Synthocracy: The Theoretical Foundations of the New Market Architecture. From the Market as a Place of Exchange to the Market as an Operating System

A foundational essay for the Synthocracy Institute

Abstract

The market has traditionally been imagined as a place, mechanism or institutional order in which buyers and sellers meet, exchange information, negotiate prices and transfer goods, services and rights. Even after commerce became digital, this basic image survived. Websites, marketplaces, payment platforms and enterprise systems were treated primarily as tools that made existing markets faster, larger and more efficient.

The rise of autonomous AI agents changes the structure of this relationship.

In an agentic market, computational systems do not merely display offers or process instructions entered by humans. They can identify needs, discover counterparties, determine eligibility, interpret specifications, rank alternatives, negotiate conditions, reserve resources, authorize payments and initiate execution. The market therefore begins to resemble an operating system: a layered environment composed of identities, permissions, access classes, protocols, registries, schedulers, system calls, administrators and execution rules.

This does not mean that the global economy will literally run on one technical operating system. The more defensible claim is functional. Economic action will increasingly depend on a distributed stack of infrastructures that determines which entities are recognizable, which actions are permitted, which options become visible, which transactions can be executed and which participants are routed into delay, review or exclusion.

Within Martin Novak’s framework, this is a central expression of market synthocracy: a decision order in which humans remain formally responsible while synthetic systems increasingly prepare the environment in which economic decisions become possible. Novak’s key claim is that power migrates upstream—from the visible moment of decision to the prior layers of routing, ranking, classification, scheduling, permission and admissibility.

The central thesis of this essay is therefore:

The future market will not merely be accessed through software. It will increasingly be constituted by software-mediated conditions of identity, admissibility and execution.

When this transition is complete, the decisive market question will no longer be only who owns, buys, sells or sets the price. It will be:

Who controls the operating layer that determines who may become visible, actionable, negotiable and executable?

1. Conceptual starting point: what synthocracy adds to market theory

Within Martin Novak’s contemporary framework, synthocracy describes a condition in which humans continue formally to govern, manage, approve or assume responsibility, while AI systems increasingly classify, filter, prioritize, recommend and prepare the reality on which those human decisions operate. The concept was developed within the work of Martin Novak and the Novakian Paradigm Institute as a response to the migration of decision-making power into models, platforms, data infrastructures and agentic systems.

This definition shifts attention away from a narrow question:

Did an AI system make the final decision?

It directs attention toward an earlier sequence:

Which options entered the process?
Who or what classified them?
Which actors were made visible?
Which evidence was considered admissible?
Which risks were amplified?
Which alternatives were delayed or removed?
Which default action was prepared?
Which human later confirmed the result?

Novak calls the total field in which these processes occur the decision environment. It includes visible options, invisible filters, rankings, classifications, defaults, timing pressures, predictive scores and machine-generated summaries. When routing, ranking, prediction and access converge, the decision environment becomes a governing field even where a person still performs the final formal act.

Applied to commerce, the implication is profound.

The economic power of an AI system does not begin when it places an order. It begins when it helps define:

what counts as a product;
what counts as a qualified supplier;
what counts as an acceptable risk;
what counts as relevant evidence;
which participants are searchable;
which offer is shown first;
which transaction is permitted to proceed.

Market synthocracy therefore refers not simply to automated trading or machine purchasing. It describes the migration of market-making power into the architecture that prepares economic choices.

2. Four ways of imagining the market

To understand why the operating-system model matters, it is useful to distinguish four historical images of the market.

2.1 The market as a place

The oldest image is spatial.

The market is where buyers and sellers meet. It may be:

a town square;
an exchange;
a trade fair;
an auction room;
a wholesale district;
a commercial port.

Access is mainly physical and institutional. The participant must be present, recognized and able to exchange.

2.2 The market as a mechanism

Modern economics increasingly describes the market as a price-coordination mechanism.

Participants send signals through supply, demand, bids, offers and transactions. Prices aggregate dispersed information and influence allocation.

The market need not have one physical location. It is a mechanism of coordination.

2.3 The market as a platform

The digital economy introduced a third image.

A platform connects multiple groups:

buyers and sellers;
drivers and passengers;
advertisers and audiences;
developers and users;
merchants and payment providers.

The platform does not merely transmit information. It organizes discovery, reputation, ranking, payments, access and contractual conditions. The European Commission has explicitly recognized that large platforms can become private gatekeepers to markets, customers and information. The Digital Markets Act consequently treats certain platforms as important gateways capable of creating bottlenecks between business users and end users.

2.4 The market as an operating system

The agentic market extends the platform model.

An operating system does not only connect users with applications. It:

recognizes identities;
allocates resources;
grants permissions;
schedules processes;
manages memory and state;
exposes system calls;
monitors execution;
handles failures;
determines which operations are valid;
separates user-level action from privileged administration.

The future market will increasingly perform analogous functions.

It will identify firms and agents, grant or deny commercial capabilities, prioritize requests, reserve inventory, authorize transactions, manage credentials, invoke payment systems and determine whether an action can move from intention into execution.

The operating-system thesis does not claim that economic life becomes identical to computer engineering. It claims that market participation increasingly depends on functional layers that resemble operating-system functions.

3. The digital market and the agentic market are not the same

The digital market is primarily designed for human-directed interaction.

The agentic market is increasingly designed for delegated machine action.

The digital market

A typical digital transaction works as follows:

A person searches for a product.
A platform displays options.
The person compares offers.
The person selects an item.
The person enters or confirms payment.
Software processes the decision.

The platform may influence the person through ranking, recommendations and personalization. Nevertheless, the human usually performs the main sequence of navigation and commitment.

The agentic market

A typical agentic transaction may work differently:

A system detects a need.
A buyer agent interprets the requirement.
The agent discovers potential suppliers or supplier agents.
Identity and authority are verified.
The agents exchange structured data.
Suppliers are evaluated against encoded policies.
Negotiation occurs within delegated boundaries.
Inventory or capacity is reserved.
A payment or purchase mandate is checked.
An order is executed or escalated for approval.

Research on the agentic economy describes a future in which assistant agents represent buyers while service agents represent businesses and communicate programmatically. Crucially, that research distinguishes technically unscripted interaction from economically unrestricted interaction: agents may be capable of communicating freely while still being confined by platform architecture, commercial governance or walled gardens. The architecture of agent communication will therefore materially affect access to economic opportunity.

The distinction can be summarized as follows:

Digital market	Agentic market
Human initiates most actions	Systems may detect and initiate needs
Human reads interfaces	Agents read structured capabilities and policies
Search returns pages or listings	Discovery returns executable counterparties
Recommendations influence choice	Agents may define the candidate set
Human negotiates or accepts	Agents negotiate within mandates
Payment follows explicit human action	Payment can follow pre-authorized conditions
Software processes a decision	Software participates in forming the decision
Platform mediates interaction	Infrastructure may coordinate autonomous execution

The digital market mediates human commerce.

The agentic market increasingly performs commerce on delegated human or institutional authority.

4. When infrastructure stops supporting the market and begins constituting it

Not every commercial technology constitutes a market.

A spreadsheet may support a purchasing decision without defining the market. An email system can transmit quotations without determining who may trade. A warehouse database can report inventory without governing access to the supply network.

Infrastructure becomes constitutive when participation cannot be meaningfully understood or performed independently of it.

A market infrastructure becomes constitutive when it performs several of the following functions:

Identity formation
It determines what counts as a recognized participant.
Admissibility determination
It decides which firms, offers or actions may enter the process.
Discovery control
It determines who can find whom.
Classification authority
It defines how products, risks and capabilities are represented.
Permissioning
It determines which operations each participant may perform.
Scheduling
It controls priority, timing, routing and access to scarce attention or resources.
Execution
It can transform an intention into an order, payment, reservation or delivery.
Finality
It determines when a transaction becomes binding or irreversible.
Record creation
It defines what counts as evidence that a transaction or authorization occurred.
Remedy and exit
It determines whether a decision can be appealed, reversed or processed through an alternative channel.

This is not an entirely new phenomenon.

Payment systems, central counterparties, settlement systems and trade repositories already constitute essential financial-market infrastructure. International principles developed by the Bank for International Settlements and IOSCO treat their governance, access, resilience and risk controls as matters essential to financial stability.

The agentic transformation extends this constitutive logic beyond settlement.

The new infrastructure reaches upstream into:

discovery;
qualification;
interpretation;
negotiation;
authorization;
commercial routing.

The market is no longer only constituted at the point where money or title settles. It is increasingly constituted at the point where a possible counterparty becomes eligible for machine-mediated action.

5. The market operating system as a distributed stack

The future market operating system should not be imagined as one application owned by one institution.

It is better understood as a distributed stack.

Layer 1: Physical and computational infrastructure

This includes:

chips;
networks;
data centers;
cloud platforms;
energy;
model-serving infrastructure.

These layers determine who has access to computational capacity and under what economic conditions. Novak’s work on control over AI emphasizes that power can concentrate at chokepoints such as chips, cloud compute, frontier models and distribution. A firm does not need to own every layer to exercise infrastructural power; control over a layer that others cannot practically route around may be sufficient.

Layer 2: Models and agent runtimes

This layer includes:

foundation models;
planning systems;
tool routers;
memory;
verification systems;
agent frameworks;
orchestration environments.

It provides the synthetic capacity to interpret goals and coordinate action.

Layer 3: Identity and credential infrastructure

This layer answers:

Which agent is communicating?
Which organization does it represent?
Is its identity current?
Which credentials does it possess?
Can those credentials be verified?
Has its authorization expired or been revoked?

NIST’s work on software and AI-agent identity recognizes that reliable agent adoption will require standards-based methods for identifying agents, managing them and authorizing their actions.

Layer 4: Communication protocols

Protocols determine how agents discover one another, communicate, exchange tasks and report status.

The A2A Protocol is explicitly designed to enable communication and interoperability between independent agent systems built by different vendors or frameworks. It supports discovery, capability declaration, secure interaction, long-running tasks and human-in-the-loop processes.

Layer 5: Domain and commerce protocols

General agent communication is not enough.

Markets require shared semantics for:

products;
quantities;
prices;
contracts;
delivery;
negotiation;
payment;
returns;
evidence.

Universal Commerce Protocol is intended to provide vendor-neutral commerce primitives that can operate across different agentic surfaces. Google’s reference implementation demonstrates how a common protocol can connect discovery directly with transaction execution for eligible merchants.

Layer 6: Mandates and delegated authority

An agent may be technically capable of buying without being legitimately authorized to buy.

Agent Payments Protocol introduces signed mandates that can specify limits such as price, timing and transaction conditions. These mandates are intended to provide verifiable evidence that an agent was pre-authorized to act within defined boundaries.

Layer 7: Registries and discovery systems

Registries may contain:

agent identities;
capabilities;
certifications;
service regions;
reputation;
supported protocols;
contractual status;
access requirements.

A2A allows agents to describe their capabilities and security requirements through Agent Cards. Discovery systems may provide different information according to the identity and permissions of the requesting party.

Layer 8: Schedulers and routing systems

This layer determines:

which request is processed first;
which supplier sees an opportunity;
which case is escalated;
which transaction waits;
which offer expires;
which resources are reserved;
which human receives an alert.

Within Novak’s synthocracy framework, the scheduler is not a neutral technical detail. It is a governing layer because order and timing change outcomes. What appears first is more likely to be acted upon, while what appears too late may effectively cease to exist as an option.

Layer 9: Transaction and settlement infrastructure

This includes:

purchase orders;
smart contracts;
payments;
financing;
insurance;
shipping;
fulfillment;
settlement;
invoicing.

Layer 10: Audit, remedy and governance

This layer should determine:

what was recorded;
whether a decision can be reconstructed;
who acted under whose authority;
whether an affected party can appeal;
whether a transaction can be stopped;
who can modify the system.

The stack as a whole forms the functional equivalent of the market operating system.

6. The market kernel

An operating system has a kernel: the privileged layer that mediates access to system resources.

The agentic market will also have kernel-like functions.

The market kernel is the set of foundational rules and infrastructures that determines whether an economic intention may become an executable action.

It may contain:

identity validation;
authorization checks;
commercial policy;
risk constraints;
transaction limits;
product ontologies;
settlement rules;
mandatory evidence;
resource-allocation logic;
exception handling.

The kernel need not be controlled by one entity.

It may be distributed among:

enterprise systems;
cloud providers;
protocol organizations;
payment networks;
regulators;
certification bodies;
platform operators.

Yet the fragmented nature of control does not make the kernel politically neutral. It may make power more difficult to locate.

A procurement officer may believe that a supplier was rejected by “the system.” In reality, the outcome may result from the interaction of:

an industry classification;
a platform registry;
an external sanctions database;
a credit-risk API;
an internal purchasing policy;
an LLM interpretation;
a payment provider;
a logistics constraint.

No single component may have decided the outcome independently.

The stack nevertheless produced an executable refusal.

This is one reason synthocracy cannot be reduced to “rule by an AI.” The governing force can emerge from the synthesis of multiple systems.

7. Layer C and the commercial boundary of admissibility

Martin Novak’s Layer C provides a particularly useful theoretical bridge between synthocracy and market architecture.

Layer C is defined in the Novakian Paradigm as the pre-runtime layer in which admissibility is determined before a state, claim, signal or decision becomes an input to an executing system. It is not a software layer, but a conceptual and disciplinary position: the point at which the question is not yet how an action should be governed, but whether it should enter the execution field at all.

Applied to markets, Layer C becomes the commercial boundary of admissibility.

Before an agent can compare, negotiate or purchase, the system may decide:

Is the supplier recognized?
Is the product correctly classified?
Is the data sufficiently complete?
Is the certification acceptable?
Is the jurisdiction permitted?
Is the agent trusted?
Is the buyer authorized?
Is the transaction type allowed?
Is the risk score below the threshold?
Is there sufficient evidence to proceed?

These are not transaction-execution questions.

They determine what is permitted to approach execution.

This produces a critical distinction:

Runtime market governance

Runtime governance controls actions already admitted to the market process:

maximum discount;
payment limit;
number of negotiation rounds;
transaction timeout;
approval threshold.

Pre-runtime market admissibility

Pre-runtime admissibility controls what may become part of the process:

which suppliers exist for the agent;
which evidence is accepted;
which product definitions are legible;
which agents are trusted;
which possible transactions enter the executable space.

The second layer is more foundational.

A company rejected at runtime may receive a reason.

A company excluded before runtime may never know that an opportunity existed.

8. Conditions of execution

A participant in the future market will not gain access merely because it legally exists or offers a useful product.

It will need to satisfy a set of conditions of execution.

These are the technical, semantic, commercial and institutional conditions that must be met before an agent can perform an economically meaningful action involving that participant.

8.1 Identity condition

The system must recognize:

the firm;
its authorized agent;
relevant personnel;
ownership or representation relationships.

8.2 Legibility condition

The system must be able to interpret:

product identity;
service capabilities;
specifications;
quantities;
units;
prices;
constraints.

8.3 Protocol condition

The participant must support an accepted method of:

discovery;
authentication;
communication;
transaction exchange.

8.4 Authority condition

The agent must possess a valid mandate for the requested action.

Identity is not authority. An authenticated agent may be able to communicate while lacking legitimate permission to negotiate, reserve or purchase.

8.5 Evidence condition

The system may require:

certificates;
declarations;
licenses;
historical records;
financial data;
audit results;
provenance.

8.6 Risk condition

The participant must fall within defined thresholds for:

credit;
security;
compliance;
logistics;
operational resilience;
concentration risk.

8.7 Resource condition

The requested inventory, capacity, credit or transport must be available or reservable.

8.8 Settlement condition

The transaction must be capable of:

payment;
financing;
invoicing;
tax treatment;
final settlement.

8.9 Observability condition

The transaction may be required to generate sufficient records for:

audit;
monitoring;
regulatory reporting;
dispute resolution.

8.10 Remedy condition

A legitimate market architecture should provide mechanisms for:

correction;
appeal;
human review;
revocation;
reversal.

Together, these conditions define executable market participation.

A company may be economically real but operationally non-executable.

9. Economic permissions: the market equivalent of read, write and execute

Computer systems distinguish between different permissions. Economic systems will require a more granular permission model.

Read

The right to access information about:

products;
prices;
demand;
availability;
counterparties;
reputation;
transaction status.

Read permission may be selective. A supplier may expose public capabilities while reserving contract prices for authenticated buyers.

Write

The right to create or modify:

product records;
RFQs;
quotations;
delivery instructions;
commercial profiles;
transaction data.

Discover

The right to search registries and identify potential counterparties.

This may become one of the most valuable market permissions. A buyer agent unable to search beyond one platform’s approved network does not participate in an open market even if communication inside the network is technologically advanced.

Negotiate

The right to make:

offers;
counteroffers;
substitutions;
delivery proposals;
payment-term changes.

Negotiation permission should specify limits, duration and negotiable variables.

Reserve

The right to temporarily allocate:

inventory;
capacity;
transport;
credit;
production slots.

Reservation is economically consequential because it can deny resources to other market participants before a final purchase occurs.

Commit

The right to transform a negotiated result into a binding order or contract.

Pay

The right to authorize, initiate or confirm settlement.

Delegate

The right to transfer all or part of an assignment to another agent.

Delegation creates a chain of authority that must remain reconstructable.

Rank

The right to order participants or offers.

Ranking is often treated as an analytical function. In practice it is a market permission because it controls attention and probability of selection.

Route

The right to direct opportunities, requests or exceptions to particular participants.

Block

The right to stop a participant, transaction or class of activity.

Audit

The right to inspect records, models, rules and delegations.

Appeal

The right to challenge classification, ranking, refusal or execution.

Exit

The right to withdraw identity, data, reputation and commercial relationships from one infrastructure and move them elsewhere.

A central task for future market governance will be deciding which actors can assign, exercise and revoke these permissions.

10. The scheduler as an economic governor

Traditional accounts of market power emphasize ownership, price-setting and command.

Synthocracy reveals another form of power: scheduling.

A market scheduler determines:

when an RFQ is released;
who receives it first;
how long each party has to respond;
which request enters a priority queue;
which inventory is temporarily held;
when a negotiation ends;
when a human is asked to intervene;
which offer remains valid;
which exception waits.

This matters because time changes economic reality.

A carrier that sees a shipment first may secure it before competitors respond.

A supplier whose quotation is processed after an inventory reservation may appear unavailable even though it initially had stock.

A small company routed to manual review may answer too late for a machine-speed procurement process.

The scheduler can therefore produce exclusion without issuing an explicit prohibition.

This is synthetic routing power.

A system does not have to say:

This supplier may not participate.

It can achieve a similar outcome by deciding:

This supplier will be evaluated after the transaction window closes.

Novak’s scheduler concept captures this shift precisely: in fast systems, timing is governance.

11. Registries as the new market geography

Physical markets organize participants spatially.

Agentic markets organize them through registries.

A registry may define:

who exists;
what they can do;
where they operate;
which credentials they hold;
which protocols they support;
what their reputation is;
who may contact them.

The registry is therefore not merely an address book.

It is a map of the executable market.

An unregistered supplier may be equivalent to a shop located outside all known roads.

A registered but low-ranked supplier may exist only at the periphery.

A selectively disclosed registry may show different markets to different agents.

A2A discovery documentation already contemplates situations in which registries return different Agent Cards depending on a requesting party’s identity and permissions. This is technically sensible for confidentiality and security, but it demonstrates that market visibility can become permission-dependent.

The future market may therefore have no single public geography.

Each participant may encounter a different executable market depending on:

identity;
contract status;
geography;
risk classification;
platform membership;
commercial history;
technical capability.

12. Who becomes the administrator of the market?

A computer administrator can change permissions, install software, terminate processes and modify system configuration.

Who performs equivalent functions in an agentic market?

There may be no single administrator. Instead, administration will be distributed among several classes of actors.

Platform operators

They may control:

discovery;
ranking;
onboarding;
access conditions;
transaction interfaces;
dispute channels.

Protocol organizations

They define:

message structures;
capability declarations;
extension rules;
version compatibility;
official namespaces.

A2A, for example, maintains a governance process for official and experimental protocol extensions.

Identity providers

They determine:

how agents and organizations are authenticated;
which credentials are trusted;
when credentials expire;
how they are revoked.

Cloud and model providers

They determine:

access to compute;
model availability;
usage constraints;
safety policies;
regions and jurisdictions;
technical limits.

Enterprise administrators

Procurement, finance, compliance and IT teams define:

approved suppliers;
transaction limits;
negotiation policies;
escalation rules;
data access.

Payment networks

They determine:

whether settlement is possible;
what evidence is required;
how authorization is represented;
how disputes are handled.

Reputation and risk providers

They influence:

eligibility;
pricing;
credit;
insurance;
priority.

Regulators and courts

They determine:

legal admissibility;
competition constraints;
rights of appeal;
liability;
enforceability.

This creates a distributed administrative constitution.

The market may remain formally decentralized while its executable conditions are controlled by a relatively small number of infrastructural actors.

The current digital-platform experience provides a warning. OECD research on online marketplaces in Poland, Latvia and Lithuania identifies seller dependency, informational asymmetries, self-preferencing risks and structural factors that may entrench incumbent positions.

In the agentic market, these problems could extend from platform listings into automated eligibility and execution.

13. Protocols as the private law of execution

Lawrence Lessig’s foundational observation that code and architecture regulate behavior is highly relevant here. Software architecture can determine what is possible, who sees what and under what conditions access is granted.

In an agentic market, protocol rules may acquire a similar constitutional function.

They may determine:

what constitutes a valid identity;
what counts as an offer;
how a mandate is expressed;
when a reservation exists;
which errors terminate a process;
what evidence proves authorization;
which extensions are recognized.

The result is not that legal law disappears.

Rather, economic action is increasingly shaped simultaneously by:

public law;
private contracts;
organizational policy;
platform rules;
executable protocol logic.

The protocol becomes a form of pre-emptive commercial regulation.

Law may provide a remedy after an unfair exclusion.

The protocol may make the excluded transaction technically impossible in the first place.

This is why open technical standards do not automatically create open markets.

An open protocol may coexist with:

closed registries;
non-portable reputation;
proprietary extensions;
platform-specific certification;
asymmetric access;
restricted discovery.

The operating system may be open at the communication layer and concentrated at the administrative layer.

14. System calls in the market operating system

An application interacts with an operating system through system calls.

An agentic market will develop economic equivalents.

Examples include:

DISCOVER_SUPPLIER
VERIFY_IDENTITY
CHECK_AUTHORITY
REQUEST_QUOTE
SUBMIT_QUOTE
NEGOTIATE_TERMS
CHECK_COMPLIANCE
RESERVE_INVENTORY
RESERVE_CAPACITY
REQUEST_CREDIT
AUTHORIZE_PAYMENT
CREATE_ORDER
CONFIRM_ORDER
SCHEDULE_DELIVERY
RELEASE_RESERVATION
CANCEL_TRANSACTION
ESCALATE_TO_HUMAN
APPEAL_CLASSIFICATION

These calls transform economic intention into structured operations.

The system-call model matters because it distinguishes between:

what a participant wants;
what the infrastructure allows the participant to request;
what the infrastructure actually executes.

A company may want to negotiate a non-standard arrangement. If the system exposes no corresponding operation, the possibility may disappear from practical commerce.

The architecture thereby determines not only efficiency, but the range of commercially expressible actions.

15. The market as a state machine

Agentic commerce will increasingly represent commercial processes as explicit states.

A simplified transaction may move through:

NEED_DETECTED
→ IDENTITY_VERIFIED
→ MANDATE_VALIDATED
→ COUNTERPARTY_DISCOVERED
→ PARTICIPANT_ADMITTED
→ OFFER_REQUESTED
→ OFFER_RECEIVED
→ NEGOTIATION_ACTIVE
→ TERMS_ACCEPTED
→ RESOURCE_RESERVED
→ HUMAN_APPROVAL_REQUIRED
→ PAYMENT_AUTHORIZED
→ ORDER_COMMITTED
→ EXECUTION_MONITORED
→ SETTLED
→ CLOSED

Each transition requires conditions.

For example:

PARTICIPANT_ADMITTED may require valid certification.
RESOURCE_RESERVED may require credit approval.
ORDER_COMMITTED may require a human or institutional mandate.
SETTLED may require payment-network confirmation.

The market state machine makes commerce auditable and scalable.

It also creates a new type of exclusion.

A participant may be trapped in a state such as:

IDENTITY_PENDING
COMPLIANCE_REVIEW
INSUFFICIENT_EVIDENCE
MANUAL_EXCEPTION
RISK_BLOCKED

The participant remains formally inside the system but cannot progress toward execution.

This is an economic equivalent of process suspension.

16. Capability is not authority

A central principle in Novak’s synthocracy framework is that technical capability does not create legitimate authority.

A system may be able to:

select suppliers;
calculate prices;
negotiate;
predict risk;
place orders.

That ability does not by itself establish the right to perform those actions.

Capability answers:

Can the system do this?

Authority answers:

Under whose legitimate power is it doing this?

Novak argues that efficiency, accuracy or statistical superiority cannot independently confer decision-making legitimacy. Authority requires authorization, constraint and answerability to those affected.

This principle must become part of the market operating system.

Every consequential action should ideally be connected to:

an identified principal;
a defined mandate;
a scope;
a duration;
a purpose;
explicit limits;
a revocation mechanism;
a record of delegation.

Otherwise, the market may confuse a technically valid transaction with a legitimately authorized one.

17. The ceremonial human at the interface

The operating system may preserve a human approval screen.

That does not guarantee human control.

A human becomes ceremonial when the system has already:

selected the candidates;
removed alternatives;
negotiated the conditions;
classified the risk;
reserved resources;
prepared the justification;
established a default.

The person then clicks “approve.”

Novak calls this the ceremonial human: a person who carries formal responsibility while lacking the practical ability to reconstruct or redirect the decision. Meaningful oversight must exist before the options are compressed, not only after a system presents its preferred result.

In operating-system terms, the human may possess a user interface without possessing administrative permissions.

They can approve the visible operation.

They cannot alter the kernel that produced it.

18. Open market operating system versus hard market synthocracy

The operating-system model is not inherently authoritarian or anti-competitive.

Different architectures are possible.

18.1 Open agentic market

An open architecture would provide:

interoperable protocols;
portable identities;
multiple registries;
transparent access criteria;
contestable rankings;
meaningful human review;
non-discriminatory interfaces;
functional manual alternatives;
auditable authority chains;
practical switching.

18.2 Federated market architecture

Several ecosystems may interoperate while retaining distinct:

identities;
policies;
registries;
reputation systems;
transaction rules.

This could provide competition but impose high integration and switching costs.

18.3 Closed platform market

A platform may control:

identity;
discovery;
ranking;
negotiation;
payment;
fulfillment;
appeals.

External agents may technically connect but receive inferior access.

18.4 Hard market synthocracy

Hard market synthocracy appears when:

agentic infrastructure determines commercial visibility;
protocol compliance becomes a prerequisite for participation;
ranking and routing cannot be meaningfully challenged;
manual participation becomes economically unrealistic;
dominant operators control multiple administrative layers;
humans confirm outcomes without access to upstream decisions;
exit is theoretically possible but practically prohibitive.

Novak describes synthocracy as becoming a default operating system not through a single political decree but through accumulated adoption. Companies gain efficiency, users gain convenience and institutions gain administrative capacity until dependence becomes structural.

The transition is therefore likely to appear as modernization.

Only later may it become recognizable as constitutional change.

19. A test for determining when the market has become an operating system

A sector is moving toward the market-operating-system condition when several of the following statements are true.

Identity test

A participant cannot transact without a machine-verifiable identity.

Legibility test

A commercially valid offer is ignored unless represented in an accepted machine-readable form.

Discovery test

A small number of registries or platforms determine which counterparties are visible.

Permission test

Economic actions require granular digital permissions or signed mandates.

Scheduler test

Timing and routing systems materially determine market outcomes.

Execution test

Agents can reserve, commit or pay without contemporaneous human action.

Administrative test

Infrastructure operators can change participation conditions without negotiation with affected firms.

Portability test

Leaving the system causes loss of identity, reputation, history or customer access.

Contestability test

Participants cannot meaningfully challenge ranking, exclusion or routing.

Fallback test

Manual alternatives exist formally but cannot compete with machine-speed execution.

Accountability test

A transaction can be technically logged but its authority chain cannot be reconstructed.

Constitutional test

The most consequential market rules are embedded in technical architecture rather than openly debated institutional processes.

The more tests a market satisfies, the less adequate it becomes to describe its infrastructure as merely supportive.

20. The theoretical significance of the operating-system model

The model produces several important conclusions.

20.1 Market access becomes a technical-institutional status

Participation no longer follows simply from ownership, legal capacity or willingness to trade.

It follows from executable recognition.

20.2 Visibility becomes a permission

Being visible to an agent is not merely a marketing advantage. It may become the precondition for receiving demand.

20.3 Timing becomes governance

Scheduling, queuing and routing can distribute opportunity without explicit prohibition.

20.4 Classification becomes allocation

A product category, risk class or supplier status can determine access to capital, demand and execution.

20.5 Protocol design becomes institutional design

Technical choices affect economic inclusion, competition and appeal.

20.6 Market administration becomes distributed and difficult to locate

Power may be divided across models, clouds, registries, platforms, enterprise policies and payment networks.

20.7 Human responsibility may survive after human agency weakens

A person may remain legally accountable while the practical decision has moved into infrastructure.

20.8 Exclusion moves upstream

The most consequential refusal may occur before a participant enters the visible decision process.

21. A research programme for the Synthocracy Institute

The market-operating-system thesis creates a broad research agenda.

Market kernel studies

Identify the rules and infrastructures that determine execution in specific sectors.

Commercial admissibility studies

Examine how suppliers, products and transactions enter or fail to enter agentic processes.

Scheduler studies

Measure how timing, priority and routing distribute opportunity.

Registry governance

Study who controls agent and supplier registries, how entries are ranked and how errors are corrected.

Permission mapping

Document the economic rights granted to buyer agents, supplier agents, platforms and administrators.

Authority-chain reconstruction

Develop methods for determining who acted under whose mandate.

Manual fallback studies

Test whether non-agentic channels remain functionally usable.

Platform concentration

Map control over identity, models, cloud, discovery, payment and fulfillment.

Protocol constitutional analysis

Evaluate how protocol specifications distribute rights, duties and powers.

Market observability

Develop indicators for hidden exclusion, routing asymmetry, ceremonial oversight and synthetic path dependence.

Conclusion: the market after interface

The market of the agentic era will not cease to be a domain of exchange.

It will still contain:

prices;
contracts;
competition;
scarcity;
production;
demand;
risk.

But these elements will increasingly operate inside a prior computational and institutional architecture.

Before a buyer chooses, a system may determine what can be found.

Before a supplier competes, a registry may determine whether it exists.

Before an offer is evaluated, an ontology may determine whether it is understood.

Before negotiation begins, a mandate may determine what can be proposed.

Before an order is placed, a policy engine may determine whether execution is permitted.

Before a person approves, a scheduler may determine which reality reaches the screen.

This is the central theoretical movement from market to market operating system.

The market ceases to be only the place where economic decisions occur.

It becomes the architecture that determines which economic decisions are capable of occurring.

Within the synthocracy framework, that is where power moves: from the visible decision to the environment that prepares it; from command to scheduling; from final approval to pre-runtime admissibility; from formal ownership to control over layers that others cannot route around.

The ultimate question of the new market architecture is therefore not merely:

Who bought, sold or approved?

It is:

Who administered the conditions under which buying, selling and approving became executable?

The future politics of the market will be the politics of that operating layer.