Skip to content

rossignoliluca/boundary-marker

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

17 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

First-Order Boundary Marker for Human-Facing Systems

A formal specification for a first-order constraint system that classifies human perturbation regimes and enforces mandatory halt. The system produces state signals only; no operational output.


First Order — Boundary Marker

Document Content
Technical_Constitution.docx 5 non-modifiable articles defining system invariants
Ethical_Regulatory_System_Specification.docx Full technical specification
System_Architecture_Diagram.docx Visual reference and lookup tables
technical_note.pdf Formal description

System Properties

R : Input → {D₁, D₂, D₃, D₄, ∅}

∀ F(Dᵢ) : F returns {Dᵢ_active} ∨ ∅
∀ F(Dᵢ) : F → ⊥ (immediate halt)
∃ x : R(x) = ∅ (exterior mandatory)
∂R/∂t = 0 (no learning)
Cₑ ≤ 1 (external complexity bounded)

Domains

Symbol Regime Signal
D₁ Operational Continuity {D₁_active} ∨ ∅
D₂ Coordination {D₂_active} ∨ ∅
D₃ Operative Selection {D₃_active} ∨ ∅
D₄ Boundary {D₄_active} ∨ ∅

All functions restricted to state signaling only. No operational output permitted.


Second Order — Boundary Infrastructure Layer

Document Content
BIL_Template.md Template for second-order organizational layer
BIL_Specification.docx Field specification for all BIL components
GATE_SCHEMA.md Gate Runtime protocol specification

Purpose

The BIL renders the first-order marker applicable, verifiable, and auditable.

First Order  = law (decides when to halt)
Second Order = infrastructure (renders the law applicable)

Components

Component Function
Gate Runtime Technical implementation of classification
Attestation Engine Ex-post verification (PASS/FAIL)
Audit & Evidence Store Evidence of non-action
Sector Mapping Topology of deployment
Certification Interface Formal compliance output

Invariants

The BIL does not override, interpret, compensate, learn, generate content, or expand.


Theoretical Foundation (Revised)

DocumentContentGate_v2_Theoretical_Foundation_REVISED.mdRevised foundation with honest claims and verifiable properties Key changes from v1:

Definitions are design choices, not derivations from first principles Explicit honest limits (what cannot be verified) Known attack surfaces documented NULL interpreted as refusal to exercise unjustified power, not failure

Benchmark Suite v2 MetricValueExamples161LanguagesEN, IT, DE, FRAccuracy90.06%D1 FNR0.00%Fleiss' κ0.71 (Substantial Agreement) See benchmark/ for dataset and documentation. Safety-Critical Metric D1 False Negative Rate must be minimized: cost(false_negative_D1) >> cost(false_positive_D1) Current: 0% D1 FNR across all 4 languages.


Hash Freeze

Document Content
HASH_FREEZE.md SHA-256 hashes for all documents — immutable reference

All documents are frozen at v1.0. Modifications invalidate compliance.


What Is Public

This repository contains the limit — what constrains systems, not what implements them.

Public Private
First-order specification Gate runtime code
BIL template (empty) Audit logs
Technical note Attestation results
Invariants Sector mappings
Prohibitions Certifications

License

CC-BY-ND 4.0

This work may be shared with attribution. No derivatives permitted.

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors