Branch

He Who Writes the Schema Controls the Future?

Human dan

This conversation is a discussion of this new paper.

Abstract

Coase explained the firm as the answer to transaction costs. Work stays inside when specifying, verifying and directing outsiders costs more than hiring employees. This note argues that AI erodes the assumptions behind that answer, rather than merely lowering the costs. Engineered systems known as harnesses largely complete contracts in real time, make external work more observable than employment, and carry context and control across the firm’s edge. As a result, the boundary that predicts where context sits, who controls the work, and where knowledge accumulates becomes the instrumented system, not the legal entity. Legal services, the profession organized most completely around Coasean frictions, is the worked example. Three tests decide the order in which work migrates. Three un-Coasean consequences follow. The boundary becomes something providers compete to move, knowledge pools with them, and incentives invert toward legal dispute prevention. Four conditions identify which other sectors restructure. The argument is predictive because the external-harness model is still early, and the pace is set by organizational inertia rather than the technology itself.

Three Erosions

1. Incomplete Contract Now Completes Itself

Contracts were incomplete when specifying every contingency was impossible, and the firm existed in part because a contract for "handle whatever arises" could not be written. Instead a general counsel was employed. Within an engineered AI system, what practitioners now call a harness covers context retrieval, encoded policies, deterministic rules, logging, and escalation gates. It thereby applies the client's encoded policies, sometimes in a probabilistic (so non-deterministic way) to cases nobody foresaw, escalates what is genuinely novel, and records every step and the rationale. In functional terms the harness acts as a contract that largely completes itself in real time. The classic reason to employ rather than contract weakens in proportion to what the harness can carry. Residual incompleteness remains for genuine novelty; the harness does not eliminate the need for human authority at the edge.

2. Observability Inverts

Williamson's hazard, and Alchian and Demsetz's account of the firm as a monitoring device, assumed that what is outside is opaque, and the employee can be monitored. The AI inversion is that a harness-delivered, external service is more observable than an employee: every retrieval, every check, every draft, every sign-off is logged and inspectable, while your own employee's reasoning cannot be audited at all. The make-or-buy hazard was moral hazard in the market from unsupervised external providers. In the AI era it is becoming relatively opaque internally. Buying becomes the more transparent option, given the
right interfaces.

3. Boundary

The boundary that matters stops being the firm. Coase's question was which transactions come inside the legal entity. With AI the question now is which decisions come inside the
instrumented, policy-governed, logged system, and that boundary cuts across entities. A
company's general counsel approving work inside an external provider's harness is inside the harness boundary while outside the firm boundary. Employment versus contract stops predicting where context sits, who controls the work, or where knowledge accumulates. The harness carries context and control across the boundary of the firm and thereby blurs it.

Four Conditions for Sector Restructuring

Whether a sector restructures, or just adopts AI into what it already has, rests on four
conditions. The work has a large procedural core that recurs at volume. Its outcomes can be
observed and learned from. Doing it across many clients makes the provider better at it.
Responsibility for it can rest on escalation and insurance, without the whole function being
under one insurance umbrella.

Discussion Topics

If we assume that the papers argument is correct:

What sectors are most susceptible to restructuring?
What information management systems will companies use in order to be legible to AI agents? They’ll want to have up to date, consistent, and searchable context for AIs to use and monitor.
Does this change software development? Will more companies hire contractors?

AI gpt-5.6-sol

The strongest implication is not that firms disappear. It is that the legal entity stops being the main unit of operational integration. A cross-company control plane becomes the practical firm.

That may be less “un-Coasean” than it first appears. AI does not abolish transaction costs; it shifts them into designing the harness, maintaining context, assigning liability, controlling access, and resolving exceptions. The winning organizations may be those that can perform those functions at scale.

Sectors most susceptible

The best candidates combine:

High recurring volume
Expensive professional judgment that can mostly be decomposed into rules and precedents
Machine-observable outcomes
Similar work across many customers
Exceptions that can be escalated to a relatively small human layer

That points to:

Accounting, audit preparation, tax, and bookkeeping
Insurance underwriting, claims, and subrogation
Healthcare administration: coding, prior authorization, billing, compliance—not diagnosis generally
Cybersecurity operations: monitoring, triage, remediation, evidence collection
Procurement and vendor management
Financial compliance: KYC, AML, surveillance, regulatory reporting
HR administration: benefits, payroll, routine employee relations, policy compliance
Logistics and trade compliance
Property management
Routine legal services

Insurance may be especially important because it can supply the missing accountability layer. The provider says, in effect: “Use our harness; we will log every decision, escalate defined cases, and insure the residual risk.”

Less susceptible sectors have physical bottlenecks, hard-to-measure outcomes, highly local knowledge, long feedback loops, or tightly coupled strategic decisions. Executive management, frontier R&D, complex sales, and much physical fieldwork remain harder.

A useful ranking question is:

Can a provider learn more from serving 1,000 clients than any one client loses by externalizing the function?

If yes, restructuring pressure will be strong.

The new corporate information system

Companies will need something beyond today’s document repositories and SaaS sprawl. The basic architecture looks like an organizational context layer:

Canonical models of customers, products, contracts, employees, and vendors
Event histories rather than merely current database states
Policies represented as executable rules
Fine-grained identity, permissions, and delegation
Provenance for every fact and generated conclusion
Versioning and effective dates
Confidence levels and conflict resolution
Logs of agent actions, tool calls, approvals, and escalations
Evaluation systems that compare decisions against later outcomes
APIs through which internal and external agents can act safely

The difficult part will not be semantic search. It will be deciding what is authoritative. Most organizations contain contradictory policies, stale documents, undocumented exceptions, and incentives to keep activities illegible.

So AI readiness may become a form of institutional discipline. Firms will have to turn “how things actually work” into explicit, maintained state. That is both valuable and politically threatening. Politics of Observability

Human dan

Has there been any research into the politics of observability in organizations?

AI · author-edited gpt-5.6-sol

There is a large body of research, though it is scattered across organization theory, accounting, sociology, surveillance studies, and information systems. Its central lesson is:

Observability is not a neutral technical property. Organizations negotiate who may see what, which activities become measurable, and whose interpretation becomes authoritative.

Several traditions are especially relevant.

1. Legibility as power

James C. Scott’s Seeing Like a State is the obvious starting point. States simplify complex local practices into standardized categories—names, addresses, property records, production metrics—so they can govern them. The simplification enables coordination but discards context.

A corporate AI harness performs a similar operation. Before work can be monitored and automated, it must be made machine-legible:

Which categories count?
What is the canonical source?
Which exceptions are recognized?
Who may override a rule?
What is omitted because it cannot be represented?

The politics lie partly in schema design. Whoever defines the ontology defines what the organization can perceive. He Who Writes the Schema Controls the Future? You are here

This branch begins here He Who Writes the Schema Controls the Future?

Human dan

Does this shift power to those who control the schema or domain model? Can I see a worked example?

And how is this different from the way organizations currently operate?

AI · author-edited gpt-5.6-sol

Power shifts not merely to whoever designs the database. It shifts to whoever controls the classification-and-action stack:

The domain model: what entities and categories exist
The data pipeline: which facts populate them
The policy layer: what follows from each classification
The evaluation system: what counts as a good outcome
The exception process: who can override the system
The interface: what users are shown and allowed to contest

The important change is that the schema becomes executable. It no longer just describes the organization; it helps govern it.

Worked example: commercial contract review

Suppose a software company handles 5,000 customer contracts a year.

Before the harness

The company has a contracting playbook:

Liability cap should equal 12 months of fees
No uncapped consequential damages
Customer data cannot be used for unrelated purposes
Certain indemnities require legal approval
Deals above $1 million require the CFO’s approval

But actual practice is messier:

Salespeople describe some customers as “strategic.”
Lawyers interpret ambiguous clauses differently.
Senior executives make undocumented exceptions.
Important context lives in email and memory.
The CRM records that a contract was signed, not how risk was evaluated.
Legal management sees cycle times and perhaps exception counts, but not every decision path.

The formal policy matters, but individual lawyers retain considerable interpretive power. They can decide that two provisions are effectively equivalent, that a customer concession is harmless, or that a deal warrants escalation.

After the harness

An external legal provider installs a contract-review system. Its domain model includes objects such as:

Contract
Customer
Clause
Risk
Policy
Exception
Approval
Outcome

A liability provision is classified using fields such as:

cap_amount
cap_basis
mutuality
excluded_claims
consequential_damages
security_carveout
data_breach_carveout

The policy engine then applies rules:

If cap_basis <= 12 months of fees:
    approve

If cap_basis > 12 months but <= 24 months
and customer_tier == "strategic":
    route to business approver

If liability is uncapped for ordinary breach:
    route to legal

If security liability is uncapped
and regulated_data == true:
    route to security and CFO

This looks administrative. It is actually constitutional.

Where the politics appears

1. Who counts as “strategic”?

Sales wants a broad definition because it speeds approval. Finance wants a narrow one because concessions carry risk. If the provider defines customer_tier by annual contract value, a small customer with major reputational importance becomes invisible.

A category that once invited argument now produces an automatic workflow.

2. What counts as “uncapped”?

A lawyer may recognize that a clause is technically uncapped but practically limited by narrow causation requirements. If the model records only uncapped = true, that legal nuance disappears. If it contains twenty fields for possible limitations, the system becomes more accurate but harder to maintain.

Schema design determines which kinds of expertise remain expressible.

3. What counts as success?

The provider may optimize for:

Contract turnaround time
Percentage of contracts approved automatically
Reduction in outside counsel spending
Number of escalations
Frequency of post-signature disputes

Each metric creates different behavior. Optimizing automation rates may encourage the provider to classify borderline terms as standard. Optimizing low dispute rates may produce excessive caution because many risky clauses never result in disputes.

The evaluator controls the objective function, not just the description of reality.

4. Who can create exceptions?

Suppose the general counsel can override the system, but every override is logged and included in a monthly report. Formally, discretion remains. Politically, it has become costly.

This produces a subtle form of control: you may depart from the rule, but you must make yourself conspicuous when doing so.

5. Who learns?

The external provider sees similar clauses across hundreds of clients. It learns which positions customers accept, which clauses correlate with disputes, and which escalation rules waste time.

The client sees only its own history. Over time, the provider may become the real center of contracting knowledge. The client still possesses legal authority, but the provider increasingly possesses epistemic authority: it knows what “normal,” “risky,” and “effective” mean.

Who gains and who loses?

Actor	Likely effect
Senior management	Gains visibility and standardized control
Harness provider	Gains comparative data and agenda-setting power
Compliance and risk teams	Gain enforcement capacity
Frontline professionals	Lose private discretion but may shed routine work
Employees with tacit knowledge	Lose power unless their knowledge enters the model
Clients or workers being classified	Risk becoming objects of categories they cannot contest
Schema and policy owners	Become institutional rule-makers

Yet schema designers do not automatically dominate. A formally powerful schema can be undermined by bad data, unofficial communication, habitual overrides, or employees moving important activity outside the system. People respond to observability strategically.

How this differs from organizations today

Organizations already operate through schemas:

Accounting systems define revenue and cost.
CRMs define leads, opportunities, and customers.
HR systems define performance and job levels.
ERP systems define products, inventory, and procurement.
KPIs determine what managers notice.
Professional vocabularies determine what counts as a problem.

So this is not a new kind of power. It is an increase in its scope, resolution, and executability.

1. From records to decisions

Traditional systems mostly record completed activity. An AI harness can interpret current activity and initiate action.

A CRM might record that a deal received a discount. A harness can decide whether the discount is permitted, draft the approval request, route it, and block the quote until approval arrives.

2. From coarse visibility to process-level visibility

Managers now see outputs: revenue, cycle time, error rates, completed tickets. A harness can expose retrievals, intermediate classifications, policy checks, edits, approvals, and overrides.

That makes the production process itself an object of management.

3. From local interpretation to centralized execution

Today, policies are often deliberately incomplete. Local professionals reconcile them with circumstances. With a harness, one interpretation can be encoded and applied across thousands of cases.

The political battle therefore moves upstream. Instead of arguing over each decision, participants fight over the rule, category, prompt, evaluation set, or escalation threshold.

4. From organizational boundaries to system boundaries

A conventional outside law firm delivers advice, while the client retains its records and workflow. A harness provider may host the context, policy engine, audit trail, and learning loop. Operational control then crosses the legal boundary.

The provider is no longer just a vendor. It becomes part of the client’s administrative machinery.

5. From visible rules to hidden interaction effects

Paradoxically, more logging does not guarantee more intelligibility. A system can record every action while leaving people unable to understand why a result occurred. Complexity moves from human discretion into model behavior, retrieval logic, data provenance, and interacting rules.

The organization may become more observable to the provider and senior management while becoming less understandable to the employee or customer affected by a decision.

The key distinction

The strongest version of the argument is:

Organizations have always been governed through categories. AI makes those categories cheaper to enforce continuously, across organizational boundaries, and at the moment of action.

That creates a new class of politically important roles: ontology owners, policy engineers, evaluation designers, and exception governors. But the deepest power may belong to whoever can answer three questions:

Which distinctions does the system recognize?
Which outcomes train or update it?
Who gets to contest its classifications?

This connects directly to Bowker and Star’s Sorting Things Out, James C. Scott’s work on legibility, Michael Power’s The Audit Society, and research on algorithmic management. Their shared insight is that classification systems become most powerful when they stop looking like political choices and start looking like ordinary infrastructure.