Noether

Philosophy

Valid evidence should survive independent recomputation rather than report editing, interface imitation or relabeling.

Architecture & control flow

1. 01

   Assign stable candidate identities

2. 02

   Run sealed challenges and actual v6.2 candidate audits

3. 03

   Bind settlements to predictions, tasks and adjudication

4. 04

   Recompute challenges, audits, settlements and index hashes

5. 05

   Issue conservative/demo/evidence tiers, fixing demo credit at zero

Architecture outline derived from this version’s control flow.

What this version changes

Repairs v6.5 integration gaps and demonstration semantics, strengthening evidence consistency.

Inputs & outputs

Inputs

Executable candidate parts, sealed tasks, preregistered predictions and evidence indexes.

Outputs

Challenge certificates, actual audits, settlement ledgers, stable indexes and recomputation reports.

Implementation & evidence scope

Default generation is still a metadata-only topic-interpolated candidate; this release primarily improves evidence governance rather than establishing a general scientific-discovery engine.

Code & bundled material

The introduction draws on bundled notes, changelogs and central code. Software tests, synthetic diagnostics and scientific effectiveness use different evidence standards.