Workflow

From raw claims to a public record people can actually inspect.

The workflow matters because consensus is not a single model output. It is a chain of actions that turns messy inputs into a visible, revisable, time-ordered record.

Workflow arc

01

Intake everything without prematurely collapsing the field.

02

Build chronology before editorial certainty takes over.

03

Publish a current best account with uncertainty still visible.

Five stages

The workflow is a sequence, not a blur.

1. Intake

Gather reports, official statements, footage, archives, timelines, and derivative commentary into one working field.

2. Normalization

Standardize timestamps, entities, claims, and source metadata so comparison becomes possible.

3. Chronology assembly

Create a time-ordered record of events, updates, reversals, and narrative shifts before declaring what it all means.

4. Contradiction mapping

Identify where accounts diverge, what evidence supports each branch, and which claims remain unstable.

5. Public publishing

Present the current best account with attached evidence, revisions, open questions, and a visible trail of how the record evolved.

6. Ongoing revision

Keep the record alive as new reporting arrives rather than treating publication as the end of the process.

Output qualities

The workflow should produce records that are usable under pressure.

Traceable

Every important statement should lead back to a chain of sources and revisions.

Time-aware

The interface should make narrative evolution visible rather than presenting the present as timeless.

Revisable

The system must absorb corrections gracefully without erasing the memory of what came before.

Operational line

“Consensus gets stronger when the process is visible enough to be challenged.”

Once the workflow is clear, the next question is what kind of system architecture can support it.

Continue to systems