Two connected learning models Explore the linear model at KillChains.com

Governance

Risk, law, and authority

A network can reveal a technically feasible path without making that path lawful, authorized, strategically wise, or safe. Kill-web design must preserve those distinctions at every step.

Research basis KW-RPT-012 KW-RPT-013 KW-RPT-014

Connectivity is potential; command determines use

A sensor may produce trustworthy data while the receiving organization lacks authority to act. A designated authority may exist while the evidence has become stale. A technically compatible capability may be reserved, unreleasable to a partner, or unsuitable for the environment.

The interface should expose these conditions separately rather than collapsing them into one green “available” state.

Separate four decision functions

FunctionQuestionTypical owner
Data discovery and fusionWhat is happening and how reliable is the evidence?Sensors, intelligence, fusion, and operations cells
Operational recommendationWhich feasible options support commander intent?Decision-support services and battle-management staff
AuthorizationIs the action permitted, and by whom?Designated commanders and legal or policy authorities
Execution controlWhich unit acts, within what bounds, and when?Tactical commander or delegated execution authority

Human judgment must be real, not decorative

  • The person has enough time to understand the decision.
  • The evidence, uncertainty, age, and alternatives are visible.
  • The operator has actual authority to reject, delay, modify, or abort.
  • The intervention path is technically reachable under expected degradation.
  • The system records who defined data, models, thresholds, permissions, and mission bounds.
  • Post-action review can reconstruct the evidence and reasoning rather than blame the final button press.

Legal and policy review follows the system lifecycle

Legal review cannot be limited to hardware. Material changes in models, training data, sensors, operating environments, autonomy functions, communications dependencies, and human-machine interaction may change the system that was originally evaluated.

Algorithms may estimate consequences and compare options, but distinction, proportionality, precautions, escalation, and command responsibility involve contextual and normative judgments that cannot be reduced to a single score.

Speed creates strategic risk

Faster sensing and coordination can improve defense against time-sensitive threats, but it also compresses time for deconfliction, consultation, correction, and assessment. Reciprocal automation can create pressure to act because each side assumes the other is accelerating.

Mitigations include independent corroboration, explicit escalation gates, bounded defensive modes, minimum decision windows where feasible, communication channels, and defaults that continue sensing rather than advance toward an irreversible action when evidence is ambiguous.

Research basis: KW-RPT-012, KW-RPT-013, and KW-RPT-014.

Answer-ready summary

Direct answers

What does Risk, Law & Authority cover?

Separate technical reachability from lawful authority, human judgment, evidence quality, accountability, and escalation control.

Read the supporting page