# Epstein Coverage Audit: Protocol and Result Ledger

**Review snapshot:** August 23, 2026
**Status:** Independently replicated; original categorical conclusion withdrawn

## Question

Did a bounded sample of major U.S. news organizations merely report individual Epstein-related
people and events, or did any located treatment assemble the same cross-channel access mechanism:
funding, institutional legitimacy, brokers, introductions, meetings, and proposed transactions?

This is not a quantitative content analysis. It cannot establish what every newsroom published,
what appeared only on television or in print, what readers saw, or what an inaccessible archive
contains.

## Outlet sample

The audit sampled first-party pages from AP, Reuters, CBS News, CNN, NBC News, ABC News, NPR.org,
*The New York Times*, *The Washington Post*, and *The Wall Street Journal*. AP wire republication
is treated as one reporting act, not many independent newsroom decisions. In the original sample,
“NPR” meant the national `npr.org` domain, not the sites of NPR member stations. The independent
replication tested that boundary, found contrary member-station coverage, and the public result was
corrected rather than using the original domain boundary to preserve a negative claim.

## Time boundary

Searches covered material available through August 23, 2026. Older stories were included when a
later release or investigation linked back to them. The audit does not compare what was knowable
before and after each document release; it asks whether a system-level treatment was located by
the snapshot date.

## Query families

Each outlet domain was searched with combinations from these families:

| Finding | Names and distinctive terms used |
|---|---|
| August 2016 lunch | `Epstein Barrack Barak Churkin lunch`, `Pritzker October surprise`, `how close race Russians`, `Clinton emails entertaining` |
| Russia access | `Epstein Jagland Putin Lavrov`, `Vitaly Churkin Epstein`, `numbers in case of trouble` |
| Institutional platform | `Epstein IPI Rod-Larsen`, `Gates Foundation IPI Epstein`, `International Peace Institute Epstein funding` |
| Bannon relationship | `Epstein Bannon documentary interview`, `Bannon Epstein messages`, `heads of state callers Epstein audio` |
| Deripaska proposal | `Epstein Zeitlin Deripaska sanctions`, `trust swap option ADR debt`, `strawman Zug sanctions solution` |
| New Mexico | `Epstein Zorro Ranch FBI search`, `New Mexico prosecution investigation`, `truth commission` |
| Release completeness | `Epstein files missing pages attachments Signal`, `six million three million DOJ release`, `redactions prison video` |

Searches used both general search indexing with an outlet-domain restriction and each outlet's
first-party result pages when available. A search-result snippet was never counted as coverage by
itself. The auditor had to locate a first-party article, transcript, live blog, or broadcast page
whose body addressed the proposition.

## Inclusion and coding rules

- **Direct coverage:** the treatment states the material event, relationship, or mechanism.
- **Component coverage:** it states an important part but not the cross-channel mechanism.
- **Context absent in located treatment:** the page was inspected and the named context was not
  present. This describes that page, not the outlet's entire history.
- **Not located:** no qualifying first-party treatment surfaced under the documented query
  families. This is not coded as “never published.”
- Contact, access, a proposal, and an executed transaction remain separate categories.
- An outlet's later correction or updated treatment supersedes an earlier novelty assessment.

## Located-result ledger

| Finding | Located first-party coverage | Audit coding | Material limit |
|---|---|---|---|
| Barrack–Barak–Churkin lunch | CBS, “Tom Barrack and Jeffrey Epstein files” | Direct meeting coverage | The located page did not include the same-thread Pritzker election questions |
| Pritzker election-question thread | WBEZ Chicago and other NPR member-station reporting, located in independent replication | Direct document coverage outside the original national-domain sample | Defeats the original claim that this thread remained unlocated; the source record still establishes a question, not an answer or coordination |
| Jagland / Putin / Lavrov access requests | AP Jagland investigation; CBS release live coverage | Direct component coverage | No single located page assembled the funded brokerage model across channels |
| Gates / IPI / Rod-Larsen route | AP coverage of the Gates Foundation external review | Direct component coverage | Institution-triggered treatment; broader access-platform thesis not located |
| Bannon relationship and documentary | AP, CBS, and CNN treatments | Direct relationship coverage | Specific access-method audio passages were not located in the sampled coverage |
| Deripaska exploratory structure | CNN covered the Epstein–Zeitlin–Deripaska contact and Bannon forward; AP/CBS reporting addressed the separate Barker/Mercury Rusal delisting | No sampled treatment located the specific strawman, Zug, trust-swap, option, debt, and ADR structure | The exploratory proposal must not be represented as the actual delisting mechanism |
| Zorro Ranch and New Mexico delay | CBS 2019; later AP/CBS search, commission, and litigation coverage | Substantial direct coverage | Claims about total federal investigative absence require current-record verification |
| DOJ release incompleteness | CBS dedicated investigation; AP contemporaneous coverage | Direct, substantial coverage | Incompleteness does not prove what withheld records contain |
| Multi-leg access architecture | WSJ, CBS, AP, and Washington Post network/mechanism treatments | Substantial synthesis coverage | No sampled treatment assembling every channel defined by this audit was located |

## First-party pages in the current public sample

- AP: `https://apnews.com/article/7bdf92fd2d742f88789b2b5d67bf9f48`
- AP: `https://apnews.com/article/a947bbf2bd307658014862d01517d21f`
- AP: `https://apnews.com/article/798b07999092aae601eae11a8ea14ce5`
- AP: `https://apnews.com/article/df33d7b80ec6c0bdb2d55564ab3a59fc`
- AP: `https://apnews.com/article/98ee07a273865c1cfc2b034ef182171a`
- AP: `https://apnews.com/article/886132de2292c14cb055d2bff09c8ccf`
- CBS: `https://www.cbsnews.com/news/tom-barrack-jeffrey-epstein-files/`
- CBS: `https://www.cbsnews.com/live-updates/epstein-files-released-doj-2026/`
- CBS: `https://www.cbsnews.com/news/epstein-files-whats-missing/`
- CBS: `https://www.cbsnews.com/news/jeffrey-epstein-whats-next-for-zorro-ranch-where-accusers-say-they-were-raped-trafficked-2019-08-31/`
- CNN: `https://transcripts.cnn.com/show/cg/date/2026-02-05/segment/01`
- CNN: `https://transcripts.cnn.com/show/cg/date/2026-02-27/segment/02`
- Washington Post: `https://www.washingtonpost.com/world/2026/02/06/epstein-files-russia-putin/`
- Washington Post: `https://www.washingtonpost.com/politics/2026/02/08/jeffrey-epstein-cia-nsa-records-requests-foia/`

This ledger records the qualifying pages used for the material comparisons. It is not a complete
bibliography of every Epstein story published by the ten outlets.

## Known discoverability bias

CBS, AP, and CNN were substantially more discoverable than paywalled or robots-restricted outlets.
That asymmetry can make their work appear more prevalent. The audit therefore makes no outlet
ranking and no numerical claim about coverage volume.

## Replication gate

### Author rerun · August 23, 2026

The two highest-risk negative findings were rerun as paired five-outlet domain searches: AP,
Reuters, CBS, CNN, and NBC; then ABC, NPR, *The New York Times*, *The Washington Post*, and *The
Wall Street Journal*. The Pritzker query combined his name or “October surprise” with Epstein,
Barrack, Barak, and Churkin. The Deripaska query combined Epstein, Zeitlin, and Deripaska with
sanctions, strawman, Zug, or “trust swap.”

The rerun located CBS's Barrack treatment and release liveblog, but neither surfaced the Pritzker
election questions. It located *Washington Post*, CBS/AP, and CNN reporting about Deripaska
sanctions matters. CNN covered the underlying Epstein–Zeitlin–Deripaska contact and Epstein's
forward to Bannon, while a *Washington Post* Epstein/Russia treatment discussed a separate
Deripaska contact. None described the specific proposal structure at issue. CNN,
NPR, and *The New York Times* also produced explicit robots-access limits in the search tooling;
that limitation is preserved rather than silently counted as a negative result.

This author rerun strengthens the search ledger but is not independent replication.

The independent reviewer was required to rerun the query families, record every added or contrary
qualifying page, and answer two questions:

1. Does any located treatment assemble the cross-channel architecture claimed as missing?
2. Does any located treatment contain the Pritzker or Deripaska context then coded as unlocated?

The completed replication below answered the first question with substantial contrary evidence and
forced the public case to narrow its claim. The result ledger preserves the review date and
superseded language.

It also answered the second question in part against the author: WBEZ Chicago and other NPR member
stations published the Pritzker election-question text in November 2025. The national `npr.org`
domain search had not surfaced those member-station pages. Because ordinary readers understand
“NPR” more broadly than a domain restriction, the public case now treats the Pritzker finding as
covered and retains only the narrower Deripaska-structure negative finding.

## Independent replication · August 23, 2026

**Reviewer posture:** Adversarial replication. The reviewer did not assume the author's novelty
conclusion and searched specifically for reporting that would defeat it. Searches covered the ten
outlets and the same August 23, 2026 boundary.

### Reproducible searches

For each of `apnews.com`, `reuters.com`, `cbsnews.com`, `cnn.com`, `nbcnews.com`,
`abcnews.go.com`, `npr.org`, `nytimes.com`, `washingtonpost.com`, and `wsj.com`, the reviewer ran:

- `site:DOMAIN Epstein Pritzker "October surprise" Barrack Barak Churkin`
- `site:DOMAIN Epstein Zeitlin Deripaska sanctions strawman Zug "trust swap"`
- `site:DOMAIN Epstein Jagland IPI Rod-Larsen Bannon Barrack Deripaska`

Follow-up searches were:

- `site:wsj.com "Epstein Met Trump Associates" Barrack Thiel Churkin Rod-Larsen`
- `site:wsj.com Epstein "set up meetings" "Peter Thiel" "Thomas Barrack" "Vitaly Churkin"`
- `site:wsj.com Epstein network Jagland Rod-Larsen Putin Bannon Gates`
- `Epstein Pritzker "how close" race Russians Barrack Barak Churkin`
- `Epstein Pritzker "Clinton emails" entertaining Barrack Churkin`
- `Epstein Zeitlin Deripaska "strawman" Zug sanctions`
- `Epstein Zeitlin Deripaska "trust swap"`

Search-result snippets were treated only as discovery aids. The findings below are based on
first-party article text or an outlet's readable newspaper PDF.

### Contrary and qualifying coverage

The replication found that the existing phrase “numerous ... person-by-person treatments”
materially understates prior synthesis:

- WBEZ Chicago's November 19, 2025 report carried the Pritzker correspondence, including the
  election-question thread that the original national-domain search missed:
  `https://www.wbez.org/politics/2025/11/19/jeffrey-epstein-thomas-pritzker-emails-deeper-ties-jb-pritzker-cousin`.

- *The Wall Street Journal*, August 31, 2023, reported Epstein's use of relationships as
  “steppingstones” toward powerful people, the Rød-Larsen introduction to Churkin, the scheduled
  Barrack–Barak–Churkin lunch, and the Thiel–Churkin meeting in one treatment. It also preserved
  the important limits that a calendar does not prove every meeting occurred and that purpose was
  often unknown. This is a multi-node mechanism treatment, not merely a person profile:
  `https://www.wsj.com/public/resources/documents/JJ1zqihQoLz0w7wMXtJk-WSJNewsPaper-8-31-2023.pdf`.
- CBS's February 14, 2026 Barrack investigation combined the continuing relationship, invitations
  and introductions, the Signal prompt, assistance to Churkin's son, the August 29 lunch, a later
  Rød-Larsen/Thiel meeting, a roughly $1 million Colony Capital stock purchase, and exchanges about
  leaked Podesta emails:
  `https://www.cbsnews.com/news/tom-barrack-jeffrey-epstein-files/`.
- *The Washington Post* assembled Jagland/Putin access attempts, Belyakov, a separate Deripaska
  visa contact, assistance to Churkin's son, and Maria Drokova in one Russia-focused treatment:
  `https://www.washingtonpost.com/world/2026/02/06/epstein-files-russia-putin/`.
- The Post's “vast web” treatment expressly described efforts to pull levers of power and grouped
  Bannon, Barrack, and Barak:
  `https://www.washingtonpost.com/politics/2025/11/14/epstein-files-associates-names/`.
- AP reported Epstein's use of Jagland's and the Nobel institution's prestige in invitations to
  global elites, including the Rød-Larsen route:
  `https://apnews.com/article/a947bbf2bd307658014862d01517d21f`.

These treatments do not assemble every element in the proposed thesis—funding, institutional
legitimacy, brokers, introductions, meetings, and proposed transactions across the political,
diplomatic, and sanctions channels—but they do assemble important multi-node mechanisms.

### Two narrow negative findings

No qualifying sampled treatment was located containing Tom Pritzker's election-question text.
CBS covered the lunch and later Podesta-email exchanges but omitted the Pritzker questions.
Exact-phrase searches surfaced social and secondary pages outside the sample, not a first-party
page from one of the ten outlets.

No qualifying sampled treatment was located describing the specific Epstein/Zeitlin Deripaska
proposal structure, including the “strawman,” Zug, trust-swap, option, debt, or ADR details. CNN did
cover the underlying Epstein–Zeitlin–Deripaska contact and Epstein's forward to Bannon. The Post
covered a different Epstein/Deripaska visa contact. AP, CBS, and CNN results about Barker, Mercury,
Rusal, or EN+ concern the separate transaction that led to actual sanctions relief. An advocacy
page outside the sample and a document index surfaced the detailed Zeitlin proposal, confirming
discoverability but not sampled coverage of that structure.

### Access and discovery limits

The search tooling explicitly reported robots restrictions for CNN, NBC, NPR, and *The New York
Times* in multiple batches, although some indexed CNN transcripts and NPR pages surfaced. WSJ
article discovery was paywall-biased and weak; its full newspaper PDF was nevertheless indexed and
readable. Reuters, AP, and ABC searches returned no qualifying result for the two narrow findings.
All such outcomes remain coded **not located**, never “not published.”

### Replication verdict and required narrowing

1. **Complete architecture:** No located treatment assembled the complete six-part, cross-channel
   architecture as defined by this protocol.
2. **Contrary breadth:** WSJ, CBS, AP, and the Post did publish substantial network and mechanism
   treatments. This defeats any suggestion that coverage was generally confined to disconnected
   person-by-person facts.
3. **Pritzker and Zeitlin/Deripaska:** Neither specific context was located in the ten-outlet sample.

Case 002's novelty claim must therefore narrow to the integration of political introductions,
institutional funding/legitimacy, and the separate sanctions proposal. The categorical headline
“The architecture wasn't” should be withdrawn or qualified because multiple sampled outlets did
explain meaningful portions of the architecture. The defensible result is: **substantial network
reporting existed; no sampled treatment was located integrating every channel defined here.**