Research record / 02 October 2026
Raven System
Activity Update
The evidence record keeps growing.
More retained source data, with a traceable boundary between collection and operational decisions.
01 / Since 4 September
What changed.
| Metric | 4 September | 2 October | Scope |
|---|---|---|---|
| UTC dates with retained source records | 204 | 232 | +28 dates; not uptime |
| Distinct payloads in one DNS dataset | 1 949 874 | 2 354 638 | +20.76%; same dataset definition |
| Rows after the September cutoff | — | 2 276 904 | 21 existing source files + 2 new host-specific paths |
| Confirmed collector families / domains | ≥11 / 10 | Historical baseline retained | No new full census |
| Paired LTE/fiber cycles | 501 | 501 historical | June 8–10; no new paired population certified |
The additional 2,276,904 rows comprise 2,111,911 rows in existing files and 164,993 in two new host-specific source files. These counts include failed probes and state records. They are not unique incidents, packets, or decisions.
02 / Retained source activity
How the record developed.
Monthly UTC-stamped rows in the fixed 23-file current source set. This is a different scope from the September archive-wide footprint. Collector coverage and cadence vary; a larger count does not establish better system performance.
| Month | Source rows | Boundary |
|---|---|---|
| 2026-02 | 12,908 | Variable collector coverage |
| 2026-03 | 17,977 | Variable collector coverage |
| 2026-04 | 1,361,409 | Variable collector coverage |
| 2026-05 | 1,012,466 | Variable collector coverage |
| 2026-06 | 2,288,353 | Variable collector coverage |
| 2026-07 | 964,425 | Variable collector coverage |
| 2026-08 | 593,546 | Variable collector coverage |
| 2026-09 | 2,327,086 | Variable collector coverage |
| 2026-10 | 131,372 | Partial through 2 Oct, 08:55 UTC |
The earliest producer-bound measurement remains 14 February 2026, 13:24:31 UTC. The 232-date historical union also includes 13 February retained records without the same producer binding; neither date count establishes uninterrupted operation.
03 / Evidence boundary
Collection is not decision validation.
All 23 captured source prefixes were stable on independent re-read; all 21 existing file prefixes matched the historical evidence. Current collection continues on two hosts. The additional host-specific sources preserve the replacement host’s identity.
This update establishes retained source-data activity. It does not establish an end-to-end Runtime decision total, customer validation, production readiness, decision lead, downtime reduction, or ROI.
The historical 501-cycle LTE/fiber comparison, Runtime processing populations, and collector/domain census remain scoped to the 4 September evidence release.
04 / Public evidence
Inspect the definitions.
Hashes identify the released bytes, not independent certification of the claims. Detailed source evidence remains private.