Burbank, Langley, Arlington

DARPA is the rarest of the three and the only one that does two unrelated jobs in this report. The larger share — 11 of 33 — is documentary, and it is recent. Section II carries an approach out of the DOJ production itself: across 2010–2013 Epstein pursued an introduction to Regina Dugan, Director of DARPA from July 2009 to March 2012, brokered by Bill Gates’s science adviser Boris Nikolic (You would like her a lot! We need to schedule a trip to DC sometime soon), kept alive by standing calendar alarms reading remind JE Regina, darpa, and planned to close at TED2012. The report states the strategic reading of that at hypothesis tier and publishes the finding that cuts against it — the interest continued after she left the agency. Nothing shows she met him or knew of the interest, and the report asserts nothing about her.

The other job is a clock, and it points the opposite way in time. Five mentions sit in Section XVI, four inside one subsection: the derivation asking what is running now and will be disclosed in 2040. The programmes it names there are dated rather than speculative — Big Mechanism (2012, machine reading of the scientific literature), the Biological Technologies Office (established 2014, as CRISPR entered wide laboratory use), and the N3 brain-computer interface line (2017–), which the report calls the successor to MKUltra’s failed attempts not through drugs and torture, but through engineering. The argument there is a clock rather than a claim: every declassified programme in the report’s own table stayed dark for two to three decades, so a 2013 cohort lands in 2033–2040.

The remaining mentions do other work again. DARPA appears in Section XVI as a credential that does not transfer: Lt. Col. John G. Blitch is genuinely documented — DARPA robotics, founder of CRASAR, ran the search-and-rescue robots at the WTC — and in 2025 UAP media he is the vetter, not the claimant. A credentialed person vouching does not convert testimony into documentation. In Section XXIV it carries the ALIAS autonomy programme into the Calabasas crash entry, where the report states the capability and then states the counter-evidence that defeats it. And in XXV it is the author disclosing the lens: Burbank, the GATE programme, and a DARPA mark on his own arm.

Lockheed is the opposite shape. Twenty-four of its 41 mentions are in Section VII, and thirteen of those are in a single subsection written for it. Where DARPA is split between a clock and a single named target, Lockheed is one argument the report works out in one place.

The claim is chronological, and the report corrects its own looser version of it first: the Skunk Works (1943) is not older than the OSS (1942). Lockheed the company is — Loughead Aircraft (1912) → Lockheed Aircraft Co. (1926) → the Gross reorganisation (1932), all predating the OSS (1942), the CIA (1947) and the NRO (1961). The private aerospace firm is older than the entire modern US intelligence apparatus — which was then built, in part, around capabilities the firm already had.

Then the documented instruments. The U-2 was a CIA programme, not the Air Force’s: Project AQUATONE, Eisenhower-approved November 1954, run by Richard Bissell, contracted directly to Kelly Johnson; Area 51 established April–July 1955 to flight-test it, and acknowledged by the Agency only in 2013. The A-12 OXCART was CIA-owned outright — twelve aircraft, operational 1965, declassified 2007. And Lockheed was system integrator on Corona, Gambit and Hexagon, the first spy satellites of an agency whose existence stayed classified until 1992.

The other half is the bribery record: ~$22M across six countries, exposed by the Church Subcommittee in 1975–76, taking down Prince Bernhard of the Netherlands, Japanese PM Kakuei Tanaka and Italian President Giovanni Leone — and producing the Foreign Corrupt Practices Act (December 1977). Today Lockheed Martin is the world’s largest arms producer for a fifteenth consecutive year (~$64.6B, SIPRI 2024), with an F-35 lifetime cost the GAO revised past $2 trillion.

Note the edge the report refuses to draw. A Lockheed↔Palantir collaboration is documented (2022). No Lockheed–Project Maven tie was found, Maven’s industry partner is Palantir directly, and the report says so rather than letting the adjacency stand in for a link.

CIA is less a subject in this report than a solvent. It appears in sixteen of the twenty-five numbered sections, once more in the front matter, and in 66 separate subsections, which means it is doing structural work rather than occupying a chapter — and a high count here is a fact about the report’s architecture, not evidence of anything.

The densest clusters are four. Section VII (62) holds the institutional machinery: the Cabinet Penetration Map across four administrations, the Task Force Orange / ISA entry on the unit that changes its classified codename roughly every two years, and Manufactured Leaders on the training pipelines. Section XVI (32) holds the declassified record itself — MKUltra, which survived only because 20,000+ documents were misfiled into financial records and escaped the 1973 destruction order — and the black-site network: 54 countries, 1,245 European Parliament-documented flights 2001–05, named sites confirmed by the European Court of Human Rights.

Section XX (24) is the Calibration Index, where the count is a scoring artefact rather than prose. Section XVIII (21) is the culture programme: the Congress for Cultural Freedom across 35 countries, Abstract Expressionism funded onto tour through MoMA’s International Council, Tom Braden’s 1967 I’m Glad the CIA Is ‘Immoral’, and the Agency’s own 100-work collection at Langley.

Two entries are worth reading for method rather than content. The two-ended camera in Section X pairs Burbank and Rochester — Lockheed built the U-2, Kodak’s Hawkeye Works made and processed the film it carried — and then refuses to draw the second line: the two industries’ collapses share no documented actor, so the report calls it a structural homology and says a homology is not a network edge. And in Section XVI the FOIA harvest logs the report’s own error at full resolution: it had claimed the NSA and DIA reading rooms were unreachable, which turned out to be a broken tool rather than a blocked agency. ‘The agency blocked us’ is a far more attractive sentence than ‘we used the wrong tool.’