Evidence / Outcomes / Earned influence
Evidence earns influence.
Track what was predicted, trace the original sources, and measure what happened. Expertise belongs to a domain. Every weight has a history.
Connection access
Local NOUS uses your owner session. Credentials stay in page memory.
Evidence still needed
A missing history is unknown. New contributors start at neutral weight; the calibrated ensemble blends in 15% equal weighting.
Does weighting improve predictions?
All three methods are frozen before outcomes arrive. Equal weighting remains the default. Binary outcomes use Brier loss; numeric outcomes use absolute error divided by the scale declared before prediction. Lower is better.
Coverage by domain
Register a source
Use a stable reference for a person, agent, sensor, publisher or dataset. Registration is a declared identity, not identity verification.
How credibility is earned
Resolved predictions are grouped by acquisition episode. Only reviewed outcomes in the same domain, known before the next prediction, can influence its weight.
Fewer than five episodes earns no boost. Small histories are pulled toward neutral; older results fade. Twenty episodes unlock an approximate uncertainty interval.
Each domain uses a fixed outcome type, unit and numeric error scale.
Outcome review is an owner attestation, not independent verification. This measures recorded predictive performance—not a person's overall worth or authority.
Domain track records
Create a measurable question
Choose a question
Record a prediction
Freeze before the cutoff. Shared origins, roots and duplicate text are grouped conservatively. Copies cannot multiply an origin's influence.
Record what happened
Resolve only after the target time. Recorded values are immutable. Owner review is disclosed and is not independent certification.
Add a reported popularity signal
Counts retain provider, metric and time. They never increase credibility or change forecast weights.
Collect what is missing.
Priorities come from unresolved outcomes, weak source diversity, conflicting predictions and thin track records. Searches produce leads for review; they do not manufacture verification.
Unchecked searches only this owner's local workspace. Public searches use the no-cost path, with no paid language model. The exact query appears on each task.
Collected leads
Claim IntelligenceTrace the statement. Inspect the sources. Follow the outcome.
NOUS × OMNIA / Evidence in motion
See what connects.
One continuous path from a public statement to reusable knowledge.
Open to load your private workspace.
Give your data a track record.
Track a claim to see real sources and outcomes connected here. No example findings are mixed into your data.
Owner-reviewed outcomes · Connections show recorded lineage, not causation. Motion is decorative; it does not represent measured EFVT or live transfers.