Run Publication Standard

This page defines the minimum conditions required to publish a result as a
GCE run. It is a boundary document: it governs what qualifies,
not what it means.

A Predictive Stress Zone (PSZ) is not a conclusion.
It is a conditional detection artifact that requires falsification disclosure to be interpretable at all.

Detection-only • Explanation firewall • Falsification required • Publication-safe by default

What this is: A normative standard for run validity. If the requirements below are not met, the work may be valuable, but it is not publishable as GCE.

What this is not: Not an interpretation guide, not a forecasting method, not a causal inference framework, and not a policy or decision tool.

Definition of a “GCE run”

A publication can claim “this is a GCE run” only if it:

  • Performs predictive assumption stress detection under falsification,
  • Maintains strict separation between detection and explanation,
  • Reports both persistence and collapse outcomes, and
  • Provides sufficient detail for an independent executor to replicate.

Non-claims (mandatory)

A GCE run must not assert or imply any of the following:

  • Causal attribution (why the stress exists)
  • Forecasts (what will happen next)
  • Prescriptions (what should be done)
  • Ranking of people, organizations, or places by “risk,” “quality,” or “competence”

If you need these, you are outside GCE by design.

Run validity requirements

1) Pre-commit perturbations (Required)

Before inspecting outcomes, declare the perturbation family you will apply to test stability.

  • Specify the set (e.g., window sizes, thresholds, smoothing, aggregation choices).
  • State the allowed parameter ranges.
  • Pre-commitment must be time-stamped (e.g., in the run log, repo, or note)
    before the first result is generated.
  • Do not add new perturbations after results are known, except as a clearly labeled follow-on study.

2) Report falsification outcomes (including collapse) (Required)

A valid run must report where detected stress persists and where it collapses.

  • Collapse is a publishable outcome.
  • Do not publish “hits only.”
  • Include at least one adversarial perturbation intended to break the detection.

3) Enforce the Explanation Firewall (Required)

Separate detection statements from interpretation. Detection must stand without narrative.

  • Include an explicit section titled Not Tested / Not Inferred.
  • Do not discuss causes, drivers, motivations, or mechanisms.
  • Do not propose interventions, remedies, or recommendations.

4) Replication sufficiency (Required)

Provide enough detail that an independent executor can reproduce the run.

  • Data source(s) and exact retrieval method (URLs, version identifiers, snapshots if applicable).
  • Exact assumption under test (worded as an assumption, not a conclusion).
  • Exact detection metrics and criteria for “stress.”
  • Perturbation set, including parameter values used.
  • Perturbation results table listing each perturbation and whether the detection
    persists or collapses.
  • Execution logs or reproducible steps (as applicable).

5) Granularity and publication safety (Required)

Do not publish results at a resolution that can reasonably be used to target or punish.

  • No individual-level units, doxxable subgroups, or small-N targeting.
  • Aggregate when necessary to avoid stigma, harassment, or re-identification.
  • Simple test: if the output can plausibly be used to target a specific person,
    small group, or small organization, do not publish at that resolution.
  • Avoid “stress labeling” language that implies defect, blame, or incompetence.

6) No leaderboards or comparative moralization (Required)

The protocol is an audit lens, not a competitive scoring system.

  • No public rankings of entities by stress level.
  • No implied “good/bad” judgments tied to stress detection.
  • No conversion of detection into reputational claims.

Mandatory publication template

Required headings (verbatim)

  • Assumption Under Test
  • Data Sources
  • Detection Method
  • Perturbation Set (Pre-committed)
  • Results: Persistence
  • Results: Collapse / Non-persistence
  • Not Tested / Not Inferred
  • Replication Notes
  • Publication Safety / Aggregation

Required language block

This document reports detection results from a Global Curiosity Engine (GCE) run.
It does not attribute causes, make forecasts, or recommend actions. Any causal,
predictive, or prescriptive interpretation is outside the scope of this run.

Common invalid patterns (non-exhaustive)

These patterns break the standard:

  • Selective reporting (publishing persistence without collapse cases)
  • Post-hoc perturbations added to “improve” results after inspection
  • Interpretation drift (“this suggests X is the cause”)
  • Decision laundering (“therefore we should do Y”)
  • Targeted granularity enabling stigma or punishment
  • Ranking / scoring entities in a comparative list
  • Headline-only publication without the falsification/collapse section

Compliance posture

GCE is a methodology, not an authority. This standard exists to preserve epistemic integrity
under publication and adversarial reading. Work that does not meet this standard should be
labeled clearly as “GCE-adjacent” or “inspired by GCE”, not as a GCE run.

Definition: “GCE-adjacent” means inspired by GCE, but missing one or more
validity requirements on this page.

Scroll to Top