Interpretive guardrail
Consistency is not confirmation
A reconstruction may be internally consistent and still poorly supported. Conversely, a model can contain substantial evidence yet remain impossible because one hard contradiction survives.
Formal chronology laboratory
Audit the complete project knowledge graph, discover candidate discrepancies, enumerate distinct conflict cores, calculate global repairs, and trace every conclusion back through events, claims, models, and sources.
Diagnosis
Internal consistency, evidence coverage, and explanatory cost are reported independently.
Interpretive guardrail
A reconstruction may be internally consistent and still poorly supported. Conversely, a model can contain substantial evidence yet remain impossible because one hard contradiction survives.
Highest-priority findings
Whole-project status
Separate counts expose declared coverage, active contradictions, review candidates, evidence coverage, and rule governance without compressing them into a single score.
Canonical ingestion
Domain coverage
Fast configuration
Set the project’s major disputed propositions without opening the complete claim catalog.
Claim-level control
Each proposition can be accepted, rejected, or left undetermined. Uncertainty is not silently converted into rejection.
Exact chronology
Dates use astronomical year numbering internally: year 0 is 1 BCE, year −1 is 2 BCE, and year −1445 is 1446 BCE.
-0002-09-11September 11, 3 BCE0000-03-01March 1, 1 BCE0030-04-01April 1, 30 CE| Claim | State | Start | End | Resolved interval |
|---|
Active chronology
Minimal unsatisfiable cores
The engine removes supersets so each displayed set contains only claims necessary for that conflict.
All findings
Minimal change analysis
Repairs are ranked first by number of changed claims and then by declared explanatory cost.
Explainable reasoning
Inspect every rule, including satisfied, violated, underdetermined, inactive, and out-of-scope rules.
Select a rule to inspect its activation condition, constraint, provenance, and result.
Dependency structure
Accepted propositions and their active dependency paths are emphasized. Cycles are reported as circularity failures.
Select a graph node to inspect its evidence and downstream dependencies.
Canonical project graph
Explore source, project-claim, event, formal-claim, and model relationships imported from the canonical datasets.
Select a node to inspect its incoming and outgoing relationships.
Symmetrical testing
Every built-in model is processed by the same rule set; no model receives an automatic coherence bonus.
| Model | Status | Hard | Tensions | Unsupported | Open premises | Core premises |
|---|
Sensitivity analysis
Change one proposition, date, or model and compare exactly what breaks, survives, or becomes newly possible.
Findings removed
Findings introduced
Logical sensitivity
Measure single-point dependencies and neighboring configuration failures without assigning historical probabilities.
| Model | Base status | Core premises | Single-point failures | Sensitivity |
|---|
Support audit
Evidence coverage is separate from logical consistency and repeated arguments are not automatically treated as independent confirmation.
Repository integrity
Scans event, claim, source, model, and legacy-rule records for structural inconsistencies and unreachable rules.
Automated review prompts
Lexical, chronological, structural, and model differences are surfaced for human review. Candidates are never promoted automatically to hard contradictions.
Complete conflict structure
Enumerate materially distinct minimal unsatisfiable cores, measure overlap, and find repairs that resolve every active core rather than one conflict at a time.
Conclusion provenance
Ask why a conclusion appears, which source paths support it, what depends on it, and which single changes would make it disappear.
Incoming derivation paths
Dependents and removal effects
Rule lifecycle
Generated and authored rules retain lifecycle status, authority, scope, provenance, and version metadata so future revisions cannot silently change conclusions.
| Rule | Status | Authority | Scope | Introduced | Author |
|---|
Session compatibility
No-code rule authoring
Author nested activation conditions, temporal or state consequences, scopes, authority levels, and machine-readable rules without editing source files.
Revisioned research states
Save named revisions in IndexedDB, restore earlier states, compare revisions, and export a complete backup archive.
Reproducibility
Every report records the active states, date intervals, conflicts, tensions, and repair sets.
Logic trace