# Strategic Cost Calibration Model

**A Methodology for the Cost Asymmetry Underlying Orange Anchor Commitments**

*Version 1.3 - Working Document*  
*May 2026*

---

> **Core invariant.** Per-commitment production cost is positive and scales linearly with volume up to operational constants, under a stated, adversary-favourable threat model. The architectural contribution is cost preservation and attribution: production cost paid by a holder persists as attributable scarcity across every system that recognises the commitment. See *Orange Anchor White Paper v2.3* for the full argument.

---

## Purpose and Scope

This document is a methodology, not a results publication.

The Orange Anchor white paper establishes a structural claim: under the stated threat model, per-commitment production cost is positive and scales linearly with volume. This document supplies the operational framework for that claim by defining the cost levers exposed by the construction, the equations that compose honest and adversary marginal costs from those levers, the named calibration profiles that bundle lever settings into deployment configurations, and the cost ratio κ as an inspection metric evaluated against a declared adversary class. It also provides one fully derived illustrative example under explicitly stated assumptions.

The cost asymmetry rests on three structural properties of the construction, each independently asserted in the white paper:

1. **Time-binding to Bitcoin’s block sequence** imposes a wall-clock floor that capital cannot compress.
2. **Memory-hard computation** caps per-server concurrency at a ceiling set by physical memory bandwidth.
3. **Per-instance physical-state witnessing** forces per-instance simulation cost on virtualised adversaries.

Each property contributes a positive component to the per-commitment cost floor. So long as each component remains positive, total cost is positive and grows linearly in N up to operational constants. The architectural claim is the structural one. This document characterises how the levers controlling each component are set and combined.

The positive per-commitment cost floor, when enforced through time-binding to Bitcoin’s block sequence and memory-hard concurrency limits, is intended to raise the marginal cost of large-scale automated participation for the class of production whose viability depends on near-zero marginal cost per object. Whether any given deployment crosses a specific economic-rationality threshold is a calibration and snapshot question deferred to the Calibration Annex. The construction does not eliminate all automated activity, nor does it address high-value or well-capitalised use cases whose margins can absorb a positive cost floor.

The model treats calibration as a deliberate, versioned act whose purpose is to produce a bounded, inspectable level of economic friction. Absolute magnitude varies with hardware, fees, and energy prices. The structural property does not.

This document sits between the architectural claim in the *Orange Anchor White Paper v2.3* and the construction-level specification in *BACC v1.9* (the *Bitcoin-Anchored Collateral Commitments* construction paper): it operationalises the white paper's cost-positivity claim using the levers and parameters that BACC v1.9 defines. Terminology used here — *burn*, *envelope*, burn-tier names, and calibration-profile names — follows *Orange Anchor Lexicon v2.7*.

**In scope**  
- Lever catalogue and directional effect characterisation  
- Cost-composition equations and key structural observations  
- Named calibration profiles with qualitative κ characterisation  
- One fully derived illustrative example  
- Sensitivity discussion for highest-leverage parameters

**Out of scope**  
- Cryptographic reductions or formal proofs (deferred to the *Bitcoin-Anchored Collateral Commitments* construction paper)  
- Predictive numerical κ values for deployment (deferred to the Calibration Annex)  
- Verifier-side policy, scoring, or weighting  
- Full threat modelling beyond the Reference Concurrent Adversary defined here

---

## 1. Reference Classes and the Cost Ratio κ

### 1.1 The cost ratio κ

κ is defined as:

> **κ = (adversary marginal cost per commitment) / (honest marginal cost per commitment)**  
> *evaluated under an explicitly declared adversary class, honest class, calibration profile, hardware-cost snapshot, and fee-market snapshot.*

κ is an inspection metric, not a security parameter. The architectural defence is the *positivity* of the per-commitment cost floor; κ characterises the *magnitude of the asymmetry* under stated conditions. A profile remains architecturally valid so long as honest cost remains practical and adversary cost remains positive; the specific ratio between them is a deployment-tuning concern.

### 1.2 Reference classes

**Honest reference class.** Commodity end-user hardware operating the commitment as a background process during otherwise-idle time. The reference instantiation in this version is a mid-range smartphone of approximately 2026 capability. Foreground use, battery-constrained operation, thermal-constrained operation, and other operational regimes are scoped separately in the *Bitcoin-Anchored Collateral Commitments* construction paper.

**Reference Concurrent Adversary (RCA-2026).** An adversary operating with substantial capital, current best-bin general-purpose hardware plus available acceleration, virtualisation tooling, and engineering depth sufficient to attempt sensor-coherence simulation. All adversary classes considered in this version are bounded below RCA-2026 in capability and therefore in compression-ratio achievability; potentially more capable classes (state-level actor, cartelised operator, hardware-vendor insider) are out of scope here and listed as open items in §11.

**Snapshot.** All hardware-cost assumptions in this version are anchored to a parameter set dated *May 2026*, drawn from publicly available cost references for memory, energy, and Bitcoin fee market conditions. The snapshot itself is documented in the Calibration Annex (forthcoming). Recalibration is expected at the cadence specified in the *Bitcoin-Anchored Collateral Commitments* construction paper.

---

## 2. Cost Lever Catalogue

The construction exposes a finite set of levers across six axes. For each lever, the model characterises the directional effect on honest marginal cost and on RCA-2026 marginal cost. Specific magnitudes are calibration choices; the *direction and shape* of the effects are structural.

### 2.1 Resource axis

| Lever | Effect on honest cost | Effect on adversary cost | Leverage |
|---|---|---|---|
| Memory commitment per instance | Near-zero marginal (device RAM is present) | Proportional to RAM cost × concurrency | High |
| CPU intensity / energy expenditure | Marginal (background process) | Proportional to power and silicon time | Medium |

### 2.2 Time axis

| Lever | Effect on honest cost | Effect on adversary cost | Leverage |
|---|---|---|---|
| Burn duration (the active commitment interval; equivalent to the *interval duration* term used in *BACC v1.9*, and to the *burn* / burn-tier terminology in *Orange Anchor Lexicon v2.7*) | Background, near-zero perceived | Wall-clock floor for active capacity | Very high |
| Anchor schedule shape | Marginal (fees) | Imposes non-compressible windows | Very high |
| Bitcoin reseed frequency | Marginal | Forces real-time block-hash consumption | High |

Burn duration is selected from a fixed set of tiers defined in the Lexicon: a *lightweight* tier of 6 blocks (~1 hour) and a *heavyweight* tier of 144 blocks (~24 hours). The calibration profiles in §5 are stated against the heavyweight tier as the reference choice for κ characterisation; the cost-positivity property holds at either tier, with the lightweight tier compressing the wall-clock floor proportionally. Operators and verifiers may treat tier choice as a policy-level signal under their own methodology (e.g. via the audit-operator BARU framework in *Orange Pages Audit Methodology v1.0*), but this is outside the cost model itself.

### 2.3 Witness axis

| Lever | Effect on honest cost | Effect on adversary cost | Leverage |
|---|---|---|---|
| Sensor channel count | Near-zero (sensors present) | Per-channel simulation engineering and runtime | Very high |
| Sampling rate per channel | Near-zero | Multiplies coherence-simulation cost | High |
| Block-triggered burst density | Background | Forces coherent real-time response | High |

### 2.4 Cryptographic axis

| Lever | Effect on honest cost | Effect on adversary cost | Leverage |
|---|---|---|---|
| Sequential function iteration count | Background | Sets compression-ratio ceiling | Foundational |
| Memory-hard checkpoint frequency | Background | Forces sustained memory pressure | High |

Specific cryptographic primitives, parameter values, and the analytical basis for compression bounds are specified in the *Bitcoin-Anchored Collateral Commitments* construction paper. This model treats those primitives as black-box cost contributors whose parameters set the resource-axis demands.

### 2.5 Network axis (opt-in)

| Lever | Effect on honest cost | Effect on adversary cost | Leverage |
|---|---|---|---|
| Real-time public publication of in-interval state to public watchers | Marginal | Forces real-time wall-clock pace during publication windows | High when engaged |

The network axis is opt-in. It functions as a voluntary quality signal for honest producers and a mandatory cost increaser for adversaries seeking to match the resulting audit score.

### 2.6 On-chain axis

| Lever | Effect on honest cost | Effect on adversary cost | Leverage |
|---|---|---|---|
| Anchor count and fee tier | Direct fee cost (potentially batched) | Direct fee cost (batching less available at scale) | Medium-high |
| Batching operator usage | Substantially reduces fee component | Limited applicability at adversary scale | Structural enabler (see §4) |

---

## 3. Cost Equations

### 3.1 Honest marginal cost per commitment

> **C_honest = E_device + F_anchor_net + W_optional**

where:
- **E_device** = energy expenditure during the burn on the producing device (background process)
- **F_anchor_net** = anchor fees per commitment after batching
- **W_optional** = voluntary opt-in network-witnessing layer fees, where engaged

### 3.2 Adversary marginal cost per commitment (RCA-2026)

> **C_adv = (Capex_amort × T_active) + P_power + F_anchor_adv + S_simulation + R_detection**

where:
- **Capex_amort** = amortised capital cost per unit time on adversary hardware
- **T_active** = active wall-clock time per commitment = max(T_burn / compression_ratio, T_floor)
- **T_floor** = non-compressible wall-clock floor imposed by anchor schedule and any engaged opt-in layers
- **P_power** = power cost during active time
- **F_anchor_adv** = anchor fees at adversary’s available batching ratio
- **S_simulation** = sensor-coherence simulation cost (engineering amortisation + per-instance runtime)
- **R_detection** = detection-risk premium

*Honesty note on S_simulation.* The term **S_simulation** carries positive weight in the architectural argument: a virtualised attacker simulating a coherent multi-channel sensor profile pays engineering and runtime overhead that a real device does not. The *direction* of that overhead is structural (virtualisation has cost; bare metal does not). The *numerical value* of `S_simulation` against current generative-simulation tooling has **no published empirical grounding** in this document; it is deliberately left as a named parameter rather than asserted as a calibrated quantity. For purposes of the cost equations and the qualitative profiles in §5, `S_simulation > 0` is asserted structurally; its magnitude is published in the Calibration Annex (forthcoming) once empirical evaluation against representative attacker tooling is completed. A reader sceptical of `S_simulation` can set it to zero and re-evaluate the equations under the remaining levers; the per-instance cost floor remains positive under all profiles with `S_simulation = 0`, but the magnitude of κ is correspondingly reduced.

*Note on compression ratio:* The compression ratio is bounded above by the slowest of sequential-function pace, memory-bandwidth saturation under per-instance working set, and checkpoint cadence. The specific bound under current best-bin hardware is developed empirically in the Calibration Annex (forthcoming) and is not asserted here.

### 3.3 Cost ratio

> **κ = C_adv / C_honest**

evaluated under the declared profile, snapshot, and reference classes.

### 3.4 Key structural observations

Three observations follow directly from the cost equations regardless of the specific numerical values assigned to any parameter. These observations are the structural output of the methodology and hold across snapshot updates.

1. **Honest cost is dominated by the amortised anchor-fee term** when batching is available; the device-energy term is marginal. Honest cost is therefore sensitive to the fee market and to batching availability.

2. **Adversary cost at scale is dominated by (Capex_amort × T_active) and by sensor-coherence simulation cost *S***; the fee component is a smaller share. Adversary cost is therefore sensitive primarily to wall-clock active time and to the engineering economics of coherence simulation.

3. **κ is primarily a function of T_active (compression ratio and T_floor), *S*, and batching asymmetry** - not primarily a function of fees, energy, or any single resource in isolation. The levers that move κ are the ones that move T_active and *S*.

These observations are snapshot-independent. The specific κ value derived from any particular numerical inputs moves with the snapshot; the structural shape of the asymmetry does not.

These observations establish that the dominant components of adversary cost at scale - active wall-clock time under the anchor schedule and sensor-coherence simulation *S* - scale with the number of commitments rather than being amortised. Any automated production model whose current economics depend on near-zero marginal cost per additional object therefore faces a structural increase in cost that grows linearly with volume.

---

## 4. Structural Dependence on Batching Operators

The construction produces a positive cost floor without any batching layer. **The architectural cost-positivity claim in the white paper does not depend on batching operators.**

However, the *magnitude* of κ in higher-cost-asymmetry profiles is substantially shaped by batching availability. Batching operators aggregate many commitments into single Bitcoin transactions, reducing honest fee cost while reducing adversary fee cost less proportionally - an adversary producing at scale has weaker batching gains because commitments cluster in time and operator behaviour may be policy-constrained against them.

Where batching is unavailable, restricted, or censored, honest cost rises and κ in fee-dominated profiles compresses accordingly.

The white paper’s structural claim - positive cost, linear in volume - is preserved with or without batching. What changes without batching is the *magnitude of asymmetry*, not the *existence of asymmetry*. A deployment that wishes to operate without batching dependency selects a profile whose κ does not rely on a fee-dominated honest cost component.

This dependency is disclosed here in full so it cannot be claimed as a hidden centralisation vector.

---

## 5. Calibration Profiles

> **All numerical κ characterisations in this section are CALIBRATION-PENDING.** The qualitative tiers (Low / Moderate / High / Very high) are structural claims derived from the lever catalogue (§2) and cost equations (§3). Specific numerical κ ranges, snapshot anchoring, and sensitivity analysis are deferred to the *Calibration Annex* (forthcoming), which will publish preliminary κ ranges under explicitly stated honest-class, adversary-class, and snapshot assumptions. Until the Annex is published, deployments SHOULD treat the qualitative tier as the calibration commitment of this document and SHOULD NOT cite numerical κ values derived from earlier drafts as authoritative.

A calibration profile is a named, versioned bundle of lever values producing a documented qualitative characterisation of κ against the declared adversary class.

This version publishes profile *definitions and qualitative κ characterisations* only. Specific numerical κ values for production deployment are deferred to the Calibration Annex, where they will be presented with full derivation, snapshot anchoring, and sensitivity analysis.

| Profile | Lever configuration | Qualitative κ vs RCA-2026 |
|---|---|---|
| **Light** | Short burn, minimal anchors, standard sensors | Low |
| **Standard** | Day-scale burn, uniform anchors, standard sensors | Moderate |
| **Hardened** | Day-scale burn, tapered anchor schedule with dense final window, full sensor set | High |
| **Maximum** | Day-scale burn, tapered + dense final window, full sensor set, opt-in network witnessing engaged | Very high |

*Batching note:* The qualitative κ characterisations above assume batching is available. When batching is unavailable, honest anchor-fee cost rises and κ regresses towards the next lower profile tier. Qualitatively, Light is least affected; Standard shows limited regression; Hardened can regress towards Standard; Maximum is expected to regress most strongly, with exact magnitudes deferred to the Calibration Annex. The construction remains cost-positive under all conditions; only the magnitude of asymmetry changes (see §4).

Profile selection is a strategic decision by the producer or operator. The model provides the structural characterisation; the Calibration Annex provides the empirical magnitudes.

---

## 6. Strategic Combinations

Three lever combinations exhibit superadditive cost asymmetry - the combined effect on adversary cost exceeds the sum of the individual lever effects, while combined honest cost remains marginal.

**Tapered anchor schedule + dense final window.** Placing anchors sparsely early and densely near the end forces the adversary to operate at honest wall-clock pace during the final dense window, regardless of compression achieved earlier. This is currently the highest-leverage structural combination.

**Sensor coherence + sustained memory pressure.** Coherent multi-channel sensor simulation under sustained memory-bandwidth load forces the adversary to simulate physical side-effects of a workload they are simultaneously trying not to perform. The two requirements interfere with each other on adversary hardware in a way they do not on a real device performing real work.

**Opt-in network witnessing + dense final window.** Real-time publication during a dense window pins the adversary to wall-clock pace during the most cost-sensitive interval of the commitment.

These combinations hold across hardware generations because they exploit relationships between resources, not absolute magnitudes of any one resource.

---

## 7. Economic Implications

The lever catalogue (§2), cost equations (§3), batching dependence (§4), profiles (§5), and strategic combinations (§6) compose to produce a set of deployment-relevant implications. Each is a direct consequence of the cost structure characterised above and holds across snapshot updates and hardware generations. None is a calibration parameter; none depends on any specific numerical value of κ.

**Cost-curve shape.** Restating the consequence of §3.4 in deployment-relevant terms: total production cost remains linearly proportional to N because per-commitment cost remains positive up to operational constants. Any automated production model whose current economics depend on near-zero marginal cost per additional object is structurally incompatible with the construction at any positive calibration.

**Capital does not compress per-instance cost.** Additional capital expended by a concurrent producer increases the number of commitments produced; it does not reduce the per-instance cost floor. The floor is set by physical properties - wall-clock time bound to Bitcoin’s block sequence, memory bandwidth under the per-instance working set, and per-instance sensor coherence - not by software headroom that capital can buy.

**The total cost gap widens with N.** Under any profile, the honest reference class incurs near-marginal device cost plus amortised anchor fees. Adversary cost is dominated by Capex_amort × T_active and by per-instance sensor-coherence simulation *S*, both of which scale linearly with the number of commitments. The total cost gap between honest and adversary production therefore grows without bound as N increases, rather than remaining a fixed offset; κ itself remains a profile- and snapshot-dependent inspection metric rather than a quantity shown here to diverge with N.

**Profile selection trades the magnitude of asymmetry, not its existence.** The Light, Standard, Hardened, and Maximum profiles bundle lever settings to deliver progressively higher κ. The construction remains cost-positive across all profiles, and under all batching conditions (§4); profile choice determines the magnitude of asymmetry, not whether asymmetry exists.

**Asymmetry holds across hardware generations.** The cost asymmetry depends on relationships between resources - time-binding versus capital, memory bandwidth versus concurrency, per-instance witnessing versus virtualisation - not on absolute magnitudes of any single resource. Honest and adversary capability scale symmetrically as hardware improves; the structural asymmetry is preserved.

**Recalibration is an operational discipline, not a structural correction.** Snapshot-anchored κ values move with hardware and fee-market changes; the structural properties that produce a positive cost floor do not. Recalibration adjusts the magnitude of inspected asymmetry under updated conditions; it does not modify the architectural claim made in the white paper.

---

## 8. Threat Adaptation and Recalibration

The model treats κ as a function of the hardware snapshot, fee-market snapshot, and adversary capability of the calibration epoch. Published κ values are valid only against the snapshot declared with each calibration release.

Adversary adaptation to published profiles is treated as a normal operational condition. As hardware improves or adversaries develop specific tooling against published lever configurations, recalibration is expected at the cadence specified in the *Bitcoin-Anchored Collateral Commitments* construction paper. Older commitments retain their structural validity - the work was performed, the anchors are real - but their *current signalling strength* under newer adversary capability is a verifier-policy concern, not an architectural one.

This recalibration discipline is operational, not aspirational. The model surfaces the dependence on snapshot conditions explicitly so it cannot be claimed as a hidden assumption.

---

## 9. Illustrative Derivation Under Stated Assumptions

This section demonstrates the method by working through one profile under one fully stated set of assumptions. **It is illustrative, not predictive.** Every assumption is named so the reader can substitute their own values and rederive.

### 9.1 Configuration

- **Profile**: Standard
- **Honest class**: Mid-range 2026 smartphone, background process during idle time
- **Adversary class**: RCA-2026 (concurrent producer, best-bin general-purpose server hardware, virtualisation, engineering depth)
- **Snapshot**: May 2026 hardware and fee-market reference (see Calibration Annex)
- **Batching**: Available; assumed ratio 100 commitments per anchor transaction

### 9.2 Stated assumptions

These values are illustrative inputs to the cost equations, not predictions about production deployment. Each is the kind of value that will be empirically anchored in the Calibration Annex.

| Parameter | Illustrative value | Notes |
|---|---|---|
| Burn duration | 24 hours | One-day Orange Anchor target |
| Anchor transactions per commitment | 2 (one start, one end) | Per construction |
| Anchor fee per transaction at snapshot fee rates | *F_tx* | Drawn from May 2026 fee-market reference |
| Batching ratio (honest) | 100 commitments per tx | Operator-dependent |
| Batching ratio (adversary), *B_adv* | Substantially lower (operator policy / clustering) | See §4 |
| Honest device energy during burn | Marginal vs idle baseline | Background process |
| Adversary server capex amortisation per active hour, *Capex_h* | *Capex_h* | Drawn from server hardware cost reference |
| Adversary compression ratio achievable | *r* (bounded above by analysis in the *Bitcoin-Anchored Collateral Commitments* construction paper) | Not asserted here |
| Adversary T_active per commitment | max(24h / r, T_floor) | T_floor set by anchor schedule |
| Adversary sensor-coherence simulation cost | *S* per instance | Engineering amortisation + runtime |

### 9.3 Composition

Honest cost composes as:

> *C_honest = E_device + (2 × F_tx / 100)*

(with *W_optional = 0*; the network-witnessing axis is not engaged under the Standard profile.)

Adversary cost composes as:

> *C_adv = (Capex_h × T_active) + P_power + (2 × F_tx / B_adv) + S + R_detection*

where *B_adv* is the adversary batching ratio defined in §9.2.

### 9.4 Illustrative honest-cost magnitude (worked example)

To anchor the cost-equation language to a concrete order of magnitude, the following worked example evaluates the honest-cost equation under one explicit set of assumptions. **The values are illustrative only.** They are intended to make the equation legible, not to commit the protocol to any specific deployment cost.

*Assumptions (all illustrative; substitute and rederive).*

- Bitcoin fee rate at snapshot: **10 sat/vB** (medium-fee mempool conditions; the fee market varies substantially between snapshots).
- Anchor transaction size: **~250 vB** per anchor transaction (typical SegWit single-input single-output with OP_RETURN commitment).
- Per-anchor fee at the assumed rate: 10 sat/vB × 250 vB = **2,500 sat ≈ $1.50 at $60,000/BTC** (snapshot-dependent).
- Per-commitment anchor cost without batching: 2 anchors × $1.50 = **$3.00**.
- Batching ratio (honest): **50 commitments per anchor transaction** (Standard profile, operator-aggregated).
- Per-commitment anchor cost with 1:50 batching: $3.00 / 50 = **$0.06** (six cents).
- Device-energy term `E_device` for a 24h background process on a mid-range 2026 smartphone: marginal vs idle baseline, typically **a few cents per commitment** at residential electricity rates.

*Composition.*

> C_honest ≈ E_device + (2 × F_tx / 50) ≈ $0.0X + $0.06 ≈ **tens of cents per commitment** under these assumptions.

*Sensitivity.*

- At a high-fee snapshot (50 sat/vB), per-commitment anchor cost rises to $0.30 + device energy — **low single dollars** per commitment.
- Without batching (1:1), per-commitment anchor cost rises to $3.00 — **a few dollars** per commitment even at the assumed 10 sat/vB rate.
- BTC/USD price moves the absolute USD figure proportionally; the satoshi-denominated cost is approximately stable across snapshots at constant fee rate.

Under the illustrative conditions stated above, per-proof honest cost falls in the range of **tens of cents to low single dollars** \u2014 the figure cited in the *Orange Anchor Integration Brief v2.2* \u00a75.4. Readers MUST substitute their own snapshot values for any production budgeting; the Calibration Annex (forthcoming) will publish numerical anchoring under a declared snapshot.

### 9.5 Scope of this example

The structural observations that fall out of this composition are stated in §3.4. This derivation is illustrative of the method - not a recommendation, prediction, or deployment specification. The Calibration Annex will provide the empirically-anchored numerical evaluation; this section’s purpose is to make the framework inspectable.

---

## 10. Scope Boundaries

The following are fixed boundaries of this document, not open items.

- **No cryptographic reductions or formal proofs.** Cryptographic primitives are treated as parameterised black boxes; their analysis belongs to the *Bitcoin-Anchored Collateral Commitments* construction paper.
- **No predictive κ values.** Specific κ magnitudes for deployment configurations are deferred to the Calibration Annex.
- **Single reference adversary class (RCA-2026).** Full threat modelling beyond this class is out of scope here.
- **Verifier-side policy excluded.** Scoring, weighting, and interpretation of κ by verifiers are out of scope.

---

## 11. Open Items

The following are in-scope concerns deferred to future versions.

- **Multi-class adversary modelling.** State-level actor, cartelised operator, and hardware-vendor insider classes are not modelled in this version.
- **Additional honest device classes.** Low-end smartphone, dedicated home server, and embedded device profiles are not covered.
- **Quantitative κ decay function.** The time-decay of κ as adversary hardware improves is currently qualitative; quantitative modelling is an open item.
- **Quantitative network-axis characterisation.** Cost effects of opt-in network witnessing are currently described directionally only.

---

## 12. Relationship to Other Documents

- **White paper (v2.2)** carries the structural cost-positivity and cost-linearity claims. This document operationalises those claims into a calibration framework.
- **Bitcoin-Anchored Collateral Commitments (BACC v1.9)** — the construction paper — specifies the cryptographic primitives, parameter values, compression-bound analysis, and recalibration cadence that this model treats as inputs.
- **Orange Anchor Lexicon v2.7** defines the canonical vocabulary used in this document, including *burn*, *envelope*, burn-tier terminology (lightweight 6-block, heavyweight 144-block), and the calibration-profile names (Light, Standard, Hardened, Maximum) used in §5.
- **BAVAI Reference** specifies the attribution index. It is referenced by the verifier-policy concerns acknowledged in §8 but is not part of the cost model itself.
- **Orange Anchor Interaction Patterns v1.0** specifies check-in and cosign, which are downstream applications of commitments and are out of scope here.
- **Calibration Annex (forthcoming)** will provide the empirical hardware-cost and fee-market snapshot, the numerical κ values for each profile, and the sensitivity analysis under perturbation of each major assumption.