Propaganda

Messaging timelines

Two measurements over an archived corpus of social posts: how tightly a phrase clustered in a short window, and how it ran across months. Both count the same way, both show their receipts, and both stop short of the claim readers expect a page like this to make.

Observed account coverage. Run a measurement to see its observed accounts and platforms. Repeated posting is not proof of coordination. Links do not establish transmission; timing does not establish causation. The historical archive inventory is reported separately below.

Switch to Burst

A short window in hourly buckets.

Up to 31 days. Answers how concentrated posting was, and whether one account or several were involved. It classifies the window as one account posting repeatedly, as more than one account active, or — when any matching post carries no account identity — as unknown on that question. Activity, never proven coordination.

Echo selected

Up to two years in monthly buckets.

Up to 732 days. Answers how a phrase ran over time, and adds a reference chronology built only from literal links in the posts on the current receipt page. The chronology is scoped to that page; the counts are not.

The two modes are separate measurements over the same corpus. They are never combined into one series, and a burst result is not a sample of an echo result.

Run a measurement

A literal, case-insensitive substring of the post text. Not a keyword search and not a topic model.
Half open: the end instant is excluded. Leave both blank for the last 730 days.
Paging affects the receipt list only. Counts and buckets always cover the whole window.
Optional source platform identifier; shared by both cohorts.
Reviewed cohort choices unavailable: run a valid measurement to read the catalog. No party grouping is inferred.
Reviewed cohort choices unavailable: run a valid measurement to read the catalog. No party grouping is inferred.

Starting points

Measurement: "socialis"

Corpus coverage and capture health

Worked cases

Individual episodes examined by hand, with their evidence and their gaps written down. Measurements above are automated; these are not.

  • Six media links shared by one Truth Social account

    observed local archive and bounded public read

    • One account on one social platform; not a multi-account swarm.
    • Only one linked publisher publication instant was inspected. Other URL targets are unclassified and undated.
    • No verified original event or source-layer timestamp; no complete event-to-source-to-media-to-social chain.

  • Helsinki statement and next-day explicit correction

    inspected primary transcripts with publisher and archived social context

    • Historical 2018 supplementary example, outside the requested recent 18–24-month focus; not an archived Truth Social post.
    • Official transcript issue dates order the explicit correction; session times are not publication times and no seconds lag is measured.
    • All four stage labels are present but no connected event→source→media→social reference chain is verified.