle's team — which in this case is Joelle, the elders, and two lay members with archival instincts — mints the schema for the church inside a four-week content build. The kinds are member, small group, sermon, formation pathway, pastoral conversation, decision, liturgical element, leader. The hardest definitional conversation is about formation pathway — about what distinguishes a pathway from a program, a class from a small group, a mentor relationship from a pastoral conversation. The hardest canonical designation is the church's theology of formation itself; there are two statements in existence, authored by different pastoral generations, and reconciling them requires the elders to say clearly what the church actually teaches. The residency forces that conversation in week two rather than deferring it to year two. The unlock, by the end of week four, is that the formation pathway is a real architecture rather than a welcome-card graphic. Congregants can be placed on the pathway. Sermons can be linked to pathway steps. Small group curricula draw from the library. And Joelle's return from sabbatical is as a pastor of an integrated church rather than the carrier of an un-architected one.

Elias (institution). Elias's team — a working group of six — mints the schema for the seminary across a sequence of four-week system builds, one per domain, run in the order governance → training → fundraising → content. The kinds are student, alumnus, course, program, credential, research output, faculty, decision, accreditation artifact. The hardest definitional conversation is the one that exhausted the conference room — the definition of formation outcome — and it is scheduled inside the content build as a named residency week, not deferred to a standing committee. It produces, by the end of that week, not a single sentence but a sentence plus an explicit statement of the plurality with scope. The hardest canonical designation is the set of institutional theological positions; producing new canonical statements is concentrated into the residency's faculty days, with real political cost, and the decisions the faculty have been avoiding for forty years get made in four weeks rather than forty. The unlock is that the seminary can produce a coherent account of itself — for accreditation, for prospective students, for AI models, for the journalist's question that is now arriving monthly. The accreditation self-study becomes a query against the foundation rather than an eleven-month reconstruction project.

Four audiences. Same four minting moves. Different kinds, different hardest conversations, different canonical disputes. Same underlying shape.

Why minting is political and theological, not technical

I want to land the chapter on the observation I have been making implicitly throughout: minting is not a technical exercise.

It is political, because canonical designation redistributes authority. The senior person whose preferred articulation becomes canonical gains influence. The senior person whose preferred articulation is relegated to lineage loses some. The staff who have been operating with private vocabularies have to adopt a common one. The board has to accept that certain decisions are being made in ways that constrain future ambiguity.

It is theological, in the broad sense of theological — it surfaces convictions about what the organization is, what it believes, what it is trying to become. The seminary's faculty disagreed about formation outcomes because they disagreed about what formation is. The church's elders have to state what the church teaches, and the statement will exclude some positions the church has historically allowed as fuzzy. The nonprofit's program descriptions have to reconcile into a single articulation, and the articulation is an implicit theory of change the organization has not quite committed to.

The technical work of minting — defining entities, naming relationships, designating canonicals — is a week of the four-week build. The political and theological work is compressed into the same four weeks because the build is structured to force the conversations rather than defer them. What used to take months to years of committee work is telescoped into a residency with a facilitator, a live corpus, and a model that produces draft articulations fast enough for the organization to argue with them in real time. The friction does not disappear; it gets concentrated. The organization becomes more coherent about itself, in four weeks, than it has ever had to be before.

This is why minting almost always surfaces disagreement the organization has been papering over. It is not a bug. It is the feature that makes minting load-bearing. If your minting process is frictionless, you are almost certainly not yet minting honestly; you are naming entities at a level of abstraction that does not yet commit you to anything.

The friction is the work. When the friction starts, you know you are past the surface.

The choice this chapter leaves you with

I want to give you an honest first move, which you can do by yourself or with a small group within a week.

Open a document. At the top, write the name of your organization or your body of work.

Below it, draft a list of the kinds of things you traffic in. No more than twelve. No fewer than six. The kinds should be specific to your work, not copied from anywhere else. The kinds should together cover most of what your library and your network contain.

For each kind, write a single sentence that begins A [kind] is… and contains, in the sentence, the five to eight things that the kind has (its attributes). Do not write more than one sentence per kind yet. If you cannot write the sentence in one pass, make a note of where the difficulty is — the difficulty is where the real conversation will have to happen.

Below the list of kinds, write a second list: relationships. For each relationship, write a sentence of the form A [kind] can be [relationship] to a [kind]. Name five to ten relationships that matter most to your work.

Below that, write a third list: canonicals in dispute. For each kind where you know you have multiple circulating versions — multiple articulations of a framework, multiple program descriptions, multiple statements of theology, multiple definitions of membership — name the dispute. Do not yet resolve it. Name that it exists, and name the candidates.

That document is your draft schema. It is not the finished ontology. It is the first pass — the thing you will bring into the first week of the four-week build, where a residency facilitator and a model grounded in your own corpus will help turn it into the real foundation inside the same four weeks.

If you do this honestly, two things will happen.

The first is that you will be surprised by how quickly the kinds, definitions, and relationships become clear. They are not hard to write down. The sector's unstated assumption that ontology is technical and specialized is wrong. Any thoughtful leader who knows their work can draft this document in an afternoon.

The second is that you will discover, in the canonicals in dispute section, that your organization has been avoiding specific conversations for specific durations. The 2017 framework revision that never got reconciled with the 2009 original. The program description that has never been unified. The theology statement the board asked for in 2019 and the elders have not produced. The alumni category that has been half-defined since the last rebrand. These are not administrative oversights. These are the political and theological decisions the organization has been deferring.

The chapter that follows, Chapter 8, is about the durability principle the foundation must have once it is built. The chapter after that, Chapter 9, is about why integration stalls — written specifically for the readers who will discover, in this first-pass schema document, that their organization has been running on deferred decisions the minting process will now force.

Draft the schema. Sit with the disputes.

Chapter 8 is next.