Disclosure
Tier 2finding · june-15-silent-ocr-republish-was-part-of-r4-prep

The June-15 silent OCR republish was almost certainly R4 pipeline pre-staging

Finding 23 ("june-15-silent-ocr-republish") documented one PDF from R1 that was silently re-OCR'd and quietly republished on June 15. R4 arrived on July 10 with all 14 of its PDFs going through the same OCR pipeline (Adobe Acrobat 26 Paper Capture Plug-in) and using a single new vendor (Highland Technologies, Inc.). Reading the R4 metadata backwards, the June-15 quiet update was almost certainly a test-fire of the new pipeline on one file before the full R4 batch.

Confirms the June-15 silent republish wasn't a one-off — it was a staging step for a broader R4 OCR/re-processing effort. The R4 batch went through the same Acrobat Paper Capture Plug-in that Finding 23 flagged, produced by the same Highland Technologies vendor stack. This is a useful piece of provenance: it tells us the DoW's tradecraft-cleanup wasn't just retrospective (fixing R1-R3 leaks in R3), it was also prospective (test-driving the R4 pipeline on live files).

Evidence

The June 15 quiet update (from Finding 23, [june-15-silent-ocr-republish](/findings/june-15-silent-ocr-republish)): - Only release_1/59_214434_sp_16_[7.18.1963].pdf changed between June 14 and June 15 - Went from image-only (0 chars text) to OCR'd (12,845 chars text) - New /Producer: Adobe Acrobat (32-bit) 26 Paper Capture Plug-in - No catalog announcement

R4 (July 10) metadata for all 14 PDFs: - /Producer: Adobe Acrobat (32-bit) 26 Paper Capture Plug-in (exact same string) - /Creator: Highland Technologies, Inc. (single vendor across all 14 files — see [r4-pdf-author-field-now-populated](/findings/r4-pdf-author-field-now-populated))

Interpretation. The June-15 file gained the Acrobat Paper Capture pass but *not* the Highland Technologies vendor stamp — the Creator field on the June-15 file is empty, not "Highland Technologies." That suggests the June-15 push was just the OCR stage of a two-stage pipeline; the R4 batch on July 10 added both the OCR pass and the Highland Technologies re-render/re-save stage. Reading forward: expect future silent republishes to add Highland Technologies to the Creator field, marking them as fully-processed through the new R4 pipeline. Watching those tags is the cheapest way to detect pipeline processing on old files.

Silent-republish sweep on 2026-07-10 (post-R4). Ran the same byte-level HEAD diff against all 424 mirror-referenced PDFs/images/etc. All 424 unchanged. No silent tweaks were bundled with the R4 push — the R4 processing pipeline is being applied only to R4 files, not backfilled to R1-R3.

Cross-refs: - [june-15-silent-ocr-republish](/findings/june-15-silent-ocr-republish) — the trigger event - [r4-pdf-author-field-now-populated](/findings/r4-pdf-author-field-now-populated) — the full R4 pipeline stamp - [tradecraft-cleanup-in-release-3](/findings/tradecraft-cleanup-in-release-3) — the earlier R3 cleanup that presaged this

Sources

www.war.gov/medialink/ufo/071026/release_04/documents
R4 PDFs (fully OCR'd + Highland Technologies stamp)