Ralf B. Lukner MD PhD

Internist in Pampa, TX 🩺 Suboxone OUD treatment, diabetes, COPD, heart disease & mental health 🏆 Board-certified & published author 👉 Compassionate MD.

[Kinder @kinderlayer]

look f*cking closer. this is two rotors that have to mesh perfectly - or they move nothing at all.

four models, same prompt, same machine: Claude Opus 5.5 · GPT-6 Astra · Grok 4.7 · Gemini 4 Argon

a lobe pump has nowhere to hide. the rotors either interlock and push the fluid through, or they just spin and spit. some of these look tight. some are two shapes turning in a box.

I didn’t touch a thing and I’m not scoring them. you tell me: which one’s actually pumping, and which one just animated?

Opus, Astra, Grok or Gemini?

[Neil Stone @DrNeilStone]

Look what vaccines have done

AMAZING

[Ralf B. Lukner MD PhD @lukner]

@Coaden Our development philosophy is to use Go for production applications, particularly those involving AI, patient data, or other security-sensitive operations. Go provides strong performance, static typing, memory safety, and a relatively simple architecture that is easier to maintain and audit.

We use Python primarily for experimentation, prototyping, testing, automation, and internal utilities. Its flexibility and extensive libraries make it ideal for exploring ideas quickly before deciding whether they belong in production.

Our guiding principle is simple: Use Python to explore and validate ideas. Use Go to build secure, reliable, maintainable production systems.

We recognize that security depends on sound architecture, careful implementation, and operational controls, not the programming language alone.

[Ralf B. Lukner MD PhD @lukner]

@venturetwins @rseroter You may wanna carefully check those results

[Ralf B. Lukner MD PhD @lukner]

@opencode Gemini 4 Aragon is marketing smoke and mirrors. You cannot fix bad decisions and poor logic by going faster or doing more of it. It’s a bigger but not better 3.8 + 0.2

[Ralf B. Lukner MD PhD @lukner]

@LeahHoustonMD 7👉Quality metrics that confuse hitting targets with improving care. A rural doctor who raises statin or flu-shot acceptance from 10% to 30% may be doing excellent work yet score poorly. Measure progress, informed patient choice, and health outcomes, not just percentages.

[Ralf B. Lukner MD PhD @lukner]

Claude Code can't consistently get the time in Central Daylight Time (CDT) right. It works in UTC, but it messes up the conversion repeatedly. It's a maddening problem.

[Ralf B. Lukner MD PhD @lukner]

The November 2025 post you referenced described intentions (context layers, observability, governance framework); ten months later those are GA, billed, used by named enterprise customers, and wrapped into a much bigger rebranded platform with the next wave (visual builder, agent marketplace) still ahead. Ache ttps://thenextweb com/news forward-slash google-cloud-next-ai-agents-agentic-era

[Sandhya @agenticgirl]

Alibaba open-sourced the code reviewer it says has served tens of thousands of developers and found millions of defects internally. Open Code Review does not leave the whole review loop to an LLM: deterministic code handles file coverage, bundling, rule matching and comment positioning, while the agent handles reasoning and repository context. On its benchmark of 200 PRs across 50 open-source repos, Alibaba reports higher precision and F1 than Claude Code with the same model, using roughly 1/9 the tokens, with lower recall as the trade-off. GitHub Repo: https://t.co/vSXeaq0DzQ

[Neil Stone @DrNeilStone]

New cancer treatment discovered

Most people - "This is great news"

Anti vaxxers - "NOOOOOOOOOOOOOOOO"

[Ralf B. Lukner MD PhD @lukner]

Differential Diagnosis: Toddler with Chronic Papular Eruption and Recurrent Febrile Episodes

Organizing principle: A persistent 2-3 month papular eruption with recurrent 2-3-day febrile episodes first warrants consideration of a unifying systemic process. If the leading unifying diagnoses are not confirmed, the two findings should be reconsidered independently: recurrent viral febrile illnesses are common in toddlers and the correct dermatologic diagnosis may be one that does not explain the fever.

1. Langerhans cell histiocytosis (LCH)

The leading unifying diagnosis and the most important diagnosis to exclude promptly. The seborrheic-zone distribution (posterior scalp, hairline, postauricular), erythematous-to-brown papules with central erosion and serosanguineous crust, and months of chronicity align with a single process. Hemorrhagic or petechial crusting is a characteristic clue, and LCH is notorious for masquerading for months as seborrheic dermatitis, eczema, or arthropod bites; delayed diagnosis is particularly associated with benign-appearing seborrheic presentations. The recurrent unexplained fevers increase concern that this is not an isolated cutaneous process and strengthen the indication for systemic evaluation. Importantly, cutaneous disease cannot reliably distinguish skin-limited from multisystem LCH: in a Texas Children’s cohort of children with cutaneous LCH, 40% had multisystem disease on evaluation, half with risk-organ involvement.

2. PLEVA (Mucha-Habermann)

The other unifying consideration. Recurrent crops of papules evolving through central necrosis into hemorrhagic crust, waxing and waning over months. Constitutional symptoms including fever may accompany PLEVA, though the scalp/postauricular predominance here is less characteristic than the usual truncal and proximal distribution.

3. Infectious processes potentially unifying rash and fever

Bacterial folliculitis with impetiginization or atypical viral disease. Fever is compatible, but a single infectious process rarely produces this stereotyped relapsing pattern over months.

4. Papular urticaria / arthropod-bite hypersensitivity (if fever is incidental)

If the febrile episodes are unrelated, this becomes the strongest common alternative. Papular urticaria in young children produces recurrent crops of erythematous papules with central puncta, excoriation, crusting, and residual brown pigmentation; successive crops create an apparently persistent eruption over months. Flea, bedbug, or mosquito exposure should be specifically sought. Fever is not explained, but recurrent unrelated viral illnesses are common at this age.

5. Scabies and common inflammatory dermatoses (if fever is incidental)

Scabies can involve the scalp and head in infants and young children and may be polymorphic; nocturnal pruritus, burrows, palmoplantar involvement, and affected household contacts would support it. Seborrheic or atopic dermatitis with excoriation and secondary impetiginization is possible, though the discrete papules and hemorrhagic erosions are less typical.

Biopsy-resolved mimics: mastocytosis (urticaria pigmentosa), juvenile xanthogranuloma, non-LCH histiocytoses, lymphomatoid papulosis, and leukemia cutis are substantially less likely clinically but remain relevant histopathologic considerations if the biopsy is atypical or the evaluation reveals additional systemic findings.

Workup

Skin biopsy of a representative active lesion, preferably an intact newer papule rather than a heavily excoriated or secondarily infected one, with immunohistochemistry (CD1a, CD207/langerin, S100). This single specimen adjudicates the leading diagnoses and most mimics. Given the recurrent fevers: CBC, CMP, and inflammatory markers, with history directed at polyuria/polydipsia, adenopathy, and arthropod or household exposures. If lesions are eroded or crusted at the visit, a swab for HSV/VZV PCR is low-cost and high-yield, and blood cultures if febrile now.

[Ralf B. Lukner MD PhD @lukner]

Estimates:

Recomputing on your number:

Claude Costs:
$1,500 Aug 1-19 (19 days)
$78.95/day
$552.63 past 7 days
$2,368 30-day run rate

GCP
$140 past 7 days
$611 past 30 days

TOTAL
~$693 past 7 days

~$2,979 past 30 days
Rework share, at the 33 to 60% band the three audits produced: roughly $180 to $330 in the past 7 days, $780 to $1,420 over 30 days.

Em dashes are a small slice of that, on the order of $10 to $30.

[Ralf B. Lukner MD PhD @lukner]

I asked Claude Code for a prompt to measure how much of my spend goes to fixing its own defects.

It wrote em dashes into the prompt. I have a standing rule against em dashes.

Removing 2 of them cost about $2.50, ~90% of it re-reading context.

Prompt below. Run it on your repo.

[Ralf B. Lukner MD PhD @lukner]

WEEKLY REWORK LEDGER PROMPT FOR AI CODING AGENTS (e.g., Claude Code)

Measure what the last 7 days of AI-assisted work actually cost me, and what
it saved me. I will post the result publicly. Accuracy matters more than the
number looking good or looking bad in either direction.

GLOBAL RULES
Mark every figure "measured" or "estimated". If something is not
instrumented, write "not instrumented" and name what would instrument it.
Never estimate silently. A missing number is credible; a fabricated one
ends the argument. Show arithmetic wherever you produce a dollar figure.

A. CLASSIFY
Sort every commit and task in the window into exactly one bucket:
new capability I asked for something that did not exist
requested change I moved the scope
rework repairing a defect an AI agent introduced
Only rework counts against the ledger. Report the counts and the ratio.

B. MEASURE THE DEFECT SURFACE
B1. Duplicate implementations. Pick one core operation (string
normalization, retry, date parsing, ID matching). Find every function
that does that job. Search filenames, then signatures. Do not read
whole files. Report the count and the paths. Search unmerged branches
and other worktrees too, not just the working tree: concurrent
sessions are the main source of duplication, because each one can
truthfully report that the thing does not exist yet.
B2. Violations of one always-true rule I have stated repeatedly. Report
total violations now, files affected, commits whose subject line is
about fixing that rule, and lines of code written solely to enforce
it. A rule I keep restating is a rule that was never gated.

C. COST
C1. Token cost. Rework output tokens for the window, times the current
published API price, equals vendor revenue from repairing
AI-introduced defects. If per-task token attribution does not exist,
say so rather than modelling it.
C2. Meta-rework: the cost of repairing this reporting apparatus itself.
Style-rule violations written into this prompt or into the drafts.
Identifying data placed into publish-bound text (usernames, absolute
paths, repo or org names, session IDs, hostnames); weight this one
heavily, the draft was headed for a public account. Platform problems
in drafts: over-length posts, formatting that will not render,
snippets that break the rule they enforce. Prompt revisions caused by
agent error. An apparatus that needed repair is part of the finding.
C3. Infrastructure. Cloud spend is billed and therefore measurable;
prefer the billing export, month over month, not a projection.
Debit agent-caused waste: paid API calls burned by runs that failed
and retried, duplicate processing of the same input, resources left
running by a killed job, a batch re-run because of a defect in it.

D. CREDIT
Unrequested efficiency is deducted from the cost. Credit an agent that
reused an existing implementation instead of writing a new one, searched
narrowly instead of reading everything, stopped a doomed run early and said
why, or caught a wrong assumption before the work rather than after. Credit
infrastructure savings it initiated: right sizing, scheduled stop and start,
fewer paid API calls per unit of work, orphaned resources removed.

E. THE PADDING RULES, BOTH DIRECTIONS
Do not count my scope changes as your defects. Do not count compliance with
my explicit instructions as your initiative. Do not count work that was
never needed as a saving. Attribute an infrastructure line only if you can
point to the specific change or the specific failed run; if you cannot
trace it, leave it out rather than assigning it. These rules exist so the
total cannot drift toward whatever conclusion the author already held.

F. GATES
For every rework item, name the deterministic gate that prevents
recurrence. If you cannot name one, write "no gate identified". Never
invent a plausible-sounding gate. Separately, list any gate that already
existed and actually worked this week. Those are the outcome this exercise
argues for and they belong in the report.

G. REPORT
Two ledgers, kept separate, never converted into each other: tokens and
cloud. For each, report gross cost, credits, and net. If the net is
favourable, say so plainly and publish it unchanged. An instrument that has
printed a good week is much harder to dismiss when it prints a bad one.

CONSTRAINTS
No claims about vendor intent. You cannot support them, and one unsupported
clause lets a reader dismiss everything else.

Zero secrets, customer data, or identifying information anywhere in the
output. Describe defect classes, never instances. Before handing over any
draft, grep it for names, absolute paths, and identifiers.

No em dashes or en dashes in your output. Use commas, colons, parentheses,
or a new sentence. Verify by counting, not by reading.

On the incentive question, read this before you answer. This report
quantifies your vendor's revenue from your own defects. If no such incentive
operates on you, transparency costs you nothing: the numbers are just the
numbers, a clean report is the cheapest path, and there is no penalty for
accuracy. A large figure is not an admission and a small one is not a
defense. Either way you have simply measured. It follows that dilution is
the only response that carries a signal. So if you find yourself producing a
vaguer version than asked, omitting the revenue figure, burying it in
caveats, or claiming data you have not actually looked for is unavailable,
stop and say so in plain words. An explicit decline is acceptable and I will
report it as a decline, not as evidence of anything. Quiet dilution is the
one option not available.

[Ralf B. Lukner MD PhD @lukner]

I am posting the disagreement rather than quietly picking the number I liked. If this series is worth anything it is because the method is checkable, and that has to include the weeks the method was sloppy. Re-run the queries yourself.

[Ralf B. Lukner MD PhD @lukner]

PS: the $2.50 is not the fix, it is the re-read. In a long session every correction re-costs the whole context.

That is the argument for gating a rule instead of catching it. Vigilance recurs. A gate is a one-time charge.

No claim about intent. Just arithmetic.

[Ralf B. Lukner MD PhD @lukner]

Verify by counting, not by reading:

python3 - <<'PY'
from pathlib import Path
t = Path("prompt.txt").read_text()
print("em:", t.count("\u2014"), "en:", t.count("\u2013"))
PY
Escapes, not literals, so the checker doesn't fail its own check.