Skip to content

Core Concepts

KTP's constraint model is not a metaphor—it's an engineering discipline. This page explains the fundamental concepts that make KTP work.

Structure over Policy

Traditional security treats cyberspace as a lawless frontier requiring policies to impose order. KTP takes a different view: digital environments can have *structure*—fundamental constraints that govern what's possible, not just what's permitted.
Policy-Based
  • Human-speed enforcement
  • Depends on interpretation
  • Easily circumvented
  • Says "you shouldn't"
Structure-Based
  • Machine-speed enforcement
  • Mathematically consistent
  • Cannot be circumvented
  • Says "you can't"

The Structural Principle

In physical reality, you don't need a policy against exceeding the speed of light. The structure of reality makes it impossible. KTP creates analogous constraints for autonomous agents.(1)

  1. The distinction between policy-based and structure-based security is foundational to KTP. See KTP-CORE Section 1.2, "The Environment-Based Solution."

The Zeroth Law

At the heart of KTP is a single, inviolable constraint:
A Autonomy
≤
E Environment

Action risk must never exceed environmental capacity

Supremacy

The Zeroth Law cannot be suspended, overridden, or circumvented by any mechanism, credential, or authority. It applies equally to all agents regardless of lineage, generation, or purpose.(1)

  1. The Zeroth Law's supremacy is established in Constitution Article I, Section 2, and specified in KTP-CORE Section 4.

What It Means

  • A (Autonomy): The intrinsic risk of the action an agent wants to take
  • E (Environment): The current Trust Score—the environment's capacity to absorb risk

The Silent Veto

The Silent Veto

When A > E

Agent Request Action with risk level A
→
Zeroth Law Check Compare A to E
→
A ≤ E Allowed
A > E Impossible

The Silent Veto is not a punishment or denial message—it's structure. The agent doesn't receive an "access denied" error—the action simply becomes impossible, like trying to walk through a wall.(1)

  1. The Silent Veto mechanism is defined in KTP-CORE Section 8, covering action risk classification and veto triggers.

Capability Attenuation

If the Zeroth Law is the constraint, capability attenuation is the enforcement mechanism. Enforcement is graduated, not binary: as environmental capacity tightens, operations are throttled, privileges are downgraded, and non-essential work is deferred. The agent experiences increasing resistance rather than a closed door.(1)

  1. Attenuation mechanics are fully specified in KTP-ATTENUATION, covering constraint types, response curves, and real-time application.

Analogy (informative)

In physical space, gravity curves spacetime. Objects don't decide to fall—they follow the curvature. Attenuation behaves the same way: agents don't decide to slow down—latency increases, compute becomes scarce, network paths narrow. The comparison is informative only; the mechanism is the graduated constraint set below.

Attenuation Mechanisms

Latency Injection Response delays increase, slowing rapid-fire attacks
Time Dilation Operations take longer, preventing timing exploits
Compute Throttling Processing capacity reduces, limiting resource consumption
Network Isolation Connectivity constrains, containing suspicious agents

Experience Score (E)

Trust in KTP is not granted by authority—it's earned through demonstrated behavior over time. The primary output of the KTP model is the Experience Score (E_trust), a live 0–100 meter of how much autonomy an agent has actually earned.

The Trust Equation

Risk Deflation

E_trust = E_base × (1 - R)

E_base Raw performance score
→
Risk (R) Environmental friction
→
E_trust Effective trust score

What drives Risk Deflation up:

  • Open vulnerabilities or failed controls (patch gaps, weak TLS, bad secrets)
  • Adversarial signals (DDoS indicators, anomaly spikes, tampering)
  • Contextual pressure (regulated data, high-stakes phase, critical audience)

Risk Deflation in Action

Clean environment: An agent scores 90 on base performance. With no risk factors (R = 0), its effective trust stays at 90.

Vulnerability detected: A security issue appears, pushing R to 0.5. The agent's effective trust instantly drops to 45—half its original score. The Zeroth Law now blocks any action requiring trust above 45.

Trust Velocity

KTP also tracks how trust is changing:(1)

  1. Trust Velocity (dE/dt) and its role in anti-gaming measures is covered in KTP-CORE Section 5.4 and Section 5.5 on Trust Score Integrity.
\[\frac{dE}{dt}\]

Rapid trust changes—either building or eroding—are themselves signals. Trust that rises too fast may indicate gaming. Trust that falls suddenly may indicate compromise.

Vector Identity

Traditional identity asks "Who are you?" and expects a static answer (credential, certificate, token).

KTP asks "What have you been doing?" and expects a trajectory.(1)

  1. Vector Identity replaces static credentials with trajectory-based authentication. See KTP-IDENTITY Section 3, "Vector Identity Model."
Traditional Identity
  • Static credentials
  • Point-in-time verification
  • "I am X"
  • Possession of secrets
Vector Identity
  • Continuous behavior
  • Trajectory analysis
  • "I have been doing Y"
  • Demonstration of patterns

The Passport Fallacy

A passport proves you were verified at some point. It says nothing about what you've done since. Vector Identity treats identity as a verb—something you continuously demonstrate through behavior, not a noun you possess.

Lineage Evolution

Sp
Sponsored Apprentice New agents operate under sponsor supervision
In
Independent Journeyman Proven agents gain independence
Gu
Guarantor Master Mature agents with full autonomy

Each phase requires demonstrated survival under real conditions—trust cannot be shortcut.(1)

  1. Lineage Evolution phases (Sponsored, Independent, Guarantor) are specified in KTP-IDENTITY Section 8.

Context Signals

To enforce the Zeroth Law, KTP must measure both A (action risk) and E (environmental capacity). Context Signals provide the measurement framework.(1)

  1. The complete Context Signals specification spans 1,644 signals. See KTP-SIGNALS for measurement definitions, aggregation rules, and instrumentation requirements.

The Seven Named Inputs

Input What It Measures Explore
evidence_density Telemetry density and volume Deep dive →
trust_trend Direction and velocity of change Deep dive →
update_resistance Resistance to rapid shifts Deep dive →
adversarial_pressure Environmental stress and anomaly load Deep dive →
moment_criticality Temporal context and decay Deep dive →
attestation_coverage Attestation coverage and visibility Deep dive →
soul Constitutional constraints (vetoes) Deep dive →

Input Interaction

These inputs don't operate in isolation. For example, high adversarial_pressure combined with low update_resistance creates rapid trust collapse, while high evidence_density with stable trust_trend indicates a healthy, predictable system.

Measurement Principles

Observable over Internal Measure what agents do, not what they "think"
Continuous over Binary Trust is a spectrum, not yes/no
Trajectory over Snapshot Single measurements are noisy; patterns matter

Where to Go Next

  • The Constitution

    See how these concepts become law in the foundational governance document.

    Constitution

  • Context Signals

    Explore the signal catalogue with interactive visualization.

    Context Signals

  • Use Cases

    See how KTP applies to real-world scenarios.

    Use Cases