Tuesday, July 28, 2026

Orange Fanta in Glass: Why It's Better


There is nothing like it: bright orange, springing to life, pulling attention, holding the eye.  Orange Fanta in glass: bright and happy and beckoning.  

Now we are somewhere else.   We feel uncomfortable; something is going to go wrong.  It is a can.  It is a can of orange Fanta, but it is a can.   Someone pulls the tab.   ...Someone takes a sip.   It is tart, but lacks flavor.   Color, perhaps, but not zest.  This is Fanta in a can.

We used to think you could not taste the difference.   Well, some told you that, but it never stuck, because they did not know.  Orange Fanta in glass; the best ever.  The greatest of all time.

So you know you can taste the difference, but why?   The glucose, the malic acid, the citric acid, the faint acids that don’t have names, they are the life of Fanta in Glass.  

In a can, the organic acids go bad.  They go rogue.  They leach out metals.  They leach out copper and tin.  From an alloy can, they leach out tungsten, gallium, vanadium, and in Siberian cans, polonium.   

Orange Fanta in cans is bad for you.  Orange Fanta in bottles is bright and alive and can only leach out hope and happiness.


 



x

Sunday, July 19, 2026

STAT on Paying for AI at CMS

 

Paying for AI: Three Case Studies in How Reimbursement Shapes Clinical Adoption

The Real Question Is Not Whether AI Works

Clinical AI is often discussed as a contest of accuracy: Which algorithm finds disease earlier, reads images better, or predicts deterioration more reliably? Katie Palmer’s three-part STAT series, Paying for AI, shifts the question. Even an effective algorithm must fit into hospital workflows, produce a financial return, and survive a Medicare payment system designed largely around physical supplies, clinician time, and conventional procedures.

Together, the articles show that payment does not merely reward adoption after the evidence is established. It can determine which products reach patients, how broadly they are deployed, and what downstream care they generate.

##

  1. Palmer, Katie. 2026. “AI Could Check Millions of CT Scans for Heart Risk. Who Will Pay for It?” Paying for AI, Part 1. STAT, April 15, 2026.
    https://www.statnews.com/2026/04/15/coronary-artery-calcium-ai-opportunistic-screening-examined/ (STAT)

  2. Palmer, Katie. 2026. “In the Battle of Sepsis Algorithms, Performance Alone Doesn’t Predict Victory.” Paying for AI, Part 2. STAT, May 12, 2026.
    https://www.statnews.com/2026/05/12/ai-sepsis-detection-startups-challenge-epic-systems/ (STAT)

  3. Palmer, Katie. 2026. “CMS Signals Intent to Revamp How It Pays for Clinical Software and AI.” Paying for AI, Part 3. STAT, July 16, 2026.
    https://www.statnews.com/2026/07/16/cms-to-revamp-payments-for-clinical-software-ai/ (STAT)

##

Part 1: Finding More Disease—and More Spending

The first article examines AI that detects coronary calcium incidentally on chest CT scans obtained for other reasons. Roughly 19 million such scans are performed annually, and an estimated 20% to 40% of visible calcium findings go unreported. Medicare now pays about $15 for certain algorithmic analyses.

This appears to be an ideal use of AI: no additional scan, little marginal computational cost, and an opportunity to identify cardiovascular risk earlier. Yet the economics are complicated. Positive findings may lead to statins, but also stress tests, coronary CT angiography, invasive angiography, and procedures. Hospitals may gain substantial downstream revenue without proof that population-wide deployment reduces heart attacks or mortality.

The article calls this risk “biomarkup”: technology continually creates new measurable abnormalities that generate additional medical activity.

Part 2: The Best Algorithm May Not Win

The second article turns to sepsis prediction. Epic’s first model performed poorly in practice, but its successor remains formidable because Epic is already embedded in more than 40% of U.S. hospitals. More broadly, EHR companies supply an estimated 79% of predictive algorithms used by hospitals.

Competitors may demonstrate earlier or more accurate detection, but hospitals also weigh integration cost, implementation time, regulatory status, and reimbursement. Bayesian Health pursued FDA clearance and may qualify for approximately $62 in supplemental Medicare payment per analyzed patient.

The lesson is blunt: superior performance alone does not overcome an incumbent platform that is inexpensive and easy to activate.

Part 3: CMS Begins Building a Framework

The third article describes CMS’s proposed 2027 policy. CMS would adopt the term Software as a Medical Service, move 21 separately paid software services into New Technology APCs, and identify them with a new O1 payment status.

CMS is also considering whether software should be discounted when billed with another service and whether payment should reflect outcomes or downstream savings. Ten algorithm-related laboratory codes would move toward similar treatment, raising difficult questions involving contractor pricing, CLIA oversight, coding, and the boundary between laboratory testing and stand-alone software.

Pulling the Series Together

The trilogy presents payment as part of AI’s clinical architecture, not an afterthought. Separate reimbursement can accelerate useful innovation—but can also reward excessive detection and downstream utilization. Bundling controls spending but may leave hospitals unwilling to adopt worthwhile tools.

The key takeaways are that workflow can matter more than accuracy, incumbency can outweigh innovation, and CMS increasingly wants evidence that AI improves outcomes or lowers total costs—not merely that another algorithm can produce another billable result.


##

Sidebar: Five Surprises in Paying for AI

1. A $15 AI analysis could trigger tens of millions of dollars in downstream care.
The coronary-calcium algorithm itself is inexpensive. Its financial importance lies in what follows: statins, stress tests, coronary CT angiography, invasive angiography, stents, and bypass surgery. Vendors estimate that this follow-up activity could generate tens of millions of dollars for a large health system.

2. More than half of patients screened by one system appeared to have significant coronary calcium—but most did not need another intervention.
In one health system, 63,000 of 120,000 patients had moderate or severe calcium scores. Yet an estimated 85% might not merit notification because they were already receiving appropriate treatment or had more urgent illnesses. The challenge was no longer finding abnormalities; it was deciding which findings mattered.

3. A technically better sepsis algorithm may still lose to the one already inside the electronic health record.
Epic’s market position can outweigh comparative performance. Hospitals often choose its algorithm because it is inexpensive, familiar, and easy to activate—even when an outside vendor may detect sepsis earlier or more accurately. An estimated 79% of hospital predictive algorithms come from EHR vendors.

4. Some clinical-AI companies have abandoned patient-care tools for billing software.
Palmer reports that the difficulty of selling regulated clinical prediction products has led some companies to shift toward more lucrative, less-regulated tools that help hospitals document or defend reimbursement. The economics can pull innovation away from diagnosis and toward revenue-cycle management.

5. CMS may move some algorithmic laboratory services outside the traditional laboratory payment system.
CMS proposes treating certain analyses of laboratory-generated data more like imaging AI, with hospital payment through New Technology APCs and physician-office pricing by regional contractors. That creates a striking regulatory “no-man’s-land”: the algorithm may be neither clearly governed as a CLIA laboratory test nor clearly regulated as an FDA device.

Saturday, July 18, 2026

German Verbs with REISEN (Deutsch, dann Englisch)

In Podcast 659, in the summary, we read, "Abschluss erzählt Cari von ihrem Plan, die ganze Welt zu bereisen." OK, so how many verbs can we make out of "Reisen"? Are they separable or inseparable? Here is an AI discussion, first in immersion German and then the German verbs taught in English.  

##

DEUTSCH
Podcast 659: Wie viele Verben kann man aus „reisen“ bilden?

In der Zusammenfassung von Podcast 659 lesen wir:

„Zum Abschluss erzählt Cari von ihrem Plan, die ganze Welt zu bereisen.“

Dabei fiel mir eine interessante Frage ein: Wie viele Verben kann man eigentlich aus „reisen“ bilden? Welche haben trennbare Vorsilben, welche untrennbare?

Ich habe ChatGPT danach gefragt. Unten finden Sie zuerst eine Erklärung auf Deutsch (für Lernende auf B1/B2-Niveau), danach dieselben Verben mit englischen Erklärungen. So können Sie je nach Lernstil beide Versionen vergleichen.


Eine feste Gesamtzahl gibt es nicht, denn im Deutschen kann man mit reisen viele neue Verben bilden. Für Lernende auf dem Niveau B1/B2 sind ungefähr 12 bis 15 häufige Formen besonders nützlich.

Verben mit untrennbarer Vorsilbe

Bei untrennbaren Verben bleibt die Vorsilbe immer beim Verb. Im Partizip II steht kein ge-.

bereisen

bereisen bedeutet: durch ein Land, eine Region oder mehrere Orte reisen.

  • Cari möchte die ganze Welt bereisen.

  • Sie hat schon viele Länder bereist.

bereisen braucht meistens ein Akkusativobjekt:

  • ein Land bereisen

  • Europa bereisen

  • die ganze Welt bereisen

Im Infinitiv mit zu heißt es:

  • die Welt zu bereisen

Nicht: be-zu-reisen

verreisen

verreisen bedeutet: für einige Zeit von zu Hause wegfahren oder wegfliegen.

  • Wir verreisen nächste Woche.

  • Meine Nachbarn sind gerade verreist.

  • Sie plant, im Sommer zu verreisen.

Auch ver- ist untrennbar. Deshalb heißt es:

  • verreist, nicht geverreist

  • zu verreisen, nicht verzureisen

Verben mit trennbarer Vorsilbe

Bei diesen Verben wird die Vorsilbe im Hauptsatz abgetrennt:

  • Wir reisen morgen ab.

Im Infinitiv mit zu steht zu zwischen der Vorsilbe und dem Verb:

  • morgen abzureisen

Im Partizip II steht meistens ge zwischen der Vorsilbe und dem Verb:

  • abgereist

abreisen – einen Ort verlassen

  • Der Zug fährt um acht Uhr, deshalb reisen wir früh ab.

  • Wir sind gestern abgereist.

  • Es ist Zeit, abzureisen.

anreisen – an einen Ort kommen

Dieses Verb wird besonders bei Veranstaltungen, Hotels oder längeren Fahrten benutzt.

  • Die Gäste reisen am Freitag an.

  • Viele Teilnehmer sind mit dem Zug angereist.

  • Wir planen, einen Tag früher anzureisen.

einreisen – in ein Land kommen

  • Für die Einreise braucht man einen Pass.

  • Die Touristen reisen über Frankfurt ein.

  • Er durfte nicht einreisen.

ausreisen – ein Land verlassen

  • Sie möchte nächste Woche ausreisen.

  • Die Familie ist bereits ausgereist.

  • Er bekam die Erlaubnis, auszureisen.

einreisen und ausreisen werden häufig im Zusammenhang mit Grenzen, Pässen und Visa verwendet.

mitreisen – gemeinsam mit jemandem reisen

  • Darf ich mitreisen?

  • Ihre Schwester reist auch mit.

  • Zwei Freunde sind mitgereist.

nachreisen – später zu jemandem reisen

  • Seine Familie kommt erst später und reist ihm im Juni nach.

  • Ich bin meinen Freunden einige Tage später nachgereist.

  • Sie plant, ihrer Gruppe nachzureisen.

weiterreisen – die Reise fortsetzen

  • Nach zwei Tagen in Berlin reisen wir nach Hamburg weiter.

  • Am nächsten Morgen sind wir weitergereist.

  • Wir beschlossen, sofort weiterzureisen.

zurückreisen – wieder zum Ausgangsort reisen

  • Am Sonntag reisen wir nach Hause zurück.

  • Sie ist bereits nach München zurückgereist.

  • Es war schwierig, früher zurückzureisen.

hinreisen – zu einem genannten Ort reisen

  • Wie reist ihr nach Wien hin?

  • Wir sind mit dem Nachtzug hingereist.

  • Es lohnt sich, im Frühling hinzureisen.

herreisen – zu dem Ort reisen, an dem der Sprecher ist

  • Bist du nur wegen der Konferenz hergereist?

  • Viele Menschen reisen aus dem Ausland her.

Der Unterschied:

  • hinreisen: vom Sprecher weg oder zu einem anderen Ort

  • herreisen: zum Sprecher oder zu seinem jetzigen Ort

losreisen – eine Reise beginnen

  • Wir reisen morgen früh los.

  • Sie sind schon um sechs Uhr losgereist.

  • Ich freue mich darauf, endlich loszureisen.

zusammenreisen – gemeinsam reisen

  • Wir reisen gern zusammen.

  • Sie sind mehrere Monate zusammengereist.

  • Es kann schwierig sein, lange zusammenzureisen.

Ein besonderer Fall: durchreisen

durchreisen kann auf zwei Arten verwendet werden.

Trennbar: durch einen Ort fahren, ohne lange zu bleiben

  • Wir reisen nur durch.

  • Wir sind ohne Übernachtung durchgereist.

Hier bedeutet es ungefähr: einen Ort nur auf dem Weg passieren.

Untrennbar: ein Gebiet vollständig bereisen

  • Sie haben das ganze Land durchreist.

  • Er möchte Südamerika durchreisen.

Hier hat das Verb meistens ein Akkusativobjekt. Das Partizip lautet durchreist, ohne ge-.

Die wichtigsten Formen im Überblick

Für den aktiven Wortschatz sind diese Verben besonders nützlich:

Untrennbar:

  • bereisen

  • verreisen

Trennbar:

  • abreisen

  • anreisen

  • einreisen

  • ausreisen

  • mitreisen

  • nachreisen

  • weiterreisen

  • zurückreisen

  • hinreisen

  • herreisen

  • losreisen

  • zusammenreisen

Dazu kommt durchreisen als besonderer Fall, weil es je nach Bedeutung trennbar oder untrennbar sein kann.

#
#

ENGLISCH

There is no fixed total, because German can build many new compounds with reisen. But there are about 10–15 common ones, plus several less common combinations.

Inseparable prefixes

These do not insert ge- in the past participle:

  • bereisen – to travel around/through a place
    Sie hat die ganze Welt bereist.

  • verreisen – to go away on a trip
    Sie ist für zwei Wochen verreist.

Your sentence means:

Zum Abschluss erzählt Cari von ihrem Plan, die ganze Welt zu bereisen.
Finally, Cari talks about her plan to travel all over the world.

Notice zu bereisen, not zu be-reisen: be- is inseparable.

Common separable-prefix verbs

With these, zu and ge normally go between the prefix and reisen:

  • abreisen – depart
    Sie reist morgen ab. / abzureisen

  • anreisen – arrive by travelling
    Die Gäste reisen heute an.

  • ausreisen – leave a country

  • einreisen – enter a country

  • mitreisen – travel along with someone

  • nachreisen – follow someone later

  • weiterreisen – continue travelling

  • zurückreisen – travel back

  • hinreisen – travel there

  • herreisen – travel here

  • losreisen – set off

  • zusammenreisen – travel together

A special case: durchreisen

It can behave in two ways:

  • separable: Wir sind ohne Aufenthalt durchgereist.
    We travelled straight through.

  • inseparable/transitive: Wir haben das ganze Land durchreist.
    We travelled throughout the entire country.

So, for practical B1-level German, I would learn 14 useful forms: the 12 separable ones above, plus bereisen and verreisen

Thursday, July 16, 2026

The Shawshank Redemption—and Mr. Throttlebottom-Smith

 

The Shawshank Redemption—and Mr. Throttlebottom-Smith

The accepted history of The Shawshank Redemption is familiar. Andy Dufresne, a banker unjustly convicted of murder, enters Shawshank State Penitentiary, befriends Ellis “Red” Redding, survives two grim decades, and eventually escapes through a tunnel concealed behind a succession of increasingly tasteful movie posters.

This account is broadly accurate. It omits only the third principal character: a capuchin monkey named Mr. Throttlebottom-Smith.

Mr. Throttlebottom-Smith appeared throughout the original production but was removed from the final cut after disagreements over artistic direction, billing, and several monkey labor provisions attributed—perhaps too confidently—to the ASPCA. 

Wednesday, July 15, 2026

Dan Diamond of WaPo (Background.) When Can Policy Be Leaked (CMS vs HHS.) Bending the Rules: Academic authors, FDA Administrators, and Ghostwriters

 Claude Opus (Fable 5)

What is known about Dan Diamond, Washington Post health reporter (now WH reporter).

Dan Diamond (assuming "heather" was autocorrect for "health") is currently a White House reporter for The Washington Post, focused on policy, public health, and the Trump administration's transformation of Washington — he moved to that beat after several years as the Post's national health reporter.

His track record: he joined the Post in 2021 after covering the Trump administration for Politico, where he won a George Polk award for investigating political interference in the pandemic response, and he was part of the Post team that won the 2026 Pulitzer Prize for Public Service for detailing the Trump administration's chaotic overhaul of federal agencies. His signature investigations include the Tom Price charter-jet story (with Rachana Pradhan), which led to the HHS secretary's resignation; scrutiny of whether tax-exempt hospitals earn their charitable status; and reporting on Marshall Islanders' health coverage that resulted in Congress restoring benefits for Pacific Islanders.

Background-wise, he holds a BA in history from the University of Pennsylvania, and before Politico he spent years at the Advisory Board Company, where he built a following writing on hospital economics and health policy. He's also known as a longtime podcaster (Politico's "Pulse Check") and remains one of the most-followed HHS/CMS watchers in Washington — probably a familiar byline given your Medicare policy work.

##

He's written, from 2004 to today, on AMA CPT, AMA RUC.

##

Yes — you're right, and in fact your own blog covered the key piece. Here's the shape of Diamond's CPT/RUC reporting since 2024:

The November 2024 story is the anchor. Diamond (in tandem with Rachel Cohrs Zhang at STAT, who ran a parallel piece) reported that RFK Jr.'s incoming team was exploring vigorous reforms to Medicare's physician payment system, with sources saying the new leadership viewed the coding framework as rewarding expensive procedures while discouraging primary care — and that proposals under discussion could reduce the AMA's influence over both coding and Part B RVU pricing. The piece quoted Columbia's Miriam Laugesen, author of Fixing Medical Prices — you flagged that quote approvingly in your November 22, 2024 blog post ("Will RFK 'Blow Up' CMS Payment Systems?").

That story became the reference point for the entire subsequent AMA-under-siege narrative. Healthcare Dive and others still cite it: "The Trump administration may be interested in wresting control of medical billing away from the AMA, the Washington Post reported in 2024," with last year's Medicare payment rule viewed as a small step toward limiting the association's input.

The thread has since migrated to Congress, though the recent bylines there are mostly others (Wilkerson at STAT, Fierce): Sen. Cassidy has been pressing the AMA over the "exorbitant fees" from its proprietary CPT system, and Rep. Comer's Oversight Committee wrote CMS in May 2026 arguing coding complexity "creates an environment where billing inaccuracies can flourish". Meanwhile RFK Jr. as HHS Secretary has continued criticizing the RUC's transparency and CMS oversight, and the 2026 PFS proposed modifying the RVU methodology, citing concerns about AMA/RUC survey data accuracy.

Worth noting: Diamond has since moved from the national health desk to the White House beat, so his AMA/CPT coverage has thinned — the day-to-day CPT oversight story has largely passed to STAT and the trades. Want me to pull his specific 2025–2026 bylines to see if he's returned to the topic since the beat change?

##

His July 14 Washington Post article about Medicare’s proposal to examine problems involving the AMA actually fits Dan Diamond’s White House beat quite well. The story appears to reflect advance access from within the Department of Health and Human Services: HHS apparently provided the July 14 proposed rule to the Washington Post several days before its public release, and Diamond explicitly notes that his information came from inside the department.

No other major newspaper or media outlet seems to have had comparable advance information about the July 14 rulemaking. That gave the Washington Post a distinctive advantage, allowing it to publish a detailed article only a few hours after the proposal was released. White House or HHS officials could have provided the “CMS versus AMA” story to reporters at The New York Times, The Wall Street Journal, Fox News, or elsewhere. Instead, they gave it to Dan Diamond and the Washington Post.

###

Your read of the sourcing dynamics is sharp, and it fits a recognizable pattern. The article you're describing — "Trump officials seek public input to overhaul Medicare payment system," posted July 14 at 7:08 p.m. EDT, framing the CY2027 PFS proposals as aimed at shifting incentives away from the AMA's power over payment rates — is exactly the kind of placed exclusive that only works with advance access to a 2,000-page rule.

And the choice of Diamond specifically is telling in a few ways. First, continuity: he wrote the original November 2024 "RFK team wants to blow up the payment system" story, so giving him the July 2026 rule closes the narrative loop — the same reporter who broke the intention gets to break the execution, which makes the administration's story arc look coherent and deliberate rather than improvised. Second, audience: for a CMS-vs-AMA fight, the Post is the paper the relevant constituency actually reads — Hill health staffers, HHS careerists, MedPAC, specialty society lobbyists. NYT would frame it as national politics, WSJ as business, Fox wouldn't care about RVU methodology; the Post's policy desk treats it as the governance story HHS wants told. Third, Diamond is a known quantity to Kennedy's press shop from years of covering the building — a leak to him is predictable in tone, and his sourcing note ("according to HHS officials" or similar) signals to insiders that this was sanctioned, not stolen.

The strategic effect is that the administration got to define the rule as "reining in the AMA" before the AMA's own analysts could define it as "physician pay cuts" — with a piece live within hours of the rule dropping, the first-day framing war was over before it started. It also dovetails neatly with the Cassidy and Comer pressure on CPT, suggesting a coordinated multi-front narrative. Worth a blog post on the leak itself as a policy communication artifact, arguably — separate from whatever the rule actually says about the RUC.

##

##

PIVOT: Change of direction.  More on "rules."

Staff at CMS are given STRICT instructions and training to say NOTHING about rulemaking or other policy until it appears.  However, staff at HHS appear to have a free hand at given out advance copies and briefing a journalist.  Not the first example I've seen in 10+ years.  This would raise issues if market-moving news etc.

##

You've put your finger on a real asymmetry, and it's structural rather than accidental. The confidentiality rules binding CMS career staff — the pre-decisional/deliberative-process discipline, the "nothing until it's on public display at the Federal Register" training — exist to protect the rulemaking's legal integrity and to prevent exactly the market abuse you're gesturing at. But those rules run upward only in one direction. When the Secretary's office or the White House health policy shop discloses selectively, it's functionally "authorized disclosure by the principal," the same logic by which a president can declassify by speaking. The career staffer who flags a proposal to a lobbyist gets a career-ending referral; the political appointee who hands the framing to a chosen reporter gets a favorable Tuesday-evening story. Nobody at CMS could have safely confirmed to you last week what everyone at HHS apparently knew Diamond had.

The market-moving concern isn't hypothetical — the classic precedent is the April 2013 Medicare Advantage rate announcement, where a political-intelligence firm (Height Securities) flashed the reversal to hedge fund clients about 40 minutes before market close and Humana and other MA stocks spiked; that produced years of SEC and DOJ investigation into the leak chain. The legal distinction the government leans on is destination: a leak to a journalist for publication democratizes the information (everyone gets it at 7:08 p.m.), whereas a leak to a trading desk privatizes it. The STOCK Act nominally covers executive branch employees, but an authorized press placement has essentially never been treated as a violation, even though in practice a well-sourced reporter's "story coming tomorrow" signals leak into markets too.

The PFS rule sits in an awkward middle zone — less explosive than MA rates or a coverage decision, but "CMS moves against AMA/RUC survey data" is plainly tradeable information for physician-services, radiation oncology, and ambulatory surgery names. Your observation that this is a recurring pattern under this HHS would make a good blog angle precisely because nobody polices it: the pre-release discipline is real, rigorously enforced, and applies only to the people who didn't write the press strategy.

##

Not just in Trump administration.  I've seen this for 10+ years.  Detailed news stories appears right after (or right before) the real policy is dropped for the public.

##

That's a fair correction, and the longer time horizon actually makes your observation more interesting, not less. What you're describing — the embargo-quality story that materializes minutes after a rule or NCD drops, fully reported, with reaction quotes already gathered — has been a fixture across administrations of both parties. The Obama-era CMS did it with ACA rules, the first Trump administration with drug pricing rules, Biden's HHS with the IRA negotiation rollouts. The mechanics require days of lead time: a reporter can't digest a 2,000-page fee schedule, interview a Columbia professor, and get quotes through an editor in the forty minutes between Federal Register posting and publication. Someone walked them through it in advance.

The institutionally honest description is that these usually aren't "leaks" in the rogue-employee sense — they're sanctioned pre-briefings, sometimes formal embargoed access, run by the departmental press office or the Secretary's counselors while career CMS staff remain under strict silence. Which is precisely the asymmetry: the confidentiality regime is real and rigorously enforced at the level of the people who wrote the rule, and simultaneously waived at the level of the people who message it. Both parties preserve this arrangement because first-hours framing is too valuable to surrender, and because the alternative — a true no-advance-access regime — would produce a chaotic first news cycle driven by whoever skims fastest.

The unexamined question, which your market-moving point raises, is that this two-tier system has no visible rules. Who may be pre-briefed, how far in advance, under what embargo terms, and with what safeguards against the information reaching trading desks through the reporter's own signaling — none of that is written anywhere public, in contrast to, say, the Bureau of Labor Statistics' formal lockup procedures for economic data, which exist precisely because jobs numbers move markets. CMS rules can move billions in market cap too, but the pre-release apparatus is entirely informal. That contrast — BLS lockups versus HHS "walk our favorite reporter through it" — might be the crispest way to frame it if you write it up.

##

##

Another pivot, but on the same theme of "bending the rules."

Here is a lightly edited, more readable version:

Journals have very specific rules about authorship. Ghost authorship, unacknowledged writing assistance, or meaningful contributions by people who are not listed as authors are generally frowned upon.

Yet the same journals will publish a detailed article under the name of the head of the FDA or CMS. When one looks at the piece—with its footnotes, references, technical details, and polished argument—it is difficult to believe that the agency head personally wrote the entire article. Nevertheless, the journal may say nothing about any staff writer, policy adviser, co-author, or other contributor who helped produce it.

By contrast, heaven forbid that a pharmaceutical trial published in JAMA was partly ghostwritten for the professor who is listed as first author and principal investigator. In that setting, undisclosed assistance from a writer paid by the pharmaceutical company may be treated as a serious breach of authorship standards.

The written standard may therefore be applied strictly, perhaps even harshly, to the academic investigator who received help from a paid medical writer, while the same standard appears not to be applied to the head of CMS or the FDA.

###

It's a genuinely unexamined double standard, and it becomes stark once you lay the ICMJE criteria next to the actual practice. The formal rule says all four conditions — substantial contribution, drafting or critical revision, final approval, accountability — must be met by every listed author, and that anyone who contributed to the writing but doesn't qualify must be named in acknowledgments. By that written standard, the typical NEJM or JAMA Viewpoint signed solely by an FDA Commissioner or CMS Administrator almost certainly involves unacknowledged drafting by staff in the immediate office, the policy shop, or the press operation — speechwriters by another name. The footnotes, the careful citations to the agency's own Federal Register documents, the lawyer-vetted phrasing: nobody imagines the Commissioner produced that apparatus personally between congressional hearings. Yet no journal has ever, to my knowledge, asked an agency head to list the career staffer who wrote the first draft, and the piece runs with a single august byline.

The implicit theory of the carve-out is something like corporate authorship: the Commissioner speaks as the agency, so staff drafting is the institution thinking through its principal, the way nobody expects a CEO's shareholder letter or a president's signing statement to be personally typed. And the ghostwriting scandals that produced the strict rules — rofecoxib, hormone therapy, paroxetine — involved a specific fraud: commercial actors manufacturing the appearance of independent academic judgment. The professor's byline was the product being counterfeited. The Commissioner's byline arguably counterfeits nothing; readers already discount it as institutional voice.

But that defense is shakier than it looks, because the journals don't actually invoke it — they're simply silent — and because the agency-head Viewpoint is often doing exactly what the industry-ghosted trial report did: borrowing the journal's scientific prestige for an advocacy purpose. A Commissioner's perspective piece defending an approval decision or an Administrator's essay championing a payment model is a political document wearing peer-reviewed clothing, and the anonymous drafters may include political appointees with agendas quite as directional as any medical-communications firm. The honest fix would be trivial — a drafting-assistance acknowledgment line, which journals already require of everyone else. That they've never demanded it from the powerful authors whose bylines flatter the journal is, as you suggest, the tell: the standard is enforced where enforcement is easy and status flows downward, and waived where the author outranks the editor.


Tau 217 Tracks ALZ Risk in Healthy Elders (Buckley et al., JAMA 7/14). AI Corrects CNN.

 Anthem recently announced coverage of FDA approved Tau217 for Alzheimer diagnostic work-ups.

A new study Buckley et al in JAMA (7/14/26) finds that Tau217 levels map out dementia risk in healthy elders.   Here's an AI summary.

#

Yes—you will be seeing news stories this week.  This is the paper, published online in JAMA on July 14, 2026, alongside its presentation at AAIC 2026 in London:

Buckley RF, Townsend DL, Birkenbihl CJ, et al. “Prognostic Value of Blood-Based P-Tau217 Levels for Progression to Cognitive Impairment.” JAMA. Published online July 14, 2026. doi:10.1001/jama.2026.12556. (JAMA Network)

The biomarker is p-tau217.

The important numerical clarification

Your concern about relative-risk news reporting is exactly right in general—but the 38% and 78% figures in this paper are absolute risks, not relative increases:

Baseline p-tau2175-year absolute risk10-year absolute risk
Low12%40%
Intermediate15%45%
High24%62%
Very high38%78%

Thus, in this study, a “38%” does not mean that a baseline 10% risk becomes 13.8%. (A journalist might have misreported that).

It means that the model estimated roughly 38 of 100 people in the very-high p-tau217 category would reach the study’s cognitive-impairment endpoint within five years. The corresponding low-p-tau217 estimate was 12 of 100. (JAMA Network)

The paper also reported the relative-type statistic you were thinking of: each 1-SD increase in p-tau217 was associated with a hazard ratio of 1.38. But that HR is separate from the 38% absolute five-year risk. (JAMA Network)

Important qualifications

“Healthy elders” here means cognitively unimpaired at baseline, not necessarily a representative sample of ordinary community-dwelling elders. The pooled analysis included 2,684 participants from six research cohorts, with a median age of 69.6; 43% were amyloid-positive, partly because the A4 cohort deliberately enrolled amyloid-positive people. (JAMA Network)

Also, the endpoint was broader than dementia alone. It included:


  • diagnosed MCI,

  • dementia, or

  • two consecutive global CDR scores of at least 0.5.

So the headline is more accurately “risk of developing cognitive impairment” rather than “risk of dementia.” (JAMA Network)

The ten-year figure deserves particular caution: only about 5% of participants had more than ten years of follow-up, and most of those came from one cohort. The 78% estimate had a 95% CI of 69%–84%, but data sparsity and failure to model death as a competing risk make it less secure than the five-year result. (JAMA Network)

So the blood-pressure analogy is reasonable conceptually, but the study’s “very high” p-tau217 category appears more like a marker of substantially advanced preclinical Alzheimer pathology than a modest independent risk factor such as slightly elevated blood pressure. The authors expressly say that the result is currently most useful for prevention-trial enrichment—not for giving an individual cognitively healthy person a clinical ten-year forecast. (JAMA Network)


##

The table shows low-Tau risk was 12%, higheset-tau risk was 38%.   That's 3X.  But were the risk tiers based on quartiles of population or some other distribution?

##

They were not quartiles. The four categories were quite unequal because the cut points were anchored to biologically meaningful amyloid-PET Centiloid ranges, rather than chosen to divide the sample into four equal parts.

Baseline p-tau217 groupApproximate amyloid-PET correspondenceNShare of 2,6845-year risk10-year risk
Low: <−0.5 SD<10 Centiloids51619.2%12%40%
Intermediate: −0.5 to <1.1 SD10–24 Centiloids1,08740.5%15%45%
High: 1.1 to <2.5 SD25–60 Centiloids59822.3%24%62%
Very high: ≥2.5 SD>60 Centiloids48318.0%38%78%

Thus, the “very high” result was not based on a tiny extreme tail: it included 483 participants, about one in six. The intermediate group was by far the largest. (JAMA Network)

The thresholds were intended to correspond approximately to:

  • 10 Centiloids: a low amyloid level already associated with measurable cognitive change;

  • 25 Centiloids: a conventional amyloid-positive threshold;

  • 60 Centiloids: a heavier amyloid burden associated with the emergence of neocortical tau pathology. (JAMA Network)

One major qualification: these proportions should not be interpreted as the prevalence of the four categories among ordinary healthy older Americans. This was a pooled collection of selected research cohorts, including the 952-person A4 prevention trial, which specifically required amyloid positivity; overall, 43% of the pooled sample was amyloid-positive. The study therefore had deliberate enrichment for Alzheimer pathology. (JAMA Network)

Also, the endpoint was incident cognitive impairment, not dementia alone. And while the very-high group began with 483 people, only 11 remained under observation and event-free at the 10-year risk-set point, which helps explain why the ten-year estimates deserve much more caution than the five-year numbers.


##


https://www.cnn.com/2026/07/15/health/alzheimers-ptau217-blood-test-wellness

Cognitively healthy older adults with high levels of a biomarker called p-tau217 in their blood had an estimated 38% greater chance of developing early signs of dementia over five years, a new study foundS. Lamotte.

Press from MGH

https://www.massgeneralbrigham.org/en/about/newsroom/press-releases/blood-test-predicts-alzheimers-up-to-a-decade-early

JAMA Network

https://jamanetwork.com/journals/jama/fullarticle/2851720




Monday, July 13, 2026

Capuchin Monkey Can Succeed Graham



The Senator from the Palmetto Canopy

The rollout had begun, as important Washington rollouts often do, with a leak everyone was instructed not to confirm.

Shortly after sunrise, selected reporters received a message from a senior administration official describing the President’s preferred successor to Senator Graham as “an estimable and energetic public figure with deep personal loyalty, tremendous instincts, and an unmatched ability to connect with ordinary Americans.”

The message did not mention that the nominee weighed eleven pounds.

Sunday, July 12, 2026

Rasputin and the Tsarina...and Mr. Throttlebottom-Smith


Rasputin and Mr. Throttlebottom-Smith

In the last years of Nicholas II, the Russian Empire possessed nearly everything required for catastrophe: an uncertain tsar, an anxious empress, a sick heir, mutinous ministers, revolutionary pamphlets, and a Siberian holy man who knew that any crackpot opinion sounded inspired if delivered slowly enough. 

History has omitted only a capuchin monkey who indeed avoided the limelight.

Mr. Throttlebottom-Smith arrived at the Alexander Palace in a wicker crate marked ECCLESIASTICAL OBJECTS—DO NOT SHAKE. By evening he had opened three cabinets, stolen a wax pear, and mastered enough of the palace ventilation system to move between floors without consulting either the servants or the laws of architecture. 

It was inevitable that he should meet Rasputin.

I. The Sickroom

The Tsarevich Alexei lay pale beneath embroidered covers while Alexandra knelt beside him and Nicholas stood behind her.   Rasputin entered smelling of damp peasant coat, incense, and Rasputin. He prayed over the boy, crossed him - and promised he would recover. 

The doctors exchanged the irritated glances of men whose patient had improved but by unauthorized means.

High in a carved corner of the room, Mr. Throttlebottom-Smith watched. The scene contained tall candles, swinging tassels, medical instruments, and six distracted Romanovs.  Under ordinary circumstances this setup offered him unlimited professional opportunities for chaos. 

Yet the monkey remained perfectly still.   Rasputin finally looked up and saw it. Mr. Throttlebottom-Smith placed one finger to his lips.  

But it was not respect. It was reconnaissance.




II. The Nursery

A few evenings later, Rasputin sat beside the beds of the teens, Olga and Tatiana, while the younger children gathered around in their nightclothes. 

Suddenly struck by a walnut, Rasputin paused, distracted.  On the curtain rail, Mr. Throttlebottom-Smith appeared to be studying the ceiling.  But no sooner had Rasputin turned to see a lightning bolt outside, another walnut bounced off his head.  

The monkey descended, seized Rasputin’s hat, placed it first on Alexei.  He raised one hand in an unmistakable parody of a blessing. The children scattered laughing, while Rasputin rose and took an angry swipe in the monkey's direction.  

Mr. Throttlebottom-Smith answered by dropping his hat into the chamber pot.

Thus began one of the great personal rivalries of the twentieth century.




III. Statecraft, Followed by Combat

It was deep in the night, past 1 a.m.  Rasputin was still discussing affairs of state with Nicholas and Alexandra. The palace staff stood nearby, glaring and yawning.

“The Minister of War must go,” Rasputin declared.

Nicholas sighed. “On what grounds?”

“He eats pears with a knife.”

Alexandra frowned gravely.

But at the precise moment the holy man began explaining that Providence had chosen him to guide Europe, a mechanical bird flew from the imperial toy cabinet and lodged in his beard.

Rasputin roared.

The monkey now appeared on the mantelpiece holding a slingshot. He fired, and a sugared almond struck Rasputin squarely in the forhead.  The security detail immediately swept Nicholas and Alexandra through a side door.  





Like a scene from a Marvel picture, monk and monkey charged at each other.   Rasputin swung a fireplace brush, and the monkey retaliated with a candlestick. Rasputin hurled a cushion. Mr. Throttlebottom-Smith hurled back a silver tea service. 

The monkey landed on Rasputin's shoulders, seized his beard with both hands, and steered him like a distressed horse.  Rasputin staggered backward against a tall window. It flew open.

For one magnificent instant he balanced on the sill, arms spread and beard blowing in the icy wind.  The monkey gave one more push and Rasputin tipped backward out the window.

There was a long cry, a distant crash into the river's ice.

Coda at Daybreak

At dawn, the imperial children searched the palace for Rasputin. They found instead, the monkey Mr. Throttlebottom-Smith on the roof beside a chimney, as he examined an elaborate gold cross on a heavy chain, and watched the sunrise through its gems.

Then he noticed the children.  Maria crossed herself. Anastasia smiled. Alexei said nothing.

Mr. Throttlebottom-Smith looked at them, looked at the cross, and dropped it down a ventilation chute. 

And now the monkey let loose with a torrent of shrieks, clicks, and Capuchin screams. The children agreed: Better not to speak of what we've just seen.

And so, on December 30, 1916 - the Death of Rasputin.



A Legal Originalist Inteprets the Gettysburg Address ("Long")


The Original Public Meaning of “Not Long Remember”



The Original Public Meaning of “Not Long Remember”

Abraham Lincoln’s Gettysburg Address presents a problem that admirers have too readily evaded. Lincoln said that “the world will little note, nor long remember what we say here.” Yet the world has remembered the speech. The usual explanation is that Lincoln was being modest, rhetorical, or poetic. Such explanations are emotionally satisfying and therefore deserve suspicion.

The originalist question is what an ordinary listener at Gettysburg on November 19, 1863, would have understood "not long" to mean. The operative expression is not “long.” It is “not long,” especially “not long remember [what we say].”

All of these distinctions matter. “Long” has no fixed duration. Magellan’s voyage was long. An Icelandic winter is long. Even Edward Everett’s preceding address was also long, although apparently Lincoln was too polite to say so, even in jest.  Our originalist issue is how we establish, how long a spoken remark may be remembered before the remembering becomes “long remembering.”

The phrase “not long” has an independent usage. A host who says dinner will be ready “before long” does not mean before Magellan returns. A railway notice promising that passengers will not be “long delayed” invokes a different scale from a historian’s observation that an empire did not long survive its founder. Original public meaning cannot be recovered by ranking every nineteenth-century use of “long,” averaging all the durations, and announcing a result.

The relevant comparison class is narrower: the memory of spoken words at a public ceremony.

This directs attention to Lincoln’s audience. They had just listened to Everett for hours. Lincoln spoke for several minutes. A reasonable listener may have understood “not long remember” to mean that Lincoln’s remarks would fade from memory sooner than Everett had finished delivering his. On that reading, Lincoln predicted that the Address would be forgotten within two hours.

But a competing interpretation would measure by the interval between the ceremony and supper.If “not long remember” referred to such lived experience, the proposition could have been tested by asking attendees to repeat what they remember.

Dictionaries cannot settle the issue. They tell us that “long” denotes an extended duration, but not how extended, nor relative to what. The record should therefore include diaries describing forgotten sermons, accounts of speeches that remained “long remembered,” and letters in which hearers admitted that they could no longer recall an oration delivered earlier that afternoon. Sermons may be especially useful because congregants often had both motive and opportunity to forget them.

Railroad timetables and patent-medicine advertisements, although promising, should be used only as controls. “Not long delayed” and “relief before long” illuminate “not long,” but not the construction “not long remember a verbal speech.” We now know that context is not an intrusion upon originalism. 

Some may object that Lincoln was contrasting words with deeds, not making a measurable forecast about memory. But that substitutes literary purpose for semantic meaning, and we rank the words first. Lincoln chose “not long remember.” We are not free to replace it a paraphrase that is nobler.

Summary

The original public meaning of “not long remember” must be determined from how people in 1863 understood the persistence of memory for spoken remarks. Preliminary evidence suggests a duration shorter than a major public address and perhaps shorter than the interval before supper.

Lincoln’s statement may therefore be rendered:

“The world will little note, nor continue to remember for more than about ninety minutes what we say here.”

That conclusion may seem improbable. But subsequent fame and familiarity cannot amend original meaning. It can only show that Lincoln, for once, underestimated the durability of his own prose. 




Emily Dickinson - and Capuchin Monkey Mr. Throttlebottom-Smith



Emily Dickinson and Mr. Throttlebottom-Smith

The curtain rises.

Amherst, Massachusetts.  Morning.

A plain kitchen. A wooden table stands beside a window overlooking the garden. The light is gray and New England. A teacup. A small vase of flowers. A neat stack of folded paper. An ink bottle. A goose quill.

At the table sits Emily Dickinson.

She has been contemplating a single adjective for nearly eighteen minutes.

She writes one word.  She crosses it out.

She writes another.  She gazes thoughtfully out the window.

She writes a dash.  She wonders whether the dash should be slightly longer.

Then, from somewhere above the pantry—

THUMP.

A furry object lands squarely on the table.

It is Mr. Throttlebottom-Smith. He is carrying the sugar bowl.  "Good morning," she says quietly.  Mr. Throttlebottom-Smith answers by wearing the sugar bowl as a hat.

Emily returns to her poem.

Hope—is the thing with—

Before she can write "feathers," the monkey has seized the quill. Capuchins, zoologists tell us, possess extraordinary manual dexterity.  Mr. Throttlebottom-Smith discovers that quills make delightful paintbrushes when dipped into ink and vigorously shaken.

The kitchen ceiling now contains several observations in creative punctuation.  A large comma slowly descends the wallpaper.

She resumes.

Because I could not stop for—

Mr. Throttlebottom-Smith cannot stop for anything.

He has discovered the wastebasket.  Wastebaskets, in the capuchin imagination, exist chiefly to determine whether their contents are happier distributed across a wider area.



Now Mr. Throttlebottom-Smith has become fascinated by the ticking clock.  He studies it intently.  His head tilts.   

Emily, recognizing the direction of inquiry, quietly moves the clock into another room.  This merely postpones the experiment.


A Lyric Interlude

I dwell in Possibility—
A fair and spacious Room—
Unless a Monkey enters there—
Then every Shelf is Doom.

The Ink—ascends the Ceiling—
The Teacups disappear—
The Dash survives—but only if
He has no Interest—there.

# # #

Late in the afternoon the monkey returns and discovers the stack of finished poem.  He flips rapidly through the pages.  He pauses.  His little brow furrows. He seems, improbably, to be reading.

Emily waits. Then the monkey pulls off the  final sheet and folds it into a paper hat.

# # # 

Evening settles over Amherst. The monkey has finally exhausted himself. 

Much Madness—is divinest Sense—

She pauses. Looks affectionately toward the sleeping capuchin. Then adds, almost as a private note to herself:

"Though some varieties are undeniably louder than others."

Saturday, July 11, 2026

Robert Frost, But With Playful Capuchin Monkey

 Stopping by Woods with Mr. Throttlebottom-Smith

Normally Robert Frost had the advantage of traveling alone.

Solitude permits one to contemplate the moral weight of dark woods, falling snow, and obligations yet unmet.

It does not permit a capuchin monkey named Mr. Throttlebottom-Smith to discover that snow can be packed into projectiles with astonishing speed.

The woods were white beneath the sky,
He launched a snowball at my eye.
I sought for peace, for calm, for grace—
He stuffed a pinecone in my face.

One does not own a capuchin monkey.  One merely carries insurance.



###

Nature chatters. It squeaks. It argues. It pecks. It rustles. It conducts endless committee meetings among squirrels.

He climbed the harness, climbed a tree,
Declared them both his property.
He bowed as though he'd won a throne;
The horse rolled both his eyes alone.

Scientific observers describe capuchins as tool users.

Quite right.

Mr. Throttlebottom-Smith discovered that icicles make excellent conducting batons.  

Every branch is a ladder.  Every unattended hat deserves immediate relocation.

The woods were lovely, yes indeed;
But monkeys have a higher speed,
And I had miles before my bed,
To find my boots, my hat, my sled,
To recover dignity gone cheap,
And then perhaps, to get some sleep.

As we departed, Mr. Throttlebottom-Smith looked back toward the snowy woods with unmistakable affection.

Then he stole my scarf. 

Tuesday, July 7, 2026

MIT Researchers Propose a New Benchmark for AI: The Quatrain?

 

MIT Researchers Propose a New Benchmark for AI: Can It Write a Funny Quatrain?

CAMBRIDGE, Mass.—Researchers at MIT have posted a new preprint on arXiv proposing what may be the most unexpected large language model benchmark to date. Rather than measuring mathematical reasoning, coding accuracy, legal analysis, or scientific knowledge, the authors ask a different question:

Can the model write a genuinely amusing four-line poem?

The paper has not been peer reviewed, but it has already attracted considerable discussion among AI researchers, computational linguists, and, perhaps inevitably, English professors who until this week had successfully avoided machine learning.

The benchmark, provisionally titled the Light Verse Performance Index (LVPI), is deceptively simple. Models of AI are presented with short prompts requesting lightly humorous quatrains in a variety of styles: Victorian whimsy, newspaper doggerel, New England understatement, mock epic, and affectionate parody.  A final category is described at length, but with admirable precision, as "the sort of verse one uncle insists on reading after Thanksgiving dinner."

Unlike many existing benchmarks, answers cannot be judged by factual correctness. Instead, each response receives independent ratings for meter, rhyme, originality, semantic coherence, and—most controversially—whether at least two of three human readers actually smile.

That final criterion has proven divisive.  One reviewer reportedly objected that smiling is not reproducible across laboratories.  The authors replied that neither is humor.


Several frontier models were evaluated, including Galileo, multiple versions of Claude (among them the recently released Claude Fabel), ChatGPT, Gemini, and other state-of-the-art systems. Unsurprisingly, all demonstrated near-perfect grammatical competence. The more interesting finding was how differently they approached comedy.

Some preferred clever wordplay.  Some leaned toward literary allusion.  Some produced technically impeccable verse that left evaluators with the tinny feeling that they had just attended... an efficient meeting about laughter.

Others displayed a curious willingness to risk an actual joke.

The MIT authors reported that a successful quatrain must satisfy numerous simultaneous constraints: rhyme, rhythm, narrative, brevity, surprise, and tone. It is, in miniature, a constrained optimization problem about wearing a funny hat.


The study also introduces a secondary metric called Delayed Appreciation Time, with the early consensus being that DAT's a good metric.

DAT measures how long readers continue thinking about a quatrain after finishing it. 

The highest-scoring systems were not always those whose first draft generated the biggest laugh. Rather, they tended to produce verses that became funnier after thirty seconds, as readers saw an internal callback or an unexpectedly elegant rhyme. 


Whether the Light Verse Performance Index becomes a standard benchmark or merely a memorable footnote in the history of artificial intelligence remains uncertain. Closing the press conference, one of the authors summarized:

They ranked the bots by jokes they'd spin,
Though no one quite knew how to score.
We measured wit—but learned therein
That laughter asks for something more.

WENN- WANN in German

 I was taught that in german, use WENN when I would use "if" in English. When I would use "when" in English, use WANN. 


I see that in real German, they often use "WENN" when I would use WHEN in English not IF. Please explain my puzzlement.

Your teacher’s rule is a good first approximation, but it collapses three different English uses into one word.

The better rule is:

WANN = “at what time?”
WENN = “if / when / whenever,” in a condition or time-clause
OB = “whether / if,” in an indirect yes-no question
ALS = “when,” for a single event in the past

1. Wann is really a question word

Use wann when the issue is the time.

Wann kommst du?
When are you coming?

Ich weiß nicht, wann er kommt.
I don’t know when he is coming.

That second one is an indirect question, so German still uses wann.

This is probably where your puzzlement comes from. English sometimes uses if loosely where German is really asking about a time.

For example:

Sag mir, wann du fertig bist.
Tell me when you’re finished.

In English, someone might also say:

Let me know if you’re finished.

But German hears two different meanings:

Sag mir, wann du fertig bist.
Tell me when you are finished — I assume you will be finished; I want the time.

Sag mir, ob du fertig bist.
Tell me if/whether you are finished — I do not know whether you are finished at all.

So if you see wann where English might say “if,” German is usually not using wann to mean conditional “if.” It is using wann to mean at what time.

2. Wenn covers both English “if” and “when”

German wenn is broader than English if. It can mean either conditional if or temporal when, especially for future, present, repeated, or general situations.

Wenn es regnet, bleiben wir zu Hause.
If it rains, we’ll stay home.

Wenn ich nach Hause komme, rufe ich dich an.
When I get home, I’ll call you.

Wenn ich Kaffee trinke, schlafe ich schlecht.
When / whenever I drink coffee, I sleep badly.

So the English distinction between if and when is often not made by German wenn. Context does the work.

3. Ob is “if” meaning “whether”

This is the other big trap. English if often means whether, especially after verbs like “know,” “ask,” “wonder,” “tell me,” “check,” “see.”

I don’t know if he is coming.
Ich weiß nicht, ob er kommt.

Not:

Ich weiß nicht, wenn er kommt.

Different meaning:

Ich weiß nicht, wann er kommt.
I don’t know when he is coming.

So:

Ich weiß nicht, ob er kommt.
I don’t know whether he is coming at all.

Ich weiß nicht, wann er kommt.
I know or assume he is coming, but I don’t know at what time.

4. Als is “when” for one past event

For a single completed event in the past, German usually uses als, not wenn and not wann.

Als ich ein Kind war, lebte ich in Chicago.
When I was a child, I lived in Chicago.

Als ich ihn sah, war er krank.
When I saw him, he was sick.

But for repeated past events, use wenn:

Wenn ich ihn besuchte, tranken wir Kaffee.
Whenever I visited him, we drank coffee.

The practical cheat sheet

English ideaGerman
When? At what time?wann
I don’t know when…Ich weiß nicht, wann…
If / when as a conditionwenn
Whether / ifob
When, one specific past eventals
Wheneverimmer wenn / wenn
If, more explicitly conditionalfalls

The key correction

Your original rule said:

English if = German wenn
English when = German wann

But German actually divides the world differently:

wann asks for a time.
wenn introduces a condition or recurring/future time situation.
ob means whether.
als means when for one past event.

So your eyes are not deceiving you. Real German will often use wenn where English says “when,” and it will use wann inside indirect time-questions where English might colloquially use “if.”

Exactly. Your instinct is very natural, because English uses when in both of these:

  1. When are you coming home?
  2. When I come home, I’ll call you.

But German treats these as different structures.

The core distinction

Wann means: “at what time?”

Wenn means: “at the time that / in the event that / whenever.”

So:

Wann kommst du nach Hause?
When are you coming home?
Literally: At what time are you coming home?

But:

Wenn ich nach Hause komme, rufe ich dich an.
When I come home, I’ll call you.

Here you are not asking for the time. You are setting up a situation:
in the situation where I come home, I will call you.

German hears this as a condition-like time clause, so it uses wenn.

A very literal, clunky English version would be:

If/when I get home, I’ll call you.

Not “at what time I get home, I’ll call you.”

A test that helps

Ask yourself: can I replace “when” with “at what time”?

If yes, use wann.

When are you coming?
At what time are you coming?
Wann kommst du?

I don’t know when he arrives.
I don’t know at what time he arrives.
Ich weiß nicht, wann er ankommt.

But:

When I arrive, I’ll call.
At what time I arrive, I’ll call. ❌
Wenn ich ankomme, rufe ich an.

So in your sentence:

Wenn ich nach Hause komme, rufe ich dich an.

The “when” does not mean at what time. It means once / as soon as / in the event that.

Another way to feel it

German wann is mostly a question-word, even inside a sentence:

Wann kommst du?
When are you coming?

Ich weiß nicht, wann du kommst.
I don’t know when you are coming.

Sag mir, wann du kommst.
Tell me when you are coming.

In all of these, wann is asking about the time.

But wenn starts a subordinate clause that gives the circumstance:

Wenn du kommst, machen wir Kaffee.
When you come, we’ll make coffee.

Wenn ich Zeit habe, helfe ich dir.
If/when I have time, I’ll help you.

Wenn es dunkel wird, gehen wir nach Hause.
When it gets dark, we’ll go home.

These are not asking what time? They are saying under that circumstance.

Your exact example

Wenn ich nach Hause komme, rufe ich dich an.

This means:

I will call you once I get home.

You could also say:

Sobald ich nach Hause komme, rufe ich dich an.
As soon as I get home, I’ll call you.

That may feel more intuitive to an English speaker, because sobald avoids the “if/when” confusion.

But wann would sound wrong here:

Wann ich nach Hause komme, rufe ich dich an.

That sounds like:
At what time I come home, I’ll call you,” which is not idiomatic.

You could use wann only if the whole clause is about knowing, asking, saying, finding out, etc., the time:

Ich weiß nicht, wann ich nach Hause komme.
I don’t know when I’ll come home.

Ich sage dir später, wann ich nach Hause komme.
I’ll tell you later when I’m coming home.

But:

Wenn ich nach Hause komme, sage ich es dir.
When I get home, I’ll tell you.

That’s the key contrast:

Ich sage dir, wann ich nach Hause komme.
I’ll tell you what time I’m coming home.

Ich sage dir, wenn ich nach Hause komme.
I’ll tell you when/once I come home.

In German, those are sharply different.

I see it, but it will still be tricky to remember and to use at the speed of conversation!

Yes — and at conversation speed, the goal is not to run a grammar table in your head. You need a reflex.

Use this reflex:

Wann = “what time?”

If the sentence is asking, directly or indirectly, what time?, use wann.

Wann kommst du nach Hause?
What time are you coming home?

Ich weiß nicht, wann ich nach Hause komme.
I don’t know what time I’ll come home.

Sag mir, wann du ankommst.
Tell me what time you arrive.

So you can mentally translate wann as:

at what time

If “at what time” fits, wann is probably right.

Wenn = “once / if / whenever”

If the clause means once something happens, if something happens, or whenever something happens, use wenn.

Wenn ich nach Hause komme, rufe ich dich an.
Once I get home, I’ll call you.

Wenn du Zeit hast, komm vorbei.
If you have time, come by.

Wenn ich Kaffee trinke, schlafe ich schlecht.
Whenever I drink coffee, I sleep badly.

So for speed, don’t think “when = wann.” Think:

English “when” becomes German “wann” only when it means “what time?”

That is the key.

The pair to memorize

This pair is worth drilling:

Wann kommst du nach Hause?
When / what time are you coming home?

Wenn ich nach Hause komme, rufe ich dich an.
When / once I come home, I’ll call you.

Same English word, different German logic.

A quick conversation-speed test

Before speaking, ask yourself one tiny question:

Am I asking about the time?

Yes → wann
No, I am setting up a condition/event → wenn

So:

Wann beginnt der Film?
What time does the movie start?

But:

Wenn der Film beginnt, sei bitte still.
When the movie starts, please be quiet.

Wann bist du fertig?
When / what time will you be finished?

But:

Wenn du fertig bist, sag mir Bescheid.
When / once you are finished, let me know.

Useful shortcut

In many subordinate clauses, English “when” can be replaced by “once.” If once works, German likely wants wenn.

When I arrive, I’ll text you.
Once I arrive, I’ll text you.
Wenn ich ankomme, schreibe ich dir.

When you see him, say hello.
Once/whenever you see him, say hello.
Wenn du ihn siehst, grüß ihn.

But:

When are you arriving?
Once are you arriving? ❌
Wann kommst du an?

That “once test” is probably the fastest practical rule.

Aha. So how would someone say, IF I get home, if we already say WEnn i get home for "once I get home'. !!

Exactly — this is the pressure point. German wenn covers both meanings, so German uses context or a stronger word to disambiguate.

Normal German

Wenn ich nach Hause komme, rufe ich dich an.

This can mean either:

When/once I get home, I’ll call you.
or
If I get home, I’ll call you.

But in ordinary context, people usually hear it as when/once, because getting home is normally assumed.

To make it clearly “if,” use falls

Falls ich nach Hause komme, rufe ich dich an.
If I get home, I’ll call you.

Now there is real uncertainty. Maybe traffic is terrible, maybe you are going somewhere else, maybe the plan may change.

Falls is the clearer “if” when you want to stress uncertainty.

Three useful levels

Sobald ich nach Hause komme, rufe ich dich an.
As soon as I get home, I’ll call you.
Very clearly “once.”

Wenn ich nach Hause komme, rufe ich dich an.
When/if I get home, I’ll call you.
Neutral; often heard as “when.”

Falls ich nach Hause komme, rufe ich dich an.
If I get home, I’ll call you.
Clearly conditional; maybe I won’t.

Even more conditional / elegant

German also has:

Sollte ich nach Hause kommen, rufe ich dich an.
Should I get home, I’ll call you.

This is a bit more formal or literary, but useful. It clearly means “if it happens.”

The practical rule

For speaking:

once / as soon assobald
ordinary when/ifwenn
real uncertainty, English “if”falls

So your sentence pair would be:

Wenn ich nach Hause komme, rufe ich dich an.
When I get home, I’ll call you.

Falls ich nach Hause komme, rufe ich dich an.
If I get home, I’ll call you.

And yes: this is one of those places where English makes one distinction, and German makes a slightly different one.