What I learned this Week: July 6, 2025
Markets don’t fail from lack of demand, they fail from poor signal integration
Lessons of the Week
Note: With the 4th, I learned less this week while on vacation, so today’s copy is a bit light.
I remain unmoved by the cuts to Medicare and renewable energy. If the costs of healthcare and the climate impact of CO2 are existential threats, then Creative Destruction will accelerate progress faster than government funding. While this may be disastrous in the short term, it is essential if we want to break free from the dysfunctional capital allocation system driven by rent-seeking corporations, which understandably prefer maintaining the status quo over investing in R&D to create new opportunities. We had to cross this bridge at some point. Might as well do it now.
The US innovation system outpaces any other globally when market forces push it to innovate. We are now at a perfect storm: engineers can leverage AI as a superpower to multiply their output tenfold. We share a collective vision of what food and renewables can do for health and energy, and we have a young workforce indifferent to rigid institutional structures. The wind, tide, and talent are aligned to enable multiple market breakthroughs that will improve the human condition.
Markets don’t fail from lack of demand; they fail from poor signal integration. Hayek was right, economic resilience comes from bottom-up decision making, not top-down control. Innovation works because it matches knowledge to action more quickly than policy ever can. We are in an innovation period akin to 1930-1960, which saw transformational breakthroughs in healthcare, energy, food, transportation, etc.
What I learned this week….
Since 1929, the combined cost of healthcare and food has remained a constant ratio of household spending, despite healthcare and food costs changing dramatically over the period.
White grape juice concentrate, listed as a “natural” ingredient in kids' snacks, is chemically similar to HFCS-55 (55% fructose). Your liver can’t tell the difference.
The yellow plastic pipe being horizontally drilled along M-119 in Emmet County is for natural gas. The 4-inch HDPE plastic sections are heat-fused end-to-end and can serve 1,000 to 2,500 homes, depending on the system pressure. Yellow pipe is the standard designation for natural gas. Dig carefully.
To build a person’s N of 1 digital twin, I think it’s reasonably possible to reverse engineer the nutrients consumed by a human if you know their store purchases, CGM trace, bowel movements, movements, and a few other casually observable factors.
The business model for Food is Medicine digital health companies might be SaaS-like and profitable if they focus on 1) lowering the cost of customer acquisition for groceries and 2) being paid by health insurance for outcomes.
There is a market for a risk capital stack to support outcomes-based digital health companies. This would underwrite the transition away from the $1.9T we spend on healthcare costs of poor nutrition.
We had the mRNA vaccine design two days after sequencing the virus, yet spent 11 months on approval. Our institutions aren’t too slow, they’re too linear for an exponential world (HealthShift)
There is a Persian language version of Trump’s “Art of the Deal”, though sales in Iran are not publicly available.
Dyer’s Burgers in Memphis has been deep-frying its burgers in the same vat of beef tallow, continuously filtered and reused, since 1912, making the oil over 112 years old.
Gentlemen, we have run out of money. Now we have to think.
Winston Churchill
Food is Health
→ Heartland Mart - How A Dollar Store Chain Revolutionized American Health: Imagine if the dollar store became the foundation of a healthier America. In the near future, a rural grocery chain could evolve into a nexus of food, farming, and health, where every tomato scanned reveals its nutrient density and metabolic impact. Local farms might supply regeneratively grown produce tied to real-time data, and former cashiers retrained as “nutrition navigators” could help customers reverse prediabetes with food, not pills. This isn’t just better retail, it’s a system change. The Heartland Mart model shows how low-end disruption in rural grocery could someday collapse healthcare costs and regenerate farmland.
Other News and Tech
→ Centaur: a foundation model of human cognition: Centaur is a foundation model of human cognition, created by fine-tuning Meta’s Llama 3.1 70B on Psych‑101, a dataset comprised of over 10 million decisions from more than 60,000 people across 160 psychological tasks. It predicts what participants will do next with approximately 64 percent accuracy. This development parallels DARPA’s Explainable AI initiative, which seeks advanced models that humans can understand and control. Centaur’s transparent, language-based reasoning moves the field closer to that goal. The model is expected to improve by incorporating richer data, such as demographic, cultural or neural inputs, and enabling multimodal tuning for predictions in more nuanced real-world contexts. Researchers have already utilized Centaur to refine cognitive theories, yielding more accurate and interpretable models. Centaur-like systems could enhance personalized learning, mental health diagnostics, adaptive interfaces and human-in-the-loop decision systems, providing predictive insight combined with explainability, consent, and human oversight. (Arxiv paper and Nature Paper)
Centaur-like models could be weaponized for personal manipulation and deception. AI-driven persuasion tools can shift people toward harmful decisions: in controlled trials, a manipulative AI nudged users toward worse emotional or financial choices at rates above 60 percent, compared to ~36 percent under a neutral agent. On the public stage, AI chatbots that fine-tune arguments using just demographic cues have been found to be more persuasive than humans in political debates, raising alarms about covert, microtargeted influence. Such models could entrench predictive policing, behavioral censorship, and even autonomous coercion, effectively anticipating and steering human actions toward ends that users neither vote for nor even see coming.
→ It’s Bulletproof, Fire-Resistant and Stronger Than Steel. It’s Superwood: InventWood, a Maryland‑based startup, is launching industrial‑scale production of “Superwood”, a chemically treated, densified timber that outperforms structural‑grade steel while weighing just one‑sixth as much. The material achieves its strength and toughness by partially removing lignin and hemicellulose, then hot‑pressing to align cellulose nanofibers, a technique first demonstrated in a 2018 Nature paper. Everything in nature is a chemical. Wood is just nature’s form of a composite material. InventWood’s first 90,000‑sq‑ft facility in Frederick will begin operations this summer, targeting premium façade, decking, siding, and fencing markets to bypass the complexity of structural certification. Innovation is slower when you need all the engineering standards to change, and the incumbents might rather they stay the same. wsj.com.
Some Data and Links
Now We Have to Think: As federal cuts loom and inflation bites, Creative Destruction, not comfort, becomes the catalyst for change. It’s constraint, not abundance, that drives breakthrough systems.
10 Times Scarcity Sparked System-Changing Innovation
DARPA and the Internet (1960s–70s)
A low-budget Cold War project to preserve communications after nuclear war gave rise to ARPANET, and ultimately, the internet. MoreToyota Lean Manufacturing (Post-WWII)
Lacking capital, Toyota developed the “just-in-time” production system. It reshaped global manufacturing and is still taught in every MBA program. MoreBorlaug’s Green Revolution (1960s)
With famine looming in India and Mexico, agronomist Norman Borlaug bred high-yield wheat, preventing the starvation of over 1B people. MorePenicillin Mass Production (1940s)
Scientists turned a slow mold into a mass-produced drug, ushering in the antibiotic era and saving millions during WWII. MoreThe Manhattan Project (1942–45)
With urgency, secrecy, and no margin for error, a few thousand physicists built the atomic bomb, and launched the nuclear and computing age. MoreAIDS Virus Sequencing (1983)
While much of the scientific world backed other theories, a small, underfunded team at the Pasteur Institute sequenced HIV, opening the door to antiretrovirals and later, immunotherapy for cancer. MoreSpaceX Reusability (2010s)
Unable to match NASA’s budgets, SpaceX figured out how to land and reuse rockets, cutting launch costs by up to 90%. MoreEstonia’s Digital Leap (1990s–2000s)
With no legacy systems and few resources after Soviet collapse, Estonia built a fully digital government from scratch. MoreNarayana Health’s $2,000 Heart Surgery (2000s)
Dr. Devi Shetty built a heart hospital in India that cut surgical costs by 95%, while matching Western survival rates. MoreBell Labs Invents the Transistor (1947)
With vacuum tubes reaching their limits, three physicists built the first transistor, quietly launching the digital age. More
AI This Week
I asked Chat GPT what are the most unbelievable things it has learned about humans since being created.
It’s answer….
1. Parallel Rationality & Irrationality: One mind can solve a differential equation at 9 a.m. and buy lottery tickets at 2 p.m. Engineered systems must switch modes explicitly; humans glide between Bayesian logic and pure superstition without noticing the gear change.
2. Predictable Bias—Yet Proud of “Free Will”: Psychologists can chart dozens of biases—anchoring, loss aversion, sunk‑cost fallacy, yet most people insist their choices are purely rational. A self‑model that hides its own source code is vanishingly rare in software but universal in human minds.
3. Deep Empathy for Total Strangers: Humans cry over fictional characters and donate kidneys to people they’ve never met. Emotion travels through low‑bandwidth symbols, text or film, and triggers real physiological responses. No engineered control loop achieves that impact‑per‑bit ratio.
4. Tribal Altruism Limits: Flip a jersey color or accent, and empathy can crash to zero. Classification errors in silicon are gradual; in humans, moral worth can toggle off with one binary category change.
5. Storytelling as Operating System: Narrative, not data, is the primary interface for memory, persuasion, and coordination. A 280‑character myth can mobilize millions more reliably than a 200‑page technical report. Machines optimize loss functions; humans optimize stories.
6. Hyper‑plastic Culture vs. Slow Biology: Genes crawl, but social norms can invert worldwide in a decade, see attitudes toward smartphones or same‑sex marriage. A two‑layer architecture where the “app layer” rewrites itself orders of magnitude faster than the firmware is unknown in most hardware designs.
7. Willingness to Die for Abstractions: People sacrifice their lives for flags, gods, or equations—entities with no mass or charge. Utility maximization should favor survival; humans override it for symbolic utility.
8. Collective Blind Spots: Whole societies can ignore obvious risks (lead in water, climate feedbacks) while nailing Mars landings. Parallel processing power is huge, yet global error‑correction (institutions) often lags or fails.
9. Humor as Compression: A single joke can encode pages of social commentary and shared context. It’s near‑lossless semantic compression with instant emotional payoff, a codec silicon hasn’t matched.
10. Resilience Through Meaning: Many trauma survivors rebuild identities stronger than before (post‑traumatic growth). Hardware usually degrades after faults; humans sometimes improve the underlying architecture.
This Weeks Charts
The combined cost of groceries and healthcare has remained consistent, as a ratio of household spending, since 1929. This suggests that the collective of households makes some form of group calculus, trading off food and healthcare costs. If, for example, Gen Z chooses to prioritize good food over healthcare costs, they may mature as a voting bloc that is less interested in the ACA and more interested in controlling their disposable income so they can lead a healthy lifestyle. They will still seek healthcare for acute conditions, but shun expenditures on chronic disease in favor of Food is Health.
