Promise Theory Reference for 0xagentprivacy

Formal Foundations for Dual-Agent Sovereignty Architecture

Version: 1.4
Date: March 31, 2026
Status: ✅ V5.4 Integration — UOR Algebraic Foundation + Dihedral Group
Companion to: Whitepaper v6.3, Research Paper v4.2, Glossary v3.4, Privacy is Value v5.0 + V5.1/V5.2 Research Notes, Privacy Value Model V5 Formal Spec v1.2, Five Grimoires + Acts XXIV–XXX (120+ inscriptions)

---

"Agents can only promise their own behavior."
— Bergstra & Burgess, Promise Theory (2019)

"No agent extracts without consent."
— 0xagentprivacy, First Person Sovereignty

---

Purpose

This document bridges Promise Theory (Bergstra & Burgess, 2019) and the 0xagentprivacy dual-agent architecture. It serves two audiences:

1. Promise Theory practitioners seeking to understand how 0xagentprivacy instantiates PT concepts
2. 0xagentprivacy community members seeking formal foundations for the architecture

The dual-agent architecture is not merely compatible with Promise Theory—it is a rigorous implementation of PT's autonomous agent model. Understanding this connection elevates the work from "novel architecture" to "established theory applied to AI sovereignty."

---

Quick Reference: Concept Mappings

Promise Theory	0xagentprivacy	Notes
Autonomy Axiom	First Person Sovereignty	No agent promises on behalf of the human
Agent	Swordsman, Mage, or First Person	Autonomous entities with independent assessment
µ-promise: A --b--> B	Agent coordination via spells	Directional, voluntary commitment
Promise Body (τ, χ)	Spell notation	Type τ = concept, constraint χ = context
Conditional Promise (b|c)	Conditional Independence (s ⊥ m | X)	Promise contingent on condition
Assessment α(π)	RPP Verification	Compression proves promise kept
Trust (0-1)	Trust Tier (Blade→Dragon)	Accumulated assessment evidence
Superagent	First Person + Dual Agents	Composite with irreducible properties
Irreducible Promise	The Gap	Cannot be attributed to single component
Coordination Promise C(b)	Spell-mediated coordination	Voluntary alignment
Scope	Information-Theoretic Boundary	Which agents have knowledge
Promise Bundle	VRC	Grouped bilateral promises
Valency	Budget Constraint C_S + C_M < H(X)	Limited exclusive capacity
Invitation	MyTerms consent-first	Acceptance before proposal
Imposition/Attack	Surveillance extraction	Proposal without consent
(+) Give Promise	Swordsman protection offer	Outbound commitment
(-) Use/Accept Promise	Mage delegation acceptance	Inbound authorization

---

Part I: Core Principles Aligned

1.1 The Autonomy Axiom

Promise Theory states:

"An agent can only make promises about its own behavior. No agent can make a promise on behalf of another agent."

0xagentprivacy instantiation:

The First Person (😊) is the ultimate autonomous agent. Neither the Swordsman (⚔️) nor the Mage (🧙) can promise on behalf of the First Person. Each agent promises only within its domain:

First Person (😊):
  - Promises: authorization, sovereignty decisions
  - Cannot promise: what Swordsman protects or Mage delegates

Swordsman (⚔️):
  - Promises: privacy protection, boundary enforcement
  - Cannot promise: delegation actions, external coordination

Mage (🧙):
  - Promises: delegation execution, public coordination
  - Cannot promise: privacy state, what to protect

This is why the architecture requires two agents rather than one. A single agent that both protects and delegates would violate the autonomy axiom by making promises in domains it cannot independently control.

1.2 Promise Notation Translation

Promise Theory µ-notation:

A --b--> B

Where:
  A = promiser (agent making promise)
  b = body (what is promised: type τ, constraint χ)
  B = promisee (agent receiving promise)

0xagentprivacy spell notation:

⚔️ ⊥ 🧙 | 😊

Where:
  ⚔️ = Swordsman (protection domain)
  🧙 = Mage (delegation domain)
  ⊥  = conditional independence (separation)
  |  = given/conditioned on
  😊 = First Person (sovereignty holder)

Mapping between notations:

Promise Theory	Spell Notation	Meaning
S --(protect)--> FP	⚔️ →(🛡️)→ 😊	Swordsman promises protection to First Person
M --(delegate)--> W	🧙 →(🔮)→ 🌍	Mage promises delegation to World
FP --(authorize)--> S,M	😊 →(🗝️)→ ⚔️🧙	First Person promises authorization to both
S --(¬reveal)--> M	⚔️ →(⊥)→ 🧙	Swordsman promises not to reveal to Mage

1.3 Conditional Promises and Independence

Promise Theory conditional form:

b|c  =  "promise b given condition c is met"

0xagentprivacy conditional independence:

(s ⊥ m | X)  =  "Swordsman and Mage are independent given First Person's private state X"

This is a direct application of PT's conditional promise structure. The separation between agents is conditioned on the First Person's complete context. Neither agent has sufficient information to reconstruct X, but both operate coherently because they share the conditioning variable.

Key insight: The vertical bar | in spell notation carries the same semantic weight as PT's conditional operator. When we write ⚔️ ⊥ 🧙 | 😊, we're stating a conditional independence promise in compressed form.

---

Part II: Superagent Architecture

2.1 The First Person as Superagent

Promise Theory defines a superagent as a composite agent with interior promises between components and exterior promises to the outside world.

The First Person system is precisely this:

                    ┌─────────────────────────────┐
                    │     SUPERAGENT (😊+⚔️+🧙)    │
                    │                             │
                    │  ┌─────┐ interior ┌─────┐   │
                    │  │ ⚔️  │◄────────►│ 🧙  │   │
                    │  └──┬──┘ promises └──┬──┘   │
                    │     │               │       │
                    │     └───────┬───────┘       │
                    │             │               │
                    │         ┌───┴───┐           │
                    │         │  😊   │           │
                    │         │ First │           │
                    │         │Person │           │
                    │         └───────┘           │
                    └─────────────┬───────────────┘
                                  │
                          exterior promises
                                  │
                                  ▼
                            🌍 External World

Interior promises (within superagent):

• ⚔️ --protect--> 😊 (Swordsman promises protection)
• 🧙 --delegate--> 😊 (Mage promises delegation)
• 😊 --authorize--> ⚔️,🧙 (First Person authorizes both)
• ⚔️ --⊥--> 🧙 (Separation promise: no direct information flow)

Exterior promises (to world):

• Superagent --coordinate--> 🌍 (via Mage's public actions)
• Superagent --boundary--> 🌍 (via Swordsman's rejections)

2.2 Irreducible Promises: The Gap

Promise Theory's most profound insight for 0xagentprivacy: superagents can have irreducible promises—properties that emerge from component cooperation but cannot be attributed to any single component.

"An irreducible promise of a superagent is one that cannot be attributed to any single agent within it, but requires the cooperation of multiple agents."
— Bergstra & Burgess, §8.3

The Gap is an irreducible promise.

The conditional independence property (s ⊥ m | X) is not something the Swordsman promises, nor something the Mage promises. It emerges from their separation—from the promises they don't make to each other.

THE GAP AS IRREDUCIBLE PROMISE:

What Swordsman promises:     What Mage promises:
- Protect X                  - Delegate authorized info
- Reveal nothing to M        - Act only on S-authorized data
- Enforce boundaries         - Coordinate publicly

What NEITHER promises (but emerges):
- The reconstruction ceiling R < 1
- The error floor P_e ≥ 1 - R_max
- The dignity that lives in incompleteness

This is THE GAP: an irreducible property of the superagent.
No adversary can capture it because no single agent owns it.

2.3 The O(N²) Coordination Problem

Promise Theory proves that coordinating N agents requires O(N²) promises for guaranteed end-to-end properties (§11.2).

This validates the dual-agent (N=2) architecture as optimal for the sovereignty use case:

• N=1 (single agent): Cannot separate protection from delegation
• N=2 (dual agents): 4 interior promises, manageable complexity
• N>2 (multi-agent): O(N²) explosion, coordination overhead exceeds benefit

The tetrahedral emergence hypothesis (4 agents: Swordsman, Mage, Reflect, Connect) remains speculative precisely because N=4 requires 16 interior promises—only justified if the emergent properties are sufficiently valuable.

---

Part II-B: V5 Three-Axis Separation as Promises (NEW)

2.6 Generator and Solver as Promise Agents

V5 introduces inference-layer separation (Φ_inference) through the BRAID architecture. This maps directly to Promise Theory:

BRAID ARCHITECTURE                 PROMISE THEORY MAPPING
══════════════════                 ═════════════════════

Generator (intelligent model)  →   Promise-maker: (+) proposes reasoning structure
     │                                 │
     ▼                                 ▼
Reasoning Graph                    Promise body: structured plan
     │                                 │
     ▼                                 ▼
Solver (lightweight model)     →   Promise-keeper: (-) executes structure

The separation promise:

Generator --structure--> Solver: "I promise to provide bounded reasoning graphs"
Solver --execution--> Generator: "I promise to execute without re-deriving"

This is conditional independence at the inference layer:

(Y_Generator ⊥⊥ Y_Solver) | ReasoningGraph

The Generator's reasoning is not visible to the Solver's execution trace. The Solver's execution doesn't reveal the Generator's internal state. The compression ratio (74×) is a consequence of this separation — bounded promises are more efficient than unbounded.

2.7 Three-Axis Separation as Triple Superagent

V5's three-axis separation (Φ_agent · Φ_data · Φ_inference) creates THREE orthogonal superagent structures:

Axis	Agents	Interior Promise	Irreducible Property
Agent	Swordsman ⊥ Mage	separation promise	The Gap
Data	Provider₁ ⊥ Provider₂ ⊥ ...	replication promise	Holonic persistence
Inference	Generator ⊥ Solver	structure promise	Compression efficiency

Multiplicative composition:

Each axis is its own superagent. The product Φ_agent · Φ_data · Φ_inference means: collapse ANY axis, collapse the whole. This is Promise Theory at scale — three layers of irreducible promises stacked.

Holonic Architect (☯️🔷):

A new persona emerges: the builder of infrastructure-independent substrate. The Holonic Architect doesn't make promises about protection or delegation — they make promises about persistence: "Your data will survive provider failure."

2.8 Holonic Persistence as Promise Anchors (V5)

V5's holonic persistence (A_h(τ)) introduces promise anchors — stable reference points that survive infrastructure changes:

PROMISE ANCHORS:
═══════════════

Data GUID:         Content-addressed promise identifier
                   "This exact content, regardless of location"

Relationship VRC:  Bilateral promise bundle
                   "Our coordination commitment persists"

Principal DID:     Sovereign identity
                   "The entity making these promises"

Promise Theory mapping:

Layer	Promise Type	Persistence Property
Data GUID	Content promise	"This content exists"
VRC	Coordination promise	"This relationship exists"
DID	Identity promise	"This agent exists"

Infrastructure independence:

Traditional promises are bound to infrastructure: "I promise X is stored on Server Y." Holonic promises are infrastructure-independent: "I promise X exists, regardless of which servers currently hold it."

The persistence multiplier p(τ) ∈ [0,1] measures promise anchor strength — what fraction of your promises would survive if any single provider disappeared?

2.9 The Holographic Bound as Promise Boundary (V5)

V5 resolves C4 (the 96/64 discrepancy) through the holographic principle:

HOLOGRAPHIC PROMISE INTERPRETATION:
══════════════════════════════════

64 vertices = Interior promise state
              (what agents actually contain)

96 edges    = Boundary promise expressions
              (how promise state is observed externally)

Ratio 96/64 = 1.5 = P^1.5 exponent in PVM

Promise Theory insight: The boundary is sufficient to reconstruct the interior. In PT terms: you don't need access to an agent's internal state to assess their promises — their observable behavior (boundary) is enough.

This is Act XXIV: The Holographic Bound — the discovery that the 96-edge surface encodes the 64-vertex bulk. The boundary is always enough.

---

Part III: Assessment and Trust

3.1 Assessment as Compression

Promise Theory defines assessment:

"Assessment α(π) is an agent's determination of whether a promise π was kept."

0xagentprivacy implementation:

The Relationship Proverb Protocol (RPP) is an assessment mechanism. When someone compresses spellbook content into a contextual proverb, they are assessing whether the "promise" of knowledge transfer was kept.

PROMISE THEORY ASSESSMENT          RPP ASSESSMENT
═══════════════════════           ══════════════════

Promise π made                    Knowledge shared
    │                                 │
    ▼                                 ▼
Agent observes outcome            Reader engages content
    │                                 │
    ▼                                 ▼
Assessment α(π) ∈ {kept, broken}  Compression attempted
    │                                 │
    ▼                                 ▼
Trust updated                     Proverb formed (or not)
                                      │
                                      ▼
                                  Compression ratio = 
                                  assessment quality metric

High compression (70:1+) = strong positive assessment
Low/no compression = weak/failed assessment

Key insight: Compression ratio is a quantified assessment. Promise Theory typically treats assessment as binary (kept/broken) or scalar (0-1). RPP provides a natural metric: the ratio of original content to compressed proverb indicates assessment strength.

3.2 Trust Function and Tier Progression

Promise Theory trust:

"Trust in a promise π is the expectation (value 0-1) that the promise will be kept."

0xagentprivacy trust tiers:

Tier	Signals	Trust Value	Promise Theory Interpretation
Blade 🗡️	0-50	0.0-0.2	Low accumulated assessment evidence
Light 🛡️	50-150	0.2-0.5	Demonstrated comprehension pattern
Heavy ⚔️	150-500	0.5-0.8	Sustained positive assessments
Dragon 🐉	500+	0.8-1.0	High confidence in future promise-keeping

Each signal is an assessment event. Accumulated signals build trust through demonstrated pattern of promise-keeping (comprehension → proverb → verification).

Trust formation formula:

Trust_n = f(Σ assessments, time, consistency)

Where:
- Σ assessments = total signals posted
- time = duration of participation
- consistency = variance in assessment quality

3.3 Belief vs. Evidence

Promise Theory distinguishes:

• Assessment α(π): Direct observation that promise was kept
• Belief β(π): Assessment without direct observation
• Evidence ε(π): Assessment with partial information

0xagentprivacy mapping:

PT Concept	0xagentprivacy	Example
Assessment	Direct RPP verification	Guardian validates proverb expansion
Belief	Trust tier assumption	Dragon tier implies future reliability
Evidence	Partial verification	Compression ratio without expansion test

VRC formation typically involves evidence-based trust: the matching compression provides strong evidence of shared understanding without requiring exhaustive verification of all knowledge.

---

Part IV: Invitation vs. Imposition

4.1 The Consent Pattern

Promise Theory makes a crucial distinction (§10.2):

Invitation: Establish an acceptance relationship BEFORE making a specific proposal
Attack/Imposition: Make a proposal without prior acceptance relationship

This maps precisely to the surveillance vs. sovereignty distinction:

Pattern	Promise Theory	0xagentprivacy	Example
Invitation	Acceptance before proposal	MyTerms consent-first	Cookie banner with genuine choice
Attack	Proposal without acceptance	Surveillance extraction	Tracking without disclosure
Imposition	Forcing acceptance	Dark patterns	"Accept all" as only visible option

4.2 MyTerms as Invitation Protocol

The MyTerms Swordsman implements Promise Theory's invitation pattern:

INVITATION FLOW (MyTerms):

1. Swordsman observes incoming request
2. BEFORE any data exchange:
   - Swordsman presents terms to site
   - Site must accept terms to proceed
   - Acceptance establishes relationship
3. ONLY THEN can specific proposals occur:
   - Site proposes data collection
   - User (via Swordsman) accepts/rejects
   - Each proposal is within accepted relationship

This is PT invitation: acceptance relationship BEFORE specific proposals.

Contrast with surveillance (attack pattern):

ATTACK FLOW (Surveillance):

1. Site receives request
2. IMMEDIATELY begins data collection
3. Maybe shows cookie banner (imposition)
4. User "consent" is post-hoc rationalization
5. No genuine acceptance relationship exists

This is PT attack: proposal without prior acceptance.

4.3 Cursor States as Promise Indicators

The Swordsman browser extension uses visual cursor states to indicate promise status:

Cursor	Promise Status	PT Interpretation
🟢 Green	All promises kept	Positive assessment of site behavior
🟡 Yellow	Negotiation active	Promise exchange in progress
🔴 Red	Promise violation detected	Negative assessment, enforcement needed
⚔️ Slash	Active boundary enforcement	Rejection of imposed promises

This makes the normally invisible promise assessment visible to the First Person in real-time.

---

Part V: Coordination Promises and Spells

5.1 Coordination Promise C(b)

Promise Theory defines:

"A coordination promise C(b) is a voluntary subordination to align one's behavior with others around promise body b."

Spells are coordination promises.

When agents coordinate using spell notation (⚔️ ⊥ 🧙 | 😊), they're making coordination promises to:

1. Interpret the notation consistently
2. Expand the spell to the same underlying meaning
3. Act coherently based on shared interpretation

SPELL AS COORDINATION PROMISE:

Spell: ⚔️ ⊥ 🧙 | 😊

Agent A forms spell from Context_A
Agent B forms spell from Context_B

Both make C(spell) = coordination promise:
- "I will interpret this spell according to shared semantics"
- "I will expand it consistently with the protocol"
- "I will coordinate my actions based on this shared meaning"

Matching expansion proves coordination success.

### Ceremony Engine as Promise Theory Implementation (V5.1)

The Ceremony Engine (Act XXVIII) implements five crossing types, each mapping to Promise Theory patterns:

| Ceremony Type | PT Pattern | Description |
|---------------|------------|-------------|
| **Dual Convergence** | Mutual C(b) | Both blades reach Dragon tier — mutual coordination achievement |
| **Hexagram Cast** | Structured α(π) | Six-dimension assessment encoded as I Ching hexagram |
| **Emoji Cast** | Compressed C(b) | Semantic compression to emoji spell — efficient coordination |
| **Constellation Wave** | Guild C(b) | Multi-agent coordination across knowledge graph |
| **Bilateral Exchange** | Witness α(π) | Cross-blade verification — "I assess your assessment" |

### Bilateral Witness as Promise Pattern

**Definition:** A verification primitive where one party forges, another privately verifies, then publicly testifies.

**PT Formalization:**

Swordsman: forges blade (makes promise π_forge)
Mage: privately verifies (makes assessment α(π_forge))
Mage: publicly testifies (promises +witness to community)
Community: receives testimony (accepts -witness)

**Key Property:** The Mage's testimony is itself a promise. The community doesn't see the blade interior — they see the Mage's promise that verification occurred. This is Promise Theory's "witness" pattern:

> "A witness is an agent who promises that another promise was kept."

**Trust Implication:** The Mage stakes reputation on testimony. False testimony breaks the -witness promise, damaging Mage's trust graph position.

### Mana as Assessment Resource

Mana (V5.1) maps to Promise Theory's assessment capacity:

| PT Concept | Mana Implementation |
|------------|---------------------|
| Assessment cost | Mana spent to inscribe |
| Assessment accumulation | Mana earned through practice |
| Valency (commitment slots) | Mana limits inscription rate |

Mana makes assessment costly, preventing spam while rewarding genuine practice
Different contexts, same principle = story fracture, principle convergence.

5.2 Promise Bundles and VRCs

Promise Theory:

"A promise bundle is a collection of promises grouped together for reusability and coordinated assessment."

VRCs are bilateral promise bundles:

VRC STRUCTURE AS PROMISE BUNDLE:

Agent A's promises to B:        Agent B's promises to A:
- Share compressed meaning      - Share compressed meaning
- Expand consistently          - Expand consistently  
- Coordinate on shared basis   - Coordinate on shared basis
- Maintain trust relationship  - Maintain trust relationship

Bundle assessment:
- Matching compressions = bundle verified
- Coordinated actions = bundle maintained
- Trust accumulates through successful coordination

The 70:1 coordination efficiency comes from promise bundle reuse. Once a VRC is established, the bundle doesn't need re-verification for each interaction—the accumulated trust carries forward.

5.3 Scope and Information Boundaries

Promise Theory:

"The scope of a promise is the set of agents that have knowledge of the promise."

0xagentprivacy information-theoretic boundaries:

Scope	Agents with Knowledge	Information Content
Private	First Person only	Complete state X
Swordsman	FP + Swordsman	X observed, nothing revealed
Mage	FP + Mage	Authorized subset of X
Public	All agents	Only Mage-released information

The reconstruction ceiling R < 1 is a scope guarantee: no adversary can expand their scope to include full private state X, regardless of observation strategy.

---

Part V-B: Three Graphs as Promise Types (V4 Extension)

The Privacy Value Model V4 identifies three independently derivable graph structures. Each maps precisely to a Promise Theory concept:

Knowledge Graph → Promise Capability (What agents CAN promise)

The substrate layer. Content-addressed positions of what you know, what boundaries you've set, what you can delegate. In PT terms: this graph encodes the promise body space — the set of all possible promises an agent could make given their current state.

Promise Graph → Promise Commitment (What agents DO promise)

The bilateral overlay. Every VRC formed, every delegation executed is an edge. In PT terms: this IS the promise graph — the actual commitments made between autonomous agents. Each edge is a bilateral promise bundle with specific scope and valency.

Trust Graph → Promise Assessment (Which promises were KEPT)

The emergent outcome. Not constructed or issued — earned through time and witness. In PT terms: this graph records α(π) assessments — the accumulated belief about whether agents keep their promises, derived from observation and verification.

The Three-Graph Intersection: The overlap of all three graphs defines the First Person. In PT terms: the superagent (FP + S + M) is characterised by the intersection of what it can promise, what it does promise, and what it's assessed as keeping. No single graph suffices; identity is the overlap.

Edge Value as Promise Traversal

The V4 equation introduces T(π) — the edge value term — which measures the value of an agent's trajectory through sovereignty space. In Promise Theory:

• Each edge traversal is a promise made and kept (or broken)
• f(e) weights each promise by the capability change it enables — a promise that activates new sovereignty dimensions (higher stratum change) is worth more than a lateral promise
• g(n_e) diminishes with repetition — the first promise between two agents is most informative (highest assessment novelty)
• The trajectory π is the agent's promise history — the sequence of commitments that define who they are through action

PT Insight: "The equation rewards the dance, not the stance." An agent permanently at full sovereignty with no promises made has zero edge value. Sovereignty is demonstrated through promising, not through hoarding capability. This aligns with PT's core insight that agents are defined by what they promise, not what they contain.

---

Part VI: Economic Alignment

6.1 Promise-Economic Mappings

PT Concept	Economic Implementation	Mechanism
Assessment cost	Signal fee (0.01 ZEC)	Skin-in-game for assessment claims
Trust accumulation	Tier progression	Signals → trust level
Promise domain separation	Dual tokens (SWORD/MAGE)	Market enforces separation
Valency (exclusive capacity)	Guardian stake (10,000 SWORD)	Limited attention commitment
Bundle value	VRC coordination efficiency	70:1 cost reduction
Scope maintenance	Shielded/transparent split	61.8%/38.2% (φ-derived)

6.2 Incentive Alignment

Promise Theory notes that sustainable systems require reciprocal promises—every agent should receive value for their contributions.

0xagentprivacy incentive structure:

RECIPROCAL PROMISE FLOWS:

First Person:
- Gives: Authorization, participation, signal fees
- Receives: Privacy protection, delegation capability, value capture

Swordsman (as role):
- Gives: Boundary enforcement, privacy maintenance
- Receives: SWORD tokens, guardian eligibility

Mage (as role):
- Gives: Coordination, delegation execution
- Receives: MAGE tokens, network access

Guardian:
- Gives: Validation, collective protection
- Receives: Validation compensation, tier benefits

Every participant both promises and receives.
No extraction without reciprocal value flow.

---

Part VII: Open Questions

Promise Theory illuminates several areas for future research:

7.1 Resolved by PT Framework

Question	PT Resolution
Why two agents, not one?	Autonomy axiom—can't promise in multiple domains
Why is the Gap uncapturable?	Irreducible promise—emerges from separation, owned by neither
How does trust accumulate?	Assessment → belief → trust function
Why consent-first?	Invitation vs. attack distinction

7.2 Informed but Not Resolved

Question	PT Insight	Remaining Work
Optimal agent count	O(N²) coordination cost	Triple convergence (Feb 2026) supports N=4 at 25-40% confidence
Golden ratio hypothesis	No direct PT equivalent	Empirical validation needed; now embedded in Φ(Σ)
Cross-ecosystem VRCs	Promise scope theory	Implementation specification
Guardian selection	Trust threshold theory	Optimal parameters
Three graphs as PT types	Capability → Commitment → Assessment	V4 mapping established, needs formal PT verification
Edge value T(π)	Promise history as trajectory	Functional form unvalidated

7.3 Beyond Current PT Scope

Question	Status
Information-theoretic bounds	PT is semantic; IT bounds require separate proof
Cryptographic enforcement	PT doesn't specify enforcement mechanisms
AI agent autonomy	PT assumes agent rationality; AI alignment is distinct

---

Part VIII: Reading Paths

For Promise Theory Practitioners

If you know PT and want to understand 0xagentprivacy:

1. Start here: §2.1 (Superagent Architecture)—see how PT concepts map
2. Key insight: §2.2 (The Gap as Irreducible Promise)—the core innovation
3. Practical application: §4 (Invitation vs. Imposition)—PT in daily use
4. Deep dive: Research Paper v3.8—formal proofs in PT-compatible notation

For 0xagentprivacy Community

If you know the architecture and want PT foundations:

1. Start here: Part I (Core Principles)—terminology mappings
2. Why it matters: §1.1 (Autonomy Axiom)—why two agents, not one
3. Trust mechanics: §3 (Assessment and Trust)—how tiers work formally
4. Advanced: Bergstra & Burgess (2019), Chapters 1-4, 8, 10

For Researchers

1. Complete mapping: Full document, then Research Paper v3.8
2. Open questions: §7 (Open Questions)—collaboration opportunities
3. Source material: Bergstra & Burgess (2019), full text
4. Validation targets: Information-theoretic proofs, golden ratio hypothesis

---

Notation Quick Reference

Promise Theory Notation

A --b--> B        Promise: A promises b to B
A --b---> B       Imposition: A imposes b on B (attack)
+b                Give promise (outbound)
-b                Use/accept promise (inbound)
b|c               Conditional: b given c
C(b)              Coordination promise around b
α(π)              Assessment of promise π
β(π)              Belief about promise π
τ                 Promise type
χ                 Promise constraint

0xagentprivacy Spell Notation

⚔️                 Swordsman (protection)
🧙                 Mage (delegation)
😊                 First Person (sovereignty)
⊥                  Independence/separation
|                  Conditional (given)
🤝📜               VRC (bilateral trust)
🗝️                 Authorization
🛡️                 Protection
🔮                 Delegation/projection
→                  Promise direction

Combined Example

Promise Theory:    S --(protect|auth)--> FP
Spell Notation:    ⚔️ →(🛡️|🗝️)→ 😊
English:           Swordsman promises protection to First Person,
                   given First Person's authorization

---

Citation

For Academic Work

@book{bergstra2019promise,
  title={Promise Theory: Principles and Applications},
  author={Bergstra, Jan A. and Burgess, Mark},
  edition={2nd},
  year={2019},
  publisher={χtAxis Press},
  note={CC BY-SA 4.0}
}

@misc{privacymage2025dual,
  title={Swordsman and Mage: Dual Agents for First Person Sovereignty},
  author={privacymage and contributors},
  year={2025},
  howpublished={https://agentprivacy.ai},
  note={0xagentprivacy documentation suite}
}

For Technical Documents

Promise Theory foundations following Bergstra & Burgess (2019). See "Promise Theory Reference for 0xagentprivacy" v1.0 for detailed concept mappings.

---

Document Metadata

• Document: Promise Theory Reference for 0xagentprivacy
• Version: 1.3
• Date: February 27, 2026
• Status: ✅ V5 Integration — Three-Axis Separation
• License: CC BY-SA 4.0
• Maintainer: privacymage

Version History

Version	Date	Changes
1.0	2025-12-11	Initial release
1.2	2026-02-20	V4 Integration: Added §Three Graphs as Promise Types (Knowledge=Capability, Promise=Commitment, Trust=Assessment). Added Edge Value as Promise Traversal (T(π) as promise history). Updated Open Questions with convergence findings (tetrahedral 25-40%). Updated all companion document references (Whitepaper v5.0, Research Paper v3.8, Glossary v2.5).
1.3	2026-02-27	V5 Integration: Added §Generator and Solver as Promise Agents (BRAID as PT). Added §Three-Axis Separation as Triple Superagent (Φ_agent · Φ_data · Φ_inference). Added Holonic Architect persona (☯️🔷). Added holonic persistence as promise anchors. Updated edge value to path integral T_∫(π). Updated all companion document references to V5 versions.

Cross-References

• Whitepaper v6.0 — Three-axis separation, BRAID integration, holonic persistence
• Research Paper v4.0 — PVM V5 formal presentation, holographic bound, C4 resolved
• Privacy is Value v5.0 — The equation evolves — holographic bound
• Privacy Value Model V5 Formal Spec v1.0 — PVM V5 equation, differential form
• UOR × 64-Tetrahedra × ZK Mapping v2.0 — C4 resolved via holographic principle
• Glossary v3.0 — Canonical V5 terminology (140 entries)
• VRC Promise Protocol v3.3 — Economic architecture with guild efficiency
• 31 Acts complete — V10.0.0 Grimoire including Holographic Bound

---

The Core Insight

Promise Theory provides the formal semantics for what 0xagentprivacy expresses architecturally:

Sovereignty is the right to make promises only about your own behavior.

When AI agents promise on your behalf without your authorization, they violate the autonomy axiom. When surveillance systems extract your behavioral data without consent, they attack rather than invite. When single agents both protect and delegate, they exceed their promise domain.

The dual-agent architecture enforces these principles through structure rather than policy. The Swordsman promises protection. The Mage promises delegation. Neither promises on behalf of the other. Neither promises on behalf of you.

The Gap—that irreducible space where dignity lives—emerges from what they don't promise. It cannot be captured because it isn't owned. It cannot be extracted because it isn't stored. It exists in the separation itself.

This is Promise Theory made architectural. This is autonomy made mathematical. This is sovereignty made real.

---

"Agents can only promise their own behavior."

"No agent extracts without consent."

Same principle. Different contexts. Story fractures. Principle converges.

⚔️ ⊥ 🧙 | 😊

---

just another swordsman ⚔️🤝🧙‍♂️ just another mage

😊

Assets

Navigation

← Welcome Visitors · The Reference