AnankeNetResearchExplore the documentation

Evidence-first research

See the pattern.
Preserve the truth.

AnankeNet explores how software can help qualified people examine unusual patterns in marketplace information—without confusing an analytical signal with a fact.

Understand the approach Read technical documentation

Private prototype · Synthetic validation only · Not operationally deployed

Review record Integrity visible
Preserved sourceOriginal representation

Origin · Capture time · Integrity reference

  1. 01
    ObservationExtracted fact
    Source-linked
  2. 02
    DerivedCalculated value
    Reproducible
  3. 03
    AnalysisExplainable signal
    Uncertain
  4. 04
    HypothesisPossible meaning
    Review required
Accountable boundaryHuman review

Inspect · Question · Disagree

Current status

Private research prototype

Synthetic validation only

No external approval claimed

Full project status

The research question

Patterns can inform a review.
They cannot make the judgment.

A single item can tell very little. Relationships across many items may raise useful questions—but a question is not a conclusion.

AnankeNet studies whether software can help reviewers notice unusual relationships, preserve the material behind an observation, and explain how an analytical result was reached.

The system is intended to support careful human review. It does not determine guilt, criminality, victims, or offenders.

What the architecture prioritises

Trust starts before analysis

Useful analytical output depends on a source that remains traceable, reasoning that remains inspectable, and uncertainty that remains visible.

01

Preserve the source

Original material remains separate from every later transformation and interpretation.

Evidence integrity
02

Keep reasoning traceable

Important output retains its origin, information layer, component, version, and history.

Provenance and audit
03

Keep judgment human

Reviewers can inspect the basis, see uncertainty, record disagreement, and correct decisions.

Human review

A disciplined information model

Every layer has a different meaning.

Keeping the layers separate lets a reviewer see what came from the source, what was calculated, where interpretation begins, and what remains uncertain.

Explore the information model
  1. 01
    Evidence

    Preserved source material

    Source
  2. 02
    Observation

    Facts extracted from evidence

    Fact
  3. 03
    Derived

    Values calculated from observations

    Calculation
  4. 04
    Analysis

    Explainable signals and limitations

    Interpretation
  5. 05
    Hypothesis

    A possible explanation

    Possibility
  6. 06
    Human review

    An accountable assessment

    Judgment

Clear by design

Built to assist.
Never to accuse.

Within scope

Support careful examination

  • Preserve source material and provenance
  • Surface unusual patterns for closer review
  • Show reasons, limitations, and uncertainty
  • Allow human reviewers to disagree

Outside scope

Make final determinations

  • Determine guilt or criminal conduct
  • Identify victims, offenders, or individuals
  • Contact or confront marketplace users
  • Replace qualified people or authorities

Go deeper

The right level of detail
for every reader.

For decision-makers

Purpose, safeguards, and limitations

A clear account of what the research explores, what it does not claim, and why accountability remains human.

Read the approach
For technical readers

Public architecture documentation

Technical principles for integrity, provenance, explainability, reproducibility, and review—without exposing sensitive internals.

Explore documentation

The working principle

Preserve facts.
Explain uncertainty.
Keep judgment human.

Research progress is meaningful only when the source remains traceable, the reasoning remains visible, and uncertainty remains intact.