Confidence and Corroboration

Confidence estimates the strength of the evidence available for an event, cluster, report, or research conclusion.

It is not a universal truth score. It is a contextual assessment of the current evidence.

Main inputs

Confidence can incorporate:

source reliability;
source verification status;
primary versus secondary evidence;
data completeness and freshness;
independent corroboration;
consistency between sources;
entity resolution quality;
structured data quality;
number and diversity of supporting observations.

Social confidence is sensitive to source identity and corroboration. A reviewed corporate account can be stronger than an unknown account, but an official account can still be mistaken, compromised, or discussing an unconfirmed plan.

Unverified standalone social material receives stricter treatment and should not produce high-confidence publication solely because it is popular.

Cross-domain corroboration

Confidence is most useful when different evidence domains agree. Examples include:

a filing followed by market reaction;
an official announcement followed by independent reporting;
on-chain activity consistent with a disclosed protocol event;
a quantitative volatility signal confirmed by observed market behavior.

Correlated sources

Several reports may derive from one original source. NataPulse attempts to distinguish source count from independent corroboration, but users should inspect the evidence list.

Confidence in reports

Report confidence also depends on evidence completeness. A report can be withheld or warned when it has no sources, one source, thin content, low confidence, or missing required sections.

Confidence changes

Confidence can change as new evidence arrives. Early market intelligence often begins with uncertainty. A lower initial confidence is not a defect when the system correctly communicates that the evidence is incomplete.