Keywords

#automation #contract #validation #protocol #judgment #responsibility #decision-making #business-process #compliance

RNC — Responsibility Normalization Contract

Responsibility Normalization Contract Protocol (v1.0)

RNC is a contract and application protocol that removes human judgment from decision-making
and returns responsibility to where it structurally belongs: the input.

It transforms contracts from

“people deciding whether something is acceptable”
into
“systems accepting input only when it fully matches rules.”

---

Why RNC Exists

Modern organizations are overloaded with invisible responsibilities:

• Interpreting vague rules
• Making discretionary exceptions
• Explaining rejection reasons
• Taking blame for user mistakes

RNC eliminates these problems by design, not by training.

---

Core Principles

• Instant decision (True / False only)
• Zero human discretion
• No interpretation, no negotiation
• Clear responsibility split
  • Organization → rules & schema
  • User → input correctness
• No “checking”, no “pending”, no “manual review”

---

What RNC Achieves

Before	After (RNC)
Manual review	Automatic validation
“Just this once”	Rule update or nothing
Human explanation	Error ID only
Implicit responsibility	Explicit responsibility
Emotional complaints	Specification questions

---

Scope

In Scope

• Contracts
• Applications
• Registrations
• Any process that depends on user input

Out of Scope

• Internal HR evaluations
• Internal approvals
• Performance reviews
  (Use a different protocol)

---

RNC Workflow

Step 0 — Target Selection

Choose processes that involve user-provided input.

---

Step 1 — Raw Input Injection

Feed the following as-is into the RNC Compiler:

• Manuals
• Terms of Service
• Application guidelines
• FAQs
• Past rejection reasons (if available)

⚠️ No summarizing, no editing, no cleanup.
Human preprocessing mixes responsibility.

---

Step 2 — Structural Extraction (Automatic)

The compiler extracts:

• Input fields
• Required / optional flags
• Data types (number, date, string, enum)
• Prohibited conditions
• Conditional branches (If / Then)
• Truth sources / references

Outputs

• Schema.json
• Rules.json

---

Step 3 — Ambiguity Detection (RII-based)

The system flags ambiguous phrases such as:

• “in principle”
• “appropriate”
• “as deemed necessary”
• “case by case”
• “at our discretion”

Output

• DesignQuestions.md

👉 Humans do not judge cases here
👉 Humans only define rules

---

Step 4 — Error Definition

The system generates:

• Error IDs
• Violated rules
• User-facing correction commands (imperative form)

Output

• ErrorCatalog.json

---

Step 5 — RNC Package Completion

RNC is complete when all are present:

• Schema.json
• Rules.json
• ErrorCatalog.json
• TruthSources.json
• PolicyVersion

---

Runtime Protocol

Input Acceptance

• Input is rejected until it fully matches
• No provisional acceptance
• No “we’ll check later”

---

Decision Output (C3)

Only this format is allowed:

{
  "result": true | false,
  "error_ids": ["E-102", "E-341"]
}