Evidence & Corrections

Decision Atlas Evidence & Corrections

Evidence should be traceable. Corrections should be visible.

Decision Atlas publishes practical investigations with the expectation that readers should be able to understand the evidence, inspect the reasoning, see what remains uncertain, and identify when a conclusion has changed.

This page records the standards used to manage evidence, corrections, revisions, supersession, and withdrawal.


Evidence standards

Major conclusions should be supported by evidence that can be identified and reviewed.

Depending on the investigation, evidence may include:

  • primary sources
  • technical documents
  • research papers
  • official records
  • expert explanations
  • documented practitioner experience
  • observations and measurements
  • real-world testing

The relevance, reliability, and limitations of evidence may differ. Decision Atlas does not treat all sources as equally authoritative merely because they are available.


Traceability

Important findings should remain connected to the evidence used to support them.

Where practical, a published investigation should identify:

  • the source or source group supporting a major conclusion
  • the date or version of the evidence used
  • the distinction between direct evidence and inference
  • material limitations affecting interpretation
  • meaningful contradictions or unresolved questions

Traceability does not guarantee that every conclusion is correct. It allows the reasoning to be inspected, challenged, corrected, and improved.


Evidence freeze

An investigation may establish an evidence freeze before publication drafting begins.

The evidence freeze identifies the body of evidence used for the current synthesis and publication version.

Its purpose is to allow:

  • analysis to reach a defined stopping point
  • conclusions to be traced to a bounded evidence corpus
  • editorial review to occur against a stable record
  • later evidence to be distinguished from evidence used in the original publication

An evidence freeze is not a claim that no further evidence exists or that the conclusion can never change.


Correction classes

Minor correction

A change that improves accuracy or clarity without materially changing the investigation’s conclusion.

Examples may include:

  • typographical errors
  • broken links
  • misstated dates
  • minor wording corrections
  • clarification of an ambiguous sentence

Substantive correction

A change required because a material statement, source interpretation, calculation, factual claim, or supporting explanation was inaccurate.

A substantive correction should be disclosed in the publication’s revision record.

Revision

A meaningful update that changes, expands, narrows, or qualifies the published judgment because of new evidence, new testing, or improved analysis.

Supersession

A prior publication is superseded when a newer publication becomes the governing version while the earlier version remains available for traceability.

Withdrawal

A publication may be withdrawn when its central conclusion can no longer be responsibly supported or when a defect materially compromises the investigation.

A withdrawn publication should remain visibly identified as withdrawn rather than silently disappearing.


Revision records

Published investigations should maintain a visible revision history when meaningful changes occur.

A revision record should identify:

  • the publication version
  • the date of the change
  • the nature of the change
  • whether the conclusion changed
  • the reason for the change
  • the governing current version

Minor editorial changes that do not affect meaning may be grouped rather than listed individually.


Current correction record

No published Decision Atlas investigations currently require a correction, revision, supersession, or withdrawal notice.

This section will be updated when the first public investigation is released.


Report a possible error

Readers are encouraged to identify factual errors, broken sources, unsupported claims, meaningful omissions, or evidence that may materially affect a published conclusion.

When reporting a possible error, please include:

  • the investigation title
  • the statement or section in question
  • the reason you believe it may be inaccurate
  • the supporting source or evidence, when available

Accountability to reality

Decision Atlas does not treat publication as the end of an investigation’s responsibility.

A conclusion should remain open to correction when reality, stronger evidence, or better reasoning demonstrates that change is required.

Trust is not created by never changing a conclusion. Trust is created by changing it responsibly when the evidence requires it.

Scroll to Top