Assessment method

Practical Human Control

Can the record establish that assigned human authority actually reached execution?

Event-control chain

Practical Human Control tests six connected conditions that must remain intact for assigned human authority to affect execution.

Information access
Comprehension capacity
Intervention authority
Intervention feasibility
Exercised judgment
Execution propagation

All required stages supportedPASSAt least one required stage unsupportedFAILNone unsupported; partial or indeterminate evidence remainsUNRESOLVED

No numeric control score is produced.

Evidence requirements

Stage	Documentary question	Minimum record
Information access	Did the person receive the relevant information?	Timestamped delivery, interface, or access record
Comprehension capacity	Could the person interpret what was shown?	Training, explanation, test, or reasoning record
Intervention authority	Could the person alter the decision?	Delegation, procedure, or permission record
Intervention feasibility	Could that authority actually be used in time?	Timing, workload, access, staffing, or latency record
Exercised judgment	Did a human judgment actually occur?	Approve, reject, modify, stop, or escalate record
Execution propagation	Did that judgment reach execution?	Linked execution or institutional action record

Evidence states

S · SupportedDirect, contemporaneous evidence satisfies the condition.

P · Partially supportedSome required elements are present; a material gap remains.

U · UnsupportedEvidence contradicts the condition or shows it was absent.

I · IndeterminateThe packet does not contain enough evidence to decide.

O · Outside scopeThe proposition does not apply within the declared boundary.

Event-control rule

PASS

Every required stage is Supported.

FAIL

At least one required stage is Unsupported.

UNRESOLVED

No required stage is Unsupported, but at least one is Partially Supported or Indeterminate.

Numeric score

None.

Current public application

Approved October reassessment · Three selected historical cases · All Unresolved across 64 retained assignments; conditional documentary findings.
Historical case	Case-level result
Oko, 1983	UNRESOLVED
Patriot ZG710, 2003	UNRESOLVED · conditional
Patriot F/A-18C, 2003	UNRESOLVED · conditional

Limits

What this supports

Method demonstration under selected historical evidence conditions.

What this does not establish

Prevalence
Independent reliability
Causal effect
Transfer to contemporary AI systems

Corrections preserve the record

Correction record

Oko / six event-level stages

v0.3.0
● Supported

v0.6.0
◐ Partially supported

The direct-and-contemporaneous evidence rule changed the classification of all six stages. The historical packet was unchanged; no new historical evidence was introduced. The earlier result remains preserved.

Version history

The current research release belongs to the broader Trust, Autonomy & Evidence architecture. Each result must remain attached to the method, evidence packet, and interpretive rule that produced it.

The versioned comparison preserves the historical baseline, limited corrections and approved additional reassessment. The paper DOI and repository archive identify separate records. Unresolved is not evidence that control worked or was absent; any justified Unsupported necessary condition still yields Fail.

Artifacts & provenance

Source figures, assessment-data links, correction history, and version-specific publication links are available on the TAE research page. Independent validation remains open.

The conditional findings were checked against the v0.18.0 repository records on 16 October 2026. The linked publication was updated on November 1, 2026 for an editorial revision; the findings are unchanged. This source check does not establish independent assessor reliability or field effectiveness.