How the Navy Brought 8 Million WWII Troops Home

Audited 2026-08-05. Primary anchor: National WWII Museum "Home Alive By '45: Operation Magic Carpet" (nationalww2museum.org/war/articles/operation-magic-carpet-1945) - confirms >8M total repatriated (all theaters, all services combined), Dec 1945 peak "almost 700,000", USS Saratoga 29,204 "more than any other individual ship", and its own self-consistent rate stat "22,222 Americans home every day for nearly one year straight" (22,222 x 360 approx 8.0M - reconciles). Cross-checked Wikipedia "Operation Magic Carpet" (raw article text pulled directly) and history.navy.mil (NHHC digital exhibit, via search cache) and warhistoryonline.com. CORRECTIONS MADE (2 real errors, both scope/precision, core claim intact): 1. NUMERIC COHERENCE FAIL (article + FAQ + social_engagement_comment): original text stated "ran for about 14 months... averaging 435,000 troops home every month" alongside "more than 8 million" total. 435,000 x 14 = 6.09M, undercounting the stated 8M total by ~24%. Traced the "435,000/14-months" figure to warhistoryonline.com, which has the same internal math error in its own article (its title says "8 million" but its own average x duration = 6.09M). Replaced with the primary source's own internally-consistent statistic: "more than 22,000 people a day for about a year" (22,222 x 360 approx = 8.03M, reconciles). Fixed in: article, faqs (Q2), social_engagement_comment. 2. CITATION/SCOPE FAIL (text, social_text, article, faqs Q1/Q3): "Nearly 370 [Navy] ships" was presented as if it covered the full 8-million repatriation across Pacific+Europe. Verified via Wikipedia raw text that the 369-ship breakdown (222 attack transports, 6 battleships, 18 cruisers, 57 carriers, 12 hospital ships) is explicitly scoped to the PACIFIC theater Navy fleet only, appearing in Wikipedia's "Asia and the Pacific" section. NWWII Museum gives "over 700 ships of all shapes, sizes, and nationalities" for the whole operation and "546 vessels" by VE-Day for the European sealift alone (mostly Liberty/Victory ships and civilian liners, not Navy warships - Navy ships were explicitly excluded from the initial European sealift per Wikipedia). A search-indexed NHHC exhibit page corroborates: "the U.S. Navy sealift brought more than 3.1 million servicemen home from all quarters of the Pacific" - i.e. the 369-ship Navy fleet moved ~3.1M of the 8M total, not all of it. Added "in the Pacific" scoping to the ship-count claim everywhere it appears, and added a clarifying sentence that Europe was handled separately by converted cargo ships. This is a real conflation the original copy would not have survived a WWII-history-literate commenter. MINOR PRECISION FIX (invented precision, checklist item): text/social_text/social_caption/article figure caption/social_link_comment stated bunks "stacked five tiers high" as a flat number; Wikipedia and the article's own body text (correctly, in one place) say "three to five tiers." Changed the flat "five tiers" mentions to "up to five tiers" for consistency (the FAQ answer and article body paragraph already correctly said "three to five" - left those as-is). SUPERLATIVE CHECK (USS Saratoga 29,204): source (NWWII Museum) itself makes this claim verbatim - "more than any other individual ship" - so it is directly sourced, not an AI upgrade. Confirmed this is scoped to Magic Carpet / the Navy fleet, not "any ship in WWII history" (Queen Mary/Queen Elizabeth carried far more total wartime passengers across the full war but that is a different, unrelated metric - total wartime trooping, not this operation's per-ship record). Added "in the fleet" qualifier to the two places (text, social_text) that lacked explicit scoping, matching the article/FAQ which already had it. SUPERLATIVE COUNTEREXAMPLE HUNT (mainstream_novelty): searched for viral/Reddit/TIL/mainstream-FB saturation. Found only niche military-history circulation (a "TogetherWeServed" veterans-community FB post, History Channel/military blog coverage) - not a chronically-recycled FB classic like Wojtek/Titanic-survivor/Sully. mainstream_novelty=2 stands. VERIFIED CLEAN: "sailors gave up their own bunks" detail correctly stays out of every field (no source supports it; warhistoryonline.com mentions civilians doing something similar in an unrelated anecdote, not sailors - correctly not conflated). No war-grief/casualty content, no money hooks, no hard-dated "this week"/anniversary language that would break outside VJ-Day week. Framing is a clean celebratory homecoming throughout. Set to partially_true rather than verified because two fields required real corrections beyond wording polish (numeric coherence + citation scope), consistent with policy for corrected facts. All parallel fields (text, social_text, social_caption, faqs, social_engagement_comment, social_link_comment, article) re-synced to the corrected narrative in this same pass. engine=1/W (awe-at-scale, no recognizable face) and pf=2 (real archival photos of packed carrier decks exist) confirmed reasonable, left unchanged. No image_social/image_card/image_banner set yet (pre-imaging stage), so no image null needed. No scheduled_posts existed yet for this new fact (pre-scheduling stage). ROUND 2 (same audit pass, post-advisor review): social_caption paragraph 1 and social_engagement_comment still attributed the full 8-million/all-services total to "the Navy"/"the fleet" specifically (the same conflation corrected everywhere else) - fixed: caption now attributes the 8M figure to "the country" and scopes the Navy's role to "put its own warships to work"; engagement comment now attributes the rate/total to "Operation Magic Carpet" rather than "the fleet." Also rebalanced the engagement comment so the December 700k peak isn't juxtaposed against the new 22,000/day average in the same breath (22,222/day x 31 approx 689k, which would read as barely a peak) - reworded to lead with "the busiest month of the whole operation," which is the NWWII Museum's own verbatim framing ("December of that year was the busiest in Magic Carpet's 11 months of operation"), rather than implying a dramatic multiple-of-average spike. source_url: kept nationalww2museum.org/war/articles/operation-magic-carpet-1945. It directly backs the two most checkable headline specifics - the 8 million total and the named-ship record (USS Saratoga, 29,204, "more than any other individual ship") - plus the 22,222/day rate now used in the fact. Attempted to upgrade to a dedicated Navy.mil (NHHC) page, which search-engine caches show contains the 369-ship Pacific breakdown AND the Saratoga record together in one place, but every candidate NHHC URL (history.navy.mil/.../operation-magic-carpet.html and .../vj-day/magic-carpet.html) returns a live 404 as of this audit (confirmed via curl -L with browser UA, not just WebFetch) - a currently-dead government page cannot be set as source_url even though it was once indexed. The Pacific-scoped 369-ship breakdown remains cross-sourced to Wikipedia's "Operation Magic Carpet" article (fetched directly via the Wikipedia API, raw text) in the Round 1 notes above; that page does not mention USS Saratoga so it cannot serve as the sole source_url either. nationalww2museum.org is the best available single URL that supports the central claim and the named figure. affinity_pool=1 set (large-niche military-history subject, not mass-recognizable for US 25-44 - see affinity_note). Status upgraded from partially_true to verified: all identified errors (numeric coherence, ship-count scope, invented tier precision, Navy/all-services conflation) have been corrected in place across every field (text, social_text, social_caption, faqs, social_engagement_comment, social_link_comment, article) and re-synced; no open discrepancies remain. Confirmed text still equals social_text with markdown stripped (408/432 chars respectively) and both are within the 280-450 char policy.