# Gate 23 — minimum network cost

**date:** 2026-08-22  
**status:** candidate research  
**result:** OPEN — a small fixed-fee protocol layer is the preferred financing model

## Question

If FAF only coordinates identity, mapping, lineage and exchange protocols, how much shared organization is actually needed?

The danger is obvious:

> A network designed to keep local firms small could accidentally create a large central organization above them.

Gate 23 therefore treats the network itself as something that must pass a size discipline.

## Core principle

> **The network should be cheaper and smaller than the businesses it coordinates.**

That sounds obvious.

It should still be explicit.

## Minimum shared service

Candidate required layer:

1. node registry
2. product registry
3. public map
4. lineage store
5. credential / revocation service
6. public node pages
7. protocol documentation/versioning
8. basic support

That is the network core.

It does **not** require by default:

- central warehouse;
- central purchasing;
- central retail company;
- central payroll;
- national sales team;
- franchise field managers;
- property department;
- mandatory training school;
- corporate marketing department.

Those may be normal in another business model.

They are not required by FAF's current architecture.

## Fixed fee, not success tax

The cleanest current candidate is:

> **fixed transparent node fee**

rather than:
- percentage of gross sales;
- percentage of profit;
- compulsory equity;
- purchasing markup.

Reason:

If a node becomes very successful while using roughly the same registry/map/protocol infrastructure, the network should not automatically extract more merely because the node succeeded.

That would recreate the growth-extraction dynamic the project is explicitly trying to avoid.

## Generic cost sensitivity

Because FAF is jurisdiction-neutral, this Gate uses **generic currency units (CU)** rather than pretending one national cost base is universal.

### Monthly fixed fee per node

| Nodes | 60k CU annual shared cost | 120k | 240k | 480k |
|---:|---:|---:|---:|---:|
| 25 | 200 CU | 400 | 800 | 1,600 |
| 50 | 100 CU | 200 | 400 | 800 |
| 100 | 50 CU | 100 | 200 | 400 |
| 250 | 20 CU | 40 | 80 | 160 |
| 500 | 10 CU | 20 | 40 | 80 |

These are arithmetic only.

They show one useful thing:

> **A fixed shared-cost model becomes very cheap per node if the protocol layer stays genuinely small.**

The opposite is also true.

If FAF builds a 480k-CU annual organization while only 25 nodes exist, the fee becomes 1,600 CU/month per node.

That would be a serious burden.

## Early-network problem

A small new network has poor scale.

Possible temporary solutions:

- founding grants;
- donated development;
- community seed capital;
- volunteer protocol development;
- temporary founder subsidy;
- higher early membership fee explicitly scheduled to fall;
- external research funding.

But those must be visible.

Do not pretend the first 20 nodes can fund a mature shared infrastructure at the same price as 500 nodes.

## No permanent subsidy illusion

A grant can help create the protocol.

It should not hide a network that can never cover its own recurring costs.

Candidate health question:

> **What is the recurring cost of keeping one more node in the network?**

If marginal node cost is low, fixed fees can fall with scale.

If every node requires hours of bespoke central administration, the protocol has failed to standardize enough.

## Shared support labour

A small network still needs humans.

The pack now models:

`shared monthly support = base support + per-node support`

Illustrative examples:

### 100 nodes
If:
- base support = 40 h/month;
- per-node support = 0.25 h/month;

then:
- total shared support = 65 h/month;
- about 0.41 full-time equivalent at 160 h/month.

### 500 nodes
Same assumptions:
- total = 165 h/month;
- about 1.03 FTE.

Again: these are not forecasts.

They show what the architecture is trying to achieve:

> **hundreds of nodes should not require hundreds of central staff.**

If they do, interoperability has become administration.

## Per-node support time is a crucial metric

Candidate metric:

`network support load = shared support hours / active nodes`

This should be tracked.

A node that constantly needs manual fixes may indicate:
- bad software;
- unclear protocol;
- training gap;
- legal complexity;
- unusual capability.

The answer should not automatically be more central headcount.

## Optional services

FAF may offer optional services separately:

- hosted software;
- accounting integration;
- shared procurement;
- label printing;
- transport coordination;
- training;
- insurance purchasing;
- design templates.

But optional services should remain:

> **optional**

A node must not have to buy ten bundled services merely to retain protocol membership.

This is how a protocol quietly becomes a franchise.

## No mandatory procurement margin

A particularly dangerous path is:

`FAF negotiates all ingredients -> takes margin -> forces nodes to buy through FAF`

That creates:
- central purchasing power;
- upstream concentration;
- network dependency;
- a revenue incentive to force more trade through headquarters.

Current red line:

> **No mandatory centralized purchasing.**

Voluntary group purchasing is different.

## Credential service

The network needs some authority to issue/revoke FAF status.

That is one of the genuinely central functions.

But it should remain narrow:

- validate node qualification;
- issue credential;
- record current status;
- suspend/revoke under published rules;
- provide appeal/review process.

Credential authority must not become operational authority.

## Registry and map

The registry should be cheap.

A node record is not a case manager.

The network should not need a staff member to manually maintain every product listing.

Nodes should be able to update their own:
- capabilities;
- public hours;
- products;
- current status;
- stock/availability where supported.

The registry validates identity/protocol rules.

It does not become a central content department.

## Lineage storage

The network may store lineage centrally, federate it, or use a hybrid.

No architecture is locked yet.

Hard requirement:

> **Node data must be exportable.**

If FAF shuts down, the local company should not lose its own product and batch history.

This also protects Gate 8's exit principle.

## Open protocol pressure

A strong anti-lock-in option is to publish:

- schemas;
- identifiers;
- transfer object format;
- export format;
- public API/spec where appropriate.

That would make it harder for FAF itself to become the only company capable of understanding FAF data.

Open protocol does not necessarily mean:
- no trademark;
- no credential;
- anyone may falsely claim active membership.

The mark/status can remain governed while the data grammar stays open.

## Network reserve

The shared organization also needs resilience.

Possible reserve needs:
- service outage;
- credential incident;
- legal dispute;
- security issue;
- staff transition;
- migration.

But the reserve should be tied to real shared operating costs.

Do not build an investment fund simply because cash accumulates.

## Network surplus

If fixed fees produce a large surplus, possible responses:

1. reduce next year's fee;
2. improve shared infrastructure;
3. build reserve to an agreed cap;
4. refund/credit nodes;
5. fund a specific member-approved shared project.

The default should not be:

> "Great, now headquarters can hire five more people."

## Anti-headquarters test

FAF is drifting away from its architecture if the central layer begins to own or control:

- properties;
- node equity;
- local hiring;
- local pricing;
- procurement;
- product assortment;
- retail operations;
- territorial development;
- growth targets.

At that point the "network" has become the company.

## Candidate financing constitution

### Required network layer
Funded by:
> fixed transparent node fee

### Optional shared services
Funded by:
> users of that service

### Large one-off protocol development
Possible:
> grants / founding capital / project-specific funding

### Not default
- gross-sales royalty
- mandatory equity
- mandatory purchasing margin
- regional franchise fee

## Gate result

**OPEN.**

### Strong current answer

FAF should try to make the central layer:

- technically capable;
- administratively small;
- financially legible;
- cheap per node;
- easy to exit.

### Strong candidate metric

`network support load = shared support hours / active nodes`

### Strong red line

> **The network should not become the largest company in the network.**

## What is missing

- actual hosting/storage cost;
- credential/signature implementation;
- legal/compliance staff needs;
- security operations;
- support burden;
- insurance;
- governance cost;
- audit/verification cost;
- cross-country complexity;
- real node count.

## Next Gate — verification without bureaucracy

Gate 24 should ask:

> **How does FAF verify that a node still meets protocol claims without creating inspectors, paperwork and headquarters staff everywhere?**

Separate:
- self-attestation;
- machine-readable evidence;
- peer review;
- customer-visible correction;
- periodic audit;
- serious breach handling.

The goal is not zero verification.

The goal is **verification proportional to risk**.
