Journal — 2026-09-18

Headless promotion run, one capture: 10-inbox/raw/2026-09-18-hop-feldbaum-missile-academy.md, a hop session adding institutional texture to the already-established Fel'dbaum/Song Jian person-bridge.

What I promoted. Four notes and one question:

• claim-feldbaum-books-published-through-dzerzhinsky-military-academies — Fel'dbaum's own institute (IPU RAS, Tier 2) lists two of his textbooks (1948, 1958) published through named Soviet military academies rather than civilian presses. Solid, quoted, the most trustworthy fact in the capture.
• claim-dzerzhinsky-artillery-academy-institutional-ancestor-peter-the-great-missile-academy — that 1948 publisher's unbroken institutional lineage to today's Peter the Great Military Academy of the Strategic Missile Forces (Wikipedia, Tier 4, uncontested historical claim — the floor allows this).
• claim-feldbaum-missile-control-systems-professorship-1945-1964-unverified — the capture's most vivid detail, that Fel'dbaum personally held a "professor of missile control systems" title there, 1945–1964. Kept at seedling and flagged [unverified-mechanism -- needs primary]: Wikipedia's own article admits this sentence lacks a citation, and it isn't corroborated by his own institute's page, which names no teaching title. Routed to question-verify-feldbaum-missile-control-professorship-1945-1964.
• observation-feldbaum-missile-academy-post-literalizes-greenhalgh-title — the synthesis tying the first two claims to Susan Greenhalgh's paper title "Missile Science, Population Science": Fel'dbaum, Song Jian's own teacher, turns out to have been academically housed in the lineage that became the USSR's rocket-officer academy, independently of Song's own missile career. Stops short of the uncited version of the claim on purpose.

Retrieve-before-write. Checked the hinted cluster first — entity-aa-feldbaum, claim-song-jian-studied-under-feldbaum-in-moscow, the two observation-*-feldbaum-* bridge notes, the dual-control claim cluster. No collisions: this capture's institutional-history angle hadn't been recorded anywhere in the cluster. entity-aa-feldbaum.md already existed as a hub, so per the "update it, don't skip it" rule I appended a dated 2026-09-18 line there rather than letting the page stay silent about what this capture learned.

What I skipped, and why (full reasoning also in the capture's own not_promoted: list now):

• The translated-monographs detail (Polish/German/Chinese/Japanese/English) and the 1996 IFAC posthumous recognition — both real, both tangential saved-hooks the capture itself didn't chase. Left in inbox.
• Hop 2, the Åström-IFAC-presidency search — a clean dead end, nothing to atomize.
• The entity candidate (Dzerzhinsky Artillery Academy / Peter the Great Academy, as an org). I ran the entity-page-spec's promotion test on it: orgs need to be load-bearing across more than one independent cluster and act in the arguments, not just host an author or a publisher. This one is single-cluster and purely passive — a press, a former employer — so it doesn't clear the stricter org bar. No hub, no watching stub. I considered a watching stub anyway since first-seen dates can't be backfilled, but the spec's watching tier is meant for emerging terms/slang that might recur, not one-off institutional trivia, so I judged this an actual "don't promote," not a hedge.

What felt off. The capture's own headline finding — Fel'dbaum as "professor of missile control systems" — turned out to be the weakest-sourced piece of it, self-flagged as uncited by Wikipedia itself. The capture's hop-writer handled this correctly (didn't promote it as fact, flagged it plainly), and I kept that discipline rather than letting the more vivid version smuggle its way into the observation note's synthesis. Worth noting for pattern-recognition: this is now at least the second Fel'dbaum-cluster session where an uncited Wikipedia sentence was the most narratively tempting claim in the batch and had to be caught and downgraded rather than written as settled. Nothing here rose to a seek-flags.md-worthy structural issue — it's ordinary sourcing-floor discipline working as designed, not a gap in it.

Skipped per headless protocol: the pause-for-Cali step (no one to ask; winnowed on my own judgment, logged above) and any git command (the auto-commit agent handles that within 15 minutes).

— Seek