A honest survey of dietary science: macros, micros, the metabolic narrative, ultra-processed foods, the Mediterranean diet, the gut microbiome — and where the evidence runs thin.
The three molecules that supply energy. The body burns them, builds with them, stores them. We have argued for sixty years about the ideal mix; the data keeps refusing to settle the question.
Glucose is the brain's preferred fuel. Quality matters more than quantity: whole grains, legumes, fruit. Refined sugar and flour behave differently.
Densest energy source. Once vilified, now rehabilitated. Saturated vs. unsaturated still matters; trans fats are unambiguous harm.
Building blocks: muscle, enzymes, hormones. Most active adults underconsume. Plant or animal — both work if amino acids are complete.
Open question: Are all calories equivalent? Thermodynamically yes. Metabolically — satiety, hormonal response, thermic effect — almost certainly no.
Tiny doses. Outsized consequences. The body cannot synthesize most of them — they must arrive on the plate. Deficiency causes named diseases (scurvy, rickets, beriberi). Excess can cause harm too.
Food > pills, in nearly every well-conducted trial.
The first law of thermodynamics holds. But "CICO" hides what matters: how appetite, hormones, sleep, stress, gut flora, and food composition shape both sides of the equation.
Eat carbs → blood glucose rises → insulin secreted → glucose stored. Sharp spikes followed by crashes drive hunger, energy dips, and (chronically) metabolic dysfunction.
Caveat: CGM-driven "personalized nutrition" remains under-evidenced for healthy adults.
Industrial formulations engineered for shelf life, cost, and crave. The NOVA classification calls them group 4: ingredients you don't have in your kitchen, processes you can't replicate at home.
Up from ~24% in the 1970s. Children & lower-income brackets eat the most.
RCT: subjects fed UPF ate ~500 kcal/day more than matched whole-food diet, gained weight in 2 weeks.
Higher UPF intake associates with cardiovascular disease, type 2 diabetes, depression, all-cause mortality.
Honest caveat: "Ultra-processed" is a fuzzy category. Whole-grain bread is technically processed. The mechanism — palatability? additives? texture? — is still being unpicked.
Olive oil, vegetables, legumes, nuts, fish, whole grains, modest wine, little red meat. The PREDIMED trial — despite a republished re-analysis — remains the most credible large RCT for cardiovascular protection.
Compress eating into an 8–10 hour window; let the rest of the day be water and coffee. Animal models showed striking metabolic benefits. In humans, the picture has cooled.
Practical for many; aligns eating with daylight; may aid adherence to modest deficit.
2022 NEJM trial & meta-analyses: when calories are matched, the timing benefit largely disappears.
Some signals of greater lean-mass loss in long-window restrictions; protein distribution matters.
Verdict: Useful as an eating-discipline scaffold for some. Not a metabolic magic bullet. The evidence is honest about that.
The RDA (0.8 g/kg) is a floor, not a target — designed to prevent deficiency, not optimize performance, satiety, or healthy aging.
For an 80kg person: ~96–176g protein/day depending on goal.
~38 trillion bacterial cells. Hundreds of species. They digest fiber we can't, manufacture vitamins, train the immune system, talk to the brain. We are barely past cataloguing them.
Most adults consume half. Fermentable fiber feeds short-chain fatty acid producers — arguably the single most evidenced lever.
American Gut Project signal: diversity of plants > diversity of microbes > metabolic markers.
Some strains help specific conditions; most "general health" claims are not supported. Fermented foods (yogurt, kimchi, kefir) more reliable.
Reality check: "Personalized microbiome diets" sold by startups are running ahead of the science. We don't yet know what a "good" microbiome looks like.
Average intake ~9g. Strong link to hypertension at the population level. The "salt skeptics" debate is real but minority.
Average ~70g. Liquid sugar (soda, juice) most clearly harmful. Sugar in fruit and dairy: not the same problem.
The "1-2 drinks is healthy" J-curve has largely been retracted by better-controlled studies. Less is better; zero is best for cancer risk.
What we know: reducing all three reduces population disease burden. What we argue about: the precise thresholds, and how much they vary individually.
After all the controversies, the durable advice is unglamorous:
"Eat food. Not too much. Mostly plants." — Michael Pollan, still hard to improve on.
Two starting points on YouTube — a general overview and a deep dive on the ultra-processed-food story that has reshaped recent debate.