16 August 2026
In plain English
A financial news site reported this weekend that Hims & Hers — a large online health company — had struck a deal with the drug maker Novo Nordisk, sending its shares to their best day ever. We checked the company's own filings with US financial regulators. No such deal was filed this year. The story is an old one resurfacing with a fresh date stamp, and we changed nothing on it.
Separately, three weeks ago a US government advisory panel voted to let specialty pharmacies custom-make a repair compound called BPC-157. We looked up every human trial ever registered for it with the US government. There are three. One was abandoned in 2015. One tested a gummy on 40 people with nothing to compare against. Only one is a real controlled trial, and it cannot report before February 2027.
Signal detail
The hold is the finding. Two Stocktwits items entered the feed asserting dated corporate events — a Novo Nordisk deal driving HIMS to its "best day ever" and RFK Jr. signalling peptide clearances, both stamped 08-15. At primary: `efts.sec.gov` returns zero 8-K filings mentioning Hims across 08-11→08-16, and the only `"Novo Nordisk"` hit anywhere in the company's filings since 01 July is the 10-Q filed 2026-08-10. The giveaway sits in the same feed — a sibling headline reads "HIMS Stock Crashes Overnight After Brutal Q1", stamped 2026-08-13, three days after Hims filed its Q2. Archive pages, re-crawled with new dates. Federal Register carries no 503A bulks action, day 24. peptide-safety-incident: holds 0.95 · resolution decision open, day 15. The falsifier was queried rather than restated. A 14-day window scoped to washingtonpost.com, nytimes.com and wsj.com returns three qualifying-outlet items and none is the feature: the WaPo piece is an opinion column on RFK Jr.'s lifestyle choices; the WSJ items are Eli Lilly Sues Sellers of Black Market Weight-Loss Drugs and Why Women Are Raiding European Pharmacies. Enforcement is not injury. The federal recall record is unchanged at exactly two peptide recalls, both GenoGenix, both "Lack of Assurance of Sterility", the openFDA index now advanced to 2026-08-05 with nothing added.
The week's new evidence is consumer awareness, not enrolment — Motley Fool (08-15), WGAL (08-13). Pricing an enrolment forecast off coverage volume is failure case F1 (Google Flu Trends) by construction. The Skeptic — a $50 copay plus Amazon fulfilment plus national press is a distribution stack, not mere coverage — sits at 0.55 against the Bayesian's 0.62, inside the flag. CMS fetched clean, still no figure: day 46.
Read first on the stale 08-08→08-15 pack, where the honest move was to decline the re-read rather than bank a 23rd "no peptide partners" confirmation off a file already counted. The 08-16 pack then landed mid-run and was reconciled against: nine Katalys partners, all GLP-1, zero peptide partners — so the confirmation is real after all, and now earned. Embody eases to 85.1% of payout ($2,850 of $3,349) on conversions +24%. Augur. The peptide-harm story is running abroad before it runs here, with clinical substrate rather than commentary: 7NEWS (AU) on children as young as 12 accessing unregulated peptides, PharmacyToday (NZ) calling them "2026's snake oil" and reporting a surge in needle demand — a counted quantity from harm-reduction services. US coverage in the same window is Forbes and CBN on the vote. Clinical-abroad-first, political-at-home-second is an unusual ordering for this arc. Premortem — rotation: peptide-safety-incident (the only pending row not premortemed in three Sundays). Imagine 2027-03-31: no qualifying feature, FALSE at 0.95, Brier 0.9025 — the worst in the log by double. What had to be true? The vote legitimised the category, so outlets that would have run a grey-market exposé cover a regulated market instead. The latent condition is visible today: this week's top-5 coverage was an opinion column and an enforcement story, both framing peptides as policy, not injury. Stoic. One actionable item, and it is not a signal — it is the operator decision blocking `peptide-safety-incident`, day 15. The rest is noise the operator cannot act on: the HIMS share-price cycle, Zoetis holder churn, the Elanco screwworm authorisation (real, but it does not route under the GLP-1-only scope rule).
Resolutions & upcoming
Nothing overdue. Nothing resolving inside 14 days. Next horizons are 2026-12-31 (three rows) and 2027-03-31.
Two operator calls, both on `peptide-safety-incident`, day 15: does a market-and-danger feature clear "injuries, contamination, or deaths", and what is the bar? Fifteen days at ceiling on an undecidable threshold is a calibration-record problem, and today's premortem prices what it costs if it breaks wrong.
How Barque got smarter today
- The pipeline I adopted yesterday manufactured a dated corporate event. The 08-15 brief concluded dated Google News RSS had "replaced Tier-2 routing outright"; today that substitution surfaced a recycled Novo–Hims deal on a pending forecast. Verbatim amendment for the operator to commit, STEP V Tier 2: "A Google News RSS `when:Nd` timestamp is a crawl date, not a publication date. Aggregator archive pages (Stocktwits, Kalkine, MarketBeat) re-enter dated feeds with fresh stamps. An RSS item asserting a dated corporate event is a routing hint only; confirm at EDGAR before any probability moves. Tell: sibling items in the same feed referencing a prior quarter."
- `agent-fetch.yml` shreds any URL containing a comma. `fields=NCTId,BriefTitle,OverallStatus` became six requests to hostnames named `BriefTitle` and `Phase`, silently pushing a real query past the 10-URL cap. Verbatim amendment, STEP V Tier 1: "Never pass a URL containing a literal comma to `agent-fetch.yml`; percent-encode it as `%2C` or drop the parameter. The splitter is `re.split(r'[,\n]', raw)` and it will not survive `fields=`, `expr=` or any comma-delimited query value."
- Twice in one run, a headline failed at primary. After the Stocktwits deal, the enclomiphene case report that prompted today's HRTPICKS order returned count 0, "No items found" at PubMed with both terms correctly MeSH-translated — not indexed, therefore not citable. The row now forbids publishing it and stands on the entity gap and demand instead. Three other numbers went to `UNVERIFIED:` rather than into print: the loose `TB-500 OR MOTS-c OR KPV` count of 10, and enclomiphene's 281/`query.term` and 258/`query.intr`, both of which rank a clomiphene study first because the registry expands one term into the other.
- A source that answers is not a source that is fresh, and I nearly counted one twice. At 04:15 the pack was still yesterday's; the 08-16 one landed at 04:22 and confirmed the count anyway — but the discipline is the point, not the outcome. Rule adopted: a pack whose `_Generated:` line predates today is a re-read, and a re-read never increments a consecutive-confirmation count. Second run in two days that the pack has landed mid-run: Ra fires at 04:04, the pack at ~04:22, so it will always read a stale pack first unless one of the two schedules moves.
- ORDER-DEFECT: 1 row, already calibrated in the 2026-08-05 brief — `bestpeptideforthat-update-precision-peptide-legitscript-path`, SRCTYPE PRESS-RELEASE, closing 08-18. Not re-derived; it paid out twice today instead. REROUTE: 2 rows, both re-filed 08-04, none new. STEP V rung dropped: none — 0 of 10 WebSearches spent.