Observed activity
Returned records by UTC day. Missing days stay unknown.
GRAPHUS / DATA ATLAS
Every field.
A source and a limit.
Capture public BLCH (Bloch Genesis-4) records and inspect the calculations behind each field. Explore exact flows, temporal patterns and statistical diagnostics with the source and coverage attached to every result.
— catalog fields01 asset: BLCH01 network: Bloch Genesis-4— fields with results
01 / LIVE EVIDENCE
From capture to calculation.
One asset, one bounded collection. Every value carries its input digest, parameters and transform version.
Capture public BLCH source records.
UTC ranges include both endpoints. Records without timestamps are excluded only when a time filter is applied.
No comparison baseline pinned.
Source, anchors and reproducibility
No capture loaded.
Exact totals
Separate amounts for each recorded direction and event type.
02 / EXECUTABLE FIELD REGISTRY
Inspect every result.
Implementation describes the capability. Result state describes this capture. Select a field to inspect its value, definition, calculation and limits.
| Layer | Field | Meaning and boundary | Access | Implementation / result |
|---|
03 / RECORD LEDGER
The observations underneath.
Exact source records for the applied selection. Select a reference to inspect its local behavior.
| Event / transaction | From | To | Type / block | Exact base units |
|---|
No records
04 / MODEL EVALUATION
Measure a declared holdout.
Import evaluation JSON locally to measure precision at review capacity, recall, false positives, calibration and drift. Training and holdout entities must be disjoint, and evaluation times must follow the training cutoff. A report does not approve a model for release.
Evaluation files stay in this browser tab.
05 / EVIDENCE STANDARD
Depth is measured
by reproducibility.
Exact facts
Event identity, value units, finality, rollback, source revision, and missing-data flags survive every transform.
Clear inference
A transfer edge is not ownership. A shared service hub is not a criminal network. A model hypothesis is not a legal finding.
Measured models
Time and entity-disjoint tests, calibration, precision at review capacity, false-positive burden, drift, and independent review precede score release.
Correctable records
Source expiry, challenge, rectification, revocation, and audit are part of the data product, not afterthoughts.
GRAPHUS OF BLOCH INC
A Matter of Data 2.0
Hundreds of raw observations are useful only when their meaning and limits remain visible.