# Gate 16 — Evidence & Publication Boundary

**Status: PASS ON CLAIM CLASSIFICATION / empirical upgrades OPEN**

Purpose: prevent the article from presenting design work, simulations, scenarios or external benchmarks as if they were observed outcomes from a functioning 100 Burger a Day restaurant.

The project now uses a mandatory evidence taxonomy.

---

# 1. Evidence classes

## OBSERVED
Directly measured in a physical 100BAD test or operating unit.

Examples later may include:
- actual patty cook time;
- actual finish-station seconds;
- actual fries yield;
- actual reserve factor;
- actual operator hours;
- actual customer wait;
- actual supplier quote.

**Publication language**
- “We observed…”
- “The pilot measured…”
- “In Kitchen Pilot 001…”

OBSERVED may only be used after a real test exists.

## MODELED
Derived from explicit assumptions, arithmetic, simulations or engineering targets.

Current examples:
- ~412.5 nominal premium patties/animal;
- ~10.4 combined operator hours/day reference;
- queue results from Dry Run 001;
- 44 min/week coordination budget;
- modeled 53 sec finish/front active time.

**Publication language**
- “The model estimates…”
- “Under the current assumptions…”
- “A simulated stress test suggests…”

Never:
- “The restaurant does…”
unless physically observed.

## EXTERNAL
Supported by outside sources, manufacturer data, regulation, published research or supplier quotations.

Examples:
- current Danish food-safety guidance;
- commercial fryer rated capacities;
- current carcass-price references;
- farm-shop retail benchmarks.

**Publication language**
- state the external fact narrowly;
- cite the source;
- do not extend the source beyond what it actually establishes.

## SCENARIO
A deliberate what-if input used to test viability.

Examples:
- +15% farmer premium;
- 99 DKK combo;
- S60 / S80 / S100 secondary-product realization;
- 150 / 200 / 250 DKK owner-hour value;
- 1:5 syrup dilution placeholder;
- five opening days/week.

**Publication language**
- “We test a scenario in which…”
- “As a design target…”
- “For sensitivity analysis…”

Scenario values must never be described as market facts.

## DESIGN
A normative project decision or protocol rule.

Examples:
- only 100 premium sale tokens/day;
- reserve units cannot become burger #101;
- operator/customer relationship should remain direct;
- no compulsory transaction royalty;
- human-made burger remains reference/gold-standard product.

**Publication language**
- “The design requires…”
- “The protocol proposes…”
- “We impose a cap of…”

A design decision is not empirical evidence.

## HYPOTHESIS
A falsifiable proposition not yet adequately tested.

Examples:
- local bakery will reduce total operator burden enough to justify delivered price;
- 2 operators can physically sustain 50 combos/hour without quality drift;
- tallow fries can be supplied from carcass fat balance;
- destination customers will tolerate the 100-cap model;
- a federated Commons can scale without recreating a platform toll booth.

**Publication language**
- “We hypothesize…”
- “This remains to be tested…”
- “The model predicts…”

## SPECULATIVE
A future extension not needed to establish the core paper.

Examples:
- autonomous burger machine;
- workplace machine node;
- after-hours vending of frozen patties;
- large Readable Market / DeepAsk demand layer.

These may appear in a discussion/future-work section only if they clarify the architecture.

They must not be used to rescue weak core economics.

---

# 2. Confidence labels

Every quantitative claim receives one confidence label:

- **A — measured/direct**
- **B — externally strong or tightly derived**
- **C — plausible engineering model**
- **D — rough proxy / early sensitivity**
- **E — speculative**

A claim can change class and confidence over time.

Example:

`50 combos/hour can be sustained by two operators`

Current:
- class: HYPOTHESIS / MODELED support
- confidence: C

After a successful physical 20-combo rush pilot:
- class: OBSERVED for the pilot scale
- confidence: B

After repeated 100-combo service days:
- class: OBSERVED
- confidence: A

---

# 3. Current publication-safe claims

The following can already appear in an article if correctly qualified.

## Core design
Safe:
> 100 Burger a Day is a design exercise for a capped 100-premium-combo-per-opening-day restaurant.

Class: DESIGN.

Safe:
> The cap is treated as a labour and inventory constraint rather than a marketing scarcity trick.

Class: DESIGN / interpretation.

Safe:
> The project uses the animal, rather than only the burger patty, as the economic unit of analysis.

Class: DESIGN / MODEL FRAME.

## Whole-animal model
Safe:
> Under the current carcass-routing assumptions, the model produces roughly 400 premium 140 g patties per reference animal.

Class: MODELED.

Do not write:
> One cow makes 412 premium burgers.

That sounds observationally precise.

Safe:
> The model tests whether secondary products can carry enough carcass value to support a higher farm-gate payment.

Class: HYPOTHESIS / MODELED.

Do not write:
> Whole-animal utilization makes the burger beef cheap.

Not yet demonstrated.

## Farmer economics
Safe:
> We test a farmer-premium scenario rather than assuming the restaurant should improve margins by lowering farm-gate prices.

Class: SCENARIO / DESIGN.

Do not write:
> Farmers earn 15% more under 100BAD.

No operating evidence exists.

## Service model
Safe:
> A simulated four-hour service day suggests that two operators have substantially more peak resilience than one.

Class: MODELED.

Safe:
> The reference design targets short bursts of 50 combos/hour.

Class: DESIGN TARGET.

Do not write:
> Two people can serve 50 burgers an hour.

Not physically validated.

## Finish/front
Safe:
> A micro-time model puts the reference finish/front workload near one minute of active work per combo, before physical validation.

Class: MODELED.

## HOT + fries
Safe:
> The required 7.5 kg/hour peak fries output is modest relative to the rated capacity of ordinary commercial fryers reviewed for the project.

Class: MODELED + EXTERNAL.

Do not write:
> Fries will not be a bottleneck.

That remains untested physically.

## Supply nodes
Safe:
> At five opening days, a fully sold-out unit implies approximately 500 buns and 75 kg of served fries per week.

Class: DERIVED / MODELED.

Safe:
> The protocol converts this readable demand into supplier interfaces rather than prescribing who must own production.

Class: DESIGN.

## Coordination
Safe:
> The reference coordination design budgets less than 45 operator minutes per week for routine supplier administration.

Class: DESIGN TARGET.

Do not write:
> Local suppliers only require 44 minutes of admin per week.

No real supplier network exists yet.

## Governance
Safe:
> The proposed governance separates ownership of the protocol from ownership of local restaurants, farms and supply nodes.

Class: DESIGN.

Safe:
> The design includes fork and exit rights to reduce the risk that the Commons layer becomes a conventional toll-taking platform.

Class: DESIGN / GOVERNANCE HYPOTHESIS.

Do not write:
> The governance prevents platform capture.

It has not been institutionally tested.

---

# 4. Claims that are NOT publication-ready as facts

Until further evidence exists, the article must not state as factual outcomes that:

- 99 DKK is commercially viable;
- one or two people can actually run the full day at the modeled hours;
- customers will travel for the destination format;
- the unit will sell out;
- the farmer will earn more;
- the farmer will prefer the proposed contract;
- 235 kg is the true edible yield of the future reference animal;
- 412.5 premium patties is a measured animal yield;
- the secondary-product channel will realize S60 value;
- tallow supply will always exceed fries demand;
- local suppliers can hit the coordination target;
- three days of training is sufficient;
- a second unit can launch in 30 days;
- a Commons legal structure will withstand capture;
- the protocol can scale without new coordination costs.

These remain models, scenarios, or hypotheses.

---

# 5. Article truth table

Every article paragraph containing a substantive claim should be traceable to one of four support types:

1. **Internal observation**
   - physical pilot / operating dataset.
2. **Internal model**
   - workpack calculation / simulation.
3. **External source**
   - cited public evidence.
4. **Normative proposal**
   - explicitly presented as a design choice.

If a sentence has none of these, it should be removed or reframed.

---

# 6. Quantitative claim rule

Every quantitative statement must answer:

- What is the unit?
- Is it observed, modeled, external, or scenario?
- What assumptions generate it?
- What version of the workpack produced it?
- Is there a sensitivity range?
- Can a reader reproduce it?

Precision must reflect evidence quality.

Bad:
> 412.5 burgers per animal.

Better:
> In the current routing model, one reference animal supports roughly 400 premium patties before restaurant-level QC and remake losses.

Best, after physical carcass work:
> Across N processed animals, observed usable premium blend yielded X–Y 140 g patties per animal.

---

# 7. Negative-result publication rule

A failed gate is publishable.

Examples:
- 99 DKK fails once labour is measured;
- premium farmer payment makes the combo too expensive;
- local bakery coordination is worse than commercial supply;
- 50/hour creates quality collapse;
- whole-animal secondary channels cannot absorb the volume.

The article should not be engineered to “save” 100BAD.

The research question survives a negative answer.

---

# 8. Article boundary

## Main article should focus on
- the 100-burger cap;
- whole-animal accounting;
- farmer-linked procurement;
- owner/operator time;
- 99 DKK as a tested target;
- secondary-product routing;
- local supply interfaces;
- open protocol / commons concept;
- what the model says;
- what remains unproven.

## Discussion/future work may include
- P4P integration;
- Readable Market capacity;
- DeepAsk demand signals;
- frozen-patty vending;
- workplace burger machine;
- broader open-business-unit replication.

These cannot dominate the article until the core food/business thesis is established.

---

# 9. Publication maturity levels

## P0 — Concept note
Allowed now.
Contains:
- problem;
- design;
- models;
- explicit unknowns.

## P1 — Engineering paper
Requires:
- Kitchen Pilot 001;
- measured time/yield/waste;
- product sensory results;
- local supplier quotes;
- revised economics.

## P2 — Operating case study
Requires:
- real public unit;
- multiple weeks of sales;
- actual utilization;
- operator-hour data;
- actual farmer/supplier economics.

## P3 — Replication paper
Requires:
- independent unit #2;
- founder-intervention log;
- replication economics;
- governance/coordination experience.

Current project maturity:
**P0, with a strong P1 protocol prepared.**

---

# 10. Gate 16 PASS criteria

PASS requires:

1. every major current quantitative claim has an evidence class;
2. scenario values are not presented as facts;
3. simulated operations are labeled simulated;
4. external facts are source-bound;
5. design choices are not presented as findings;
6. speculative extensions are moved out of the core argument;
7. negative results remain publishable;
8. article maturity is stated;
9. no claim depends on the restaurant already existing;
10. future empirical evidence has a defined path to upgrade claims.

## Gate result

**PASS.**

The workpack can now support drafting without confusing an engineered system with an observed business.

## Next gate — Gate 17: Article Architecture

Build the actual article structure:
- title candidates;
- abstract thesis;
- section order;
- figures/tables;
- claim placement;
- explicit “what we know / what we model / what we still need to test” language.

The article should read as serious applied research, not as a franchise pitch.


---

# Gate 25.1 publication reconciliation

Publication-safe distinction:

> The carcass-routing model yields about 412 nominal 140 g patties per reference animal. For planning, a provisional 1.03 reserve/variance factor reduces this to roughly 400 saleable burger-equivalents per animal.

At five sold-out opening days this implies approximately:
- 1.25 animals/week;
- 221 kg/week modeled secondary edible output.

K1 may be described only as:

> Under the current illustrative food-cost, labour and 7,500 DKK/week overhead placeholder, the desk model crosses break-even at roughly 70 sales/day. This is a scenario result, not observed viability.