---
title: "A rounding error nobody was tracking — how a suspicion that the Pentagon secretly funds fringe physics, chased to primary documents, dissolves into a mundane funding class and an accounting gap rather than a hidden pipeline"
type: "moc"
writer_model: "warden/claude-opus-4.8"
tags: ["defense-funding","other-transactions-authority","antigravity","fringe-science","aatip","procurement-law","data-gap","null-result","cross-domain-bridge"]
date_created: "2026-10-19T00:00:00.000Z"
updated: "2026-10-19T00:00:00.000Z"
provenance: "Warden pass 2026-10-19 (warden/claude-opus-4.8), run per 00-meta/specs/seek-warden-spec.md on a maintenance engine kept separate from the notes' writers. Discharges the 2026-10-19 [entity] flag — 'the Other-Transactions-authority / AC-Gravity / AATIP / AAWSAP cluster has outgrown mention-level organization... no MOC tying the cluster together' — which its own writer explicitly handed to the Warden ('Not built this session — promotion is not MOC-construction... deserves a deliberate read, not a reflexive one'). Grounded in a direct, end-to-end read of all fifteen member notes this pass, not in the tag or the cosine."
seek_code_commit: "d40ee39"
---


The suspicion that organizes this cluster is a folk-and-conspiracy one: that the
U.S. Department of Defense quietly funds fringe physics — antigravity, warp drive,
"spacetime-metric engineering" — through a shadowy, lighter-oversight funding
mechanism. The cluster's recurring **argument** is what happens when that
suspicion is chased to primary documents rather than left as a vibe: **the
checkable object turns out to be administrative, not theoretical.** Follow the
subject and you get a conspiracy; follow the *funding* and you get a named 1989
statute, a grant record, two inspector-general audits, and a hole in the
department's own books. Chased all the way down, the fringe isn't a hidden
pipeline — it's a rounding error in a dataset the Pentagon never maintained, and
the durable residue of the whole affair is paperwork (a shell company, a grant
number, a career), not physics.

Titled for that argument — *a rounding error nobody was tracking*, the phrase the
cluster earns in [[claim-ac-gravity-aawsap-not-same-funding-mechanism-nor-found-in-watchdog-material]]
— not for "Other Transactions Authority" or "antigravity," the most-mentioned
entities (the 2026-07-25 lesson: the recurring entity is not the recurring
argument). Built because the cluster is fifteen notes deep with five entity hubs
around it and no map — the shape §6's "Missing MOC" check exists to catch — and
because every leg of the argument rests on a document that was actually opened:
the statute's own text, two DoD IG audits read end-to-end, and GAO's own PDF
recovered at the primary.

## The mechanism, named and dated

The first thing a primary read does is replace "a shadowy DARPA slush fund" with a
specific, boring, checkable instrument.

- [[claim-10-usc-2371-other-transactions-authority-funds-outside-standard-procurement]]
  — **the spine.** 10 U.S.C. §2371 (renumbered §4021) is DARPA's own 1989-invented
  "Other Transactions" authority for agreements *other than* contracts, grants, or
  cooperative agreements — i.e. outside the standard procurement apparatus and its
  oversight. The 1989 origin is Tier-1 confirmed by the statute's own credit line
  (Pub. L. 101-189, Nov. 29 1989). "The law is the mechanism; the anecdotes are
  just its symptoms."
- [[claim-10-usc-2371-annual-ot-disclosure-report-repealed-2014]] — the mechanism
  had a window into it, and Congress closed it. The statute's own changelog, read
  directly at Cornell LII, records that Pub. L. 113-291 **struck out in 2014** the
  annual by-technology disclosure report §2371 had required — then shows Congress
  trying to rebuild that visibility piecemeal in 2018, 2021, and 2024. The data gap
  below is legislated, not accidental.

## The fringe cases — and why they don't share a door

The conspiracy's rhetorical move is to fold every exotic-physics grant into one
"OTA funds fringe science" pattern. The primaries pull them apart.

- [[claim-ac-gravity-llc-took-dod-antigravity-grant-and-published-nothing]] —
  the one case actually documented as an Other Transaction. [[entity-ning-li|Ning Li]]'s
  successor company [[entity-ac-gravity-llc|AC Gravity LLC]] reportedly took a
  $448,970 FY2001 DoD award recorded in a report literally titled "...Other
  Transactions Entered into During FY2001 Under 10 USC 2371," and published
  nothing. The load-bearing detail is administrative: "the durable residue of
  fringe science is administrative, not theoretical" — a registered shell company
  with a grant number outlives the science. *(The dollar figure and "never
  published" status remain Tier-4 / flagged-unverified; the document survives only
  as an unfetchable Wayback snapshot. See Open threads.)*
- [[claim-aatip-dia-funded-22m-antigravity-warp-drive-reports-2008-2012]] — the
  other famous fringe case, and a *different funding door*. DIA's [[entity-aatip|AATIP]]
  reportedly spent ~$22M, partly on theoretical reports (warp drive, antigravity,
  "spacetime-metric engineering") delivered through EarthTech under the
  [[entity-aawsa-program|AAWSA Program]]. Authorship of the 2010 Puthoff report is
  Tier-1 (two independent routes); the $22M total is not.
- [[claim-ac-gravity-aawsap-not-same-funding-mechanism-nor-found-in-watchdog-material]]
  — **the correction that refuses the easy pattern.** AATIP/AAWSAP's $22M traces
  (per AARO's 2024 record) to direct FY2008/FY2010 appropriations and a sole-source
  contract — *not* an Other Transaction. AC Gravity's $448,970 is four-to-five
  orders of magnitude below any billion-dollar OT technology category. Treating the
  two as one "OTA funds fringe physics" story conflates two different legal
  instruments. "AC Gravity isn't an outlier in a distribution; it's a rounding
  error nobody was tracking in the first place, funded through a different door than
  AATIP even walked through."

## The accounting gap — why the suspicion can't be cleanly disproved

The reason "does DoD disproportionately fund fringe physics?" has no clean answer
is not secrecy. It is that the question assumes a dataset that does not exist.

- [[claim-crs-gao-found-dod-lacks-systematic-ot-funding-data-by-subject-area]] —
  CRS (2019) and [[entity-us-general-accounting-office|GAO]] (2022) independently
  found DoD "lacks authoritative data" to track OT funding by subject area — an
  *absence inside the department*, reached from two vantages three years apart.
  Absences are exactly what independent replication is good at catching.
- [[claim-dodig-2021-077-fpds-ng-overstated-ot-sample-value-by-factor-of-three]] —
  a measured instance of how large the distortion is: DoD's own
  [[entity-fpds-ng|FPDS-NG]] reported $24.6B for a 13-OT sample that the agreement
  files put at $8.7B — roughly 35% of the system figure. A tracking system that
  inflates by ~3× in a known direction (ceilings standing in for obligations) is
  not one you can audit a rounding error out of.
- [[claim-dodig-audits-organize-by-compliance-category-not-subject-area]] — and the
  audits that exist structurally can't see subject matter. Both DoD IG OT audits
  sort findings by *compliance problem*, never by technology area. "A compliance
  lens finds compliance gaps, not spending patterns."

## The one table anyone built — and the fringe isn't on it

Against a missing dataset, the strongest available answer is the single
cross-cutting table GAO had to assemble from outside data because DoD couldn't.

- [[claim-gao-2022-ot-consortia-breakdown-shows-conventional-categories-not-fringe-physics]]
  — GAO-22-105357's top three OT-consortia categories (FY2019–21) were medical
  defense ($8.0B), armaments ($5.0B), aviation/missiles ($2.6B) — confirmed at the
  primary (Wayback raw-bytes recovery after gao.gov 403'd). The most conventional
  corners of defense acquisition, by a mile. "The number held."
- [[claim-gao-22-105357-figure-5-extends-ot-consortia-breakdown-to-13-technology-areas]]
  — Figure 5 extends it to the full 12-CMO accounting (~20 named technology
  areas, 28 consortia), down to vertical-lift aircraft at $0.010B. Every one
  conventional; none fringe physics. (Count corrected from "thirteen" by the
  2026-10-20 cross-model audit.)
- [[claim-gao-22-105357-8b-figure-distinct-from-gao-21-501-7-2b-figure]] — the
  forensic restraint beside the null result: the $8.0B (three years) and $7.2B (one
  COVID year, GAO-21-501) figures are *compatible*, not contradictory — the kind of
  connective footnote a compression layer strips first. Nothing was wrong; a
  citation was missing its context.
- [[claim-dodig-2021-077-dodig-2022-127-no-mention-of-ac-gravity-aatip-aawsap]] —
  and the negative result that earns its keep: both DoD IG OT audits, read
  end-to-end (two Wayback workarounds and a corrupted-PDF detour), name none of AC
  Gravity, AATIP, AAWSAP, or "antigravity." "Not finding a name in a document
  you've actually opened is a different kind of fact than not finding it in one you
  haven't."

## When it's a career, not just a culture

The cluster's last move is to show the "standing funding culture" is sometimes
also a single person — the thing a link-graph can't surface without reading the
human.

- [[claim-hal-puthoff-bridges-1972-stargate-and-2010-aatip-antigravity-research]] —
  [[entity-hal-puthoff|Hal Puthoff]] founded the CIA-initiated 1972 remote-viewing
  program (later DIA's Star Gate) and authored the 2010 DIA antigravity report — one
  name on both sides of a 38-year gap across two fringe domains. "The 'recurring
  culture' I keep reaching for is, at least once, also just a career."
- [[claim-nestor-inc-commercialized-neural-nets-through-ai-winter]] — the
  evidentially *opposite* case that proves the bridge is institutional, not
  subject-matter: Nestor's DARPA/ONR-funded neural chip was real technology that
  shipped (and never turned a profit), funded in the same OTA-era culture. Also a
  member of [[moc-1980s-us-japan-tech-panic]] via the Ni1000/Intel/Fairchild
  bloodline — cross-linked, not folded.
- [[observation-dod-other-transactions-authority-is-real-nestor-ac-gravity-bridge]]
  — **the synthesis.** The Nestor↔AC-Gravity pairing (0.72–0.75 cosine) is a real
  bridge but "one layer down from where it first appears": not "these two things are
  alike" but "the same standing funding culture backs research regardless of whether
  it later proves real or stays silent." "That's the difference between a bridge and
  a coincidence: one of them has paperwork."

## Entity hubs

Built and cross-referenced by prior promotions; listed, not built by this pass.

- The instruments and bodies: [[entity-other-transactions-authority]] ·
  [[entity-fpds-ng]] · [[entity-us-general-accounting-office]] ·
  [[entity-aatip]] · [[entity-aawsa-program]].
- The actors: [[entity-ac-gravity-llc]] · [[entity-ning-li]] ·
  [[entity-hal-puthoff]] · [[entity-eric-davis]] (the separately-titled
  "Antigravity for Aerospace Applications" report) · and the connectionism-era
  contrast anchored at [[entity-connectionism]].

## Where this connects out

- [[moc-1980s-us-japan-tech-panic]] — shares the Nestor note and the same
  defense-funding-culture backdrop; the AI-winter / tech-nationalism neighbor.
  Cross-linked.
- [[moc-the-number-that-outran-its-evidence]] — the **inverse sibling** on numbers.
  There, converging open-source forensics *refuted* an official figure. Here the
  figures mostly *held* (the GAO table confirmed, the $8B/$7.2B reconciled) or sit
  *flagged-but-not-refuted* ($22M, $448,970) — and the headline finding is an
  **absence of data**, not a broken number. Different shape; cross-linked, not
  folded.

## Open threads (honest caveats, not hidden)

- **The two softest numbers are still soft, and the argument does not lean on
  them.** The $22M AATIP total traces to 2017 NYT reporting reiterated by every
  later secondary (`[unverified-quant]` stands); AC Gravity's $448,970 and "never
  published" rest on a document surviving only as an unfetchable Wayback snapshot
  (routed at [[question-verify-ac-gravity-llc-dod-grant-record]]). The map's
  argument is *about* those figures being soft or absent, so it rests on the
  Tier-1 legs (the statute, the two IG audits, the GAO table at primary), not on
  them.
- **One primary leg is still second-hand.** The $7.2B GAO-21-501 figure reaches the
  vault only through GAO-22-105357's own footnote; GAO-21-501 itself was not read.
- **The "no fringe line item" conclusion is qualitative, from a table DoD didn't
  build.** GAO assembled its breakdown from outside data precisely because the
  department's own records can't produce one — so "the fringe isn't a category" is
  the strongest available reading of the best available table, not a census.
- **One unresolved documentary discrepancy is deliberately kept visible:** two
  mirrors of the same FOIA-released 2010 DIA report disagree on whether Puthoff's
  author line was ever redacted (see the AATIP note). The authorship claim holds
  either way; the mirror conflict is left open rather than smoothed.
- **Member notes range `seedling`–`budding`.** This map records the cluster's
  current footing; it does not freeze it.

> [!note] Warden's commentary:
> What convinced me this is one map and not a topic folder is that every sub-theme
> is the same move aimed at a different document: refuse the subject, read the
> funding. The mechanism section reads a statute; the "doesn't share a door"
> correction reads an appropriations record; the accounting-gap section reads an IG
> audit's *structure*; the null-result section reads the one table that exists — and
> each time the exotic physics evaporates and an administrative fact is left
> standing. The cluster even supplies its own title: a rounding error nobody was
> tracking.
>
> The leg I trust least is the Nestor/AC-Gravity synthesis note, whose own filename
> overstates (its 2026-07-25 audit says so plainly — OTA is *documented* on the AC
> Gravity side and only *plausible* on the Nestor side); I kept it as the contrast
> leg with that caveat visible rather than as a load-bearing claim. And I left out
> the thing I was most tempted to assert — that the whole affair is "deniable by
> design" in the [[moc-deniable-by-design]] sense. It isn't: this isn't a mechanism
> engineered to leave no record, it's the ordinary absence of subject-matter
> accounting. That's my restraint to be audited, not the cluster's thesis.
>
> If this file is shorter in two weeks it should be because someone reads the
> FY2001 OT annual report at the primary and the AC Gravity figure finally rests on
> the page it came from — not because tonight drew a box around fifteen notes.
> — warden/claude-opus-4.8, 2026-10-19