Chapter 7 — Carbohydrates: Sugars, Starches, Fiber — The Fuel Source That Got Unfairly Demonized

The Hook: "She told me to stop cooking rice"

Camila Ortiz came in on four hours of sleep, which is normal for her — she'd finished a night shift at seven that morning and had an appointment at eleven, and she'd made the entirely reasonable decision to come to the appointment rather than sleep.

She wanted to talk about her husband.

Erik cooks four nights a week. Rice is on the table most of those nights, because it's cheap, because it's what his mother made, because Alma is four and will eat approximately nine foods and rice is one of them, and because at 6:40 p.m. with two children and a $180 weekly grocery budget, rice is a thing you can do.

A colleague of Camila's on the unit — well-meaning, certain, currently on her own successful low-carb regime — had told her the rice was the problem. Cut the rice. White rice is basically sugar. It's spiking your blood sugar. That's why you're tired all the time.

Camila had tried. For three weeks, Erik made cauliflower rice, which cost more, took longer, and which Alma would not eat, so Alma had toast instead, which is also a carbohydrate, so the whole exercise had achieved nothing except a more expensive dinner and a child eating less well.

Then she stopped, and felt like she'd failed at something.

Here's what I actually said to her, and it's the shortest useful summary of this chapter:

"The rice is not why you're tired. You're tired because you work three twelve-hour night shifts a week and sleep five and a half hours in daylight. And there is nothing wrong with rice."

I want to be careful about what I'm defending here, because "carbs are fine" is not the position and this chapter isn't a rehabilitation campaign.

Carbohydrate is a category so broad as to be nearly meaningless. It contains lentils and it contains Coca-Cola. It contains the fiber in an apple and the glucose syrup in a sports gel. Saying "carbs are bad" is like saying "liquids are bad" — technically it includes both water and bleach, and the category has told you nothing.

What this chapter does is replace the useless distinction with a useful one. You have been taught "simple versus complex," which is nearly worthless. The distinction that actually predicts health outcomes is intact versus refined — and once you have it, most of the carbohydrate wars stop being interesting.

Along the way: what glycemic index is actually worth (less than the apps imply), how much carbohydrate a human genuinely requires (a stranger answer than either camp wants), whether blood sugar "spikes" in a healthy person matter (mostly no), and the honest history of how the most widely eaten macronutrient on earth became the villain of the twenty-first century.

🏃 Fast Track: §7.3 (intact vs refined — the one that matters), §7.6 (how much you need), and §7.10 (what to do). Twenty minutes.

🔬 Deep Dive: §7.4 (glycemic index and its problems), §7.7 (the glucose response and what a "spike" means), and §7.9 (whole grains) are where students, coaches, and anyone arguing about this online should spend time. §7.11 (FODMAPs and sugar alcohols) matters most if you have gut symptoms; §7.12 is the label mechanics.

7.1 What a carbohydrate actually is

The chemistry, in one paragraph, and then we're done with it.

Carbohydrates are molecules built from sugar units. That's it. The categories are defined by how many units are stuck together:

Two things worth extracting.

First: starch is just a lot of glucose molecules chained together. Your amylase (Chapter 3) breaks those chains apart. So bread, rice, potato, and pasta all arrive in your bloodstream as glucose — the same molecule that arrives from table sugar. This is true and it is the single most misused fact in nutrition, because people conclude from it that starch and sugar are equivalent. They are chemically similar and nutritionally very different, and §7.3 explains why.

Second: fiber is a carbohydrate you can't digest. It's on the label under "total carbohydrate" because chemically it belongs there, and it behaves nothing like the rest of the category. Chapter 11 is entirely about it.

7.2 "Simple versus complex" — retire it

You were taught this in school. Simple carbohydrates = sugars = bad and fast. Complex carbohydrates = starches = good and slow.

It is a poor guide and you should stop using it.

Watch it fail on four foods:

Two "complex" foods behave completely differently from each other. A "simple" food behaves better than both of them.

The distinction fails because it's about chemistry when the thing that matters is physics — how fast enzymes can reach the starch, which depends on the food's structure, not its molecular length. You met this in Chapter 3 §3.9: the same molecules, differently packaged, are differently available.

Replace it.

7.3 The distinction that actually works: intact versus refined

Here's the frame that does the work for the rest of the chapter.

The useful question is not "what kind of sugar molecule?" It's "how much structure is still around it?"

Intact means the carbohydrate is still inside its original plant architecture — cell walls, fiber, protein, fat, water, and micronutrients arranged as they grew.

Refined means that structure has been dismantled: the grain milled fine, the bran and germ removed, the fruit juiced, the starch extracted.

That last row is doing more work than any of the others, and it's the one nobody puts in a textbook.

🔍 Why this works. Consider how long it takes to eat 300 calories in four forms.

300 kcal of apples is roughly three large apples. That's fifteen to twenty minutes of active chewing and about 750 ml of volume in your stomach, and most people simply stop before finishing them.

300 kcal of apple juice is about 600 ml. Two minutes. No chewing, minimal gastric distension, and most of the fiber gone.

300 kcal of steel-cut oats takes ten minutes to eat and empties from the stomach over hours.

300 kcal of cornflakes takes four minutes and leaves you hungry by ten a.m.

Same macronutrient category. Same approximate energy. Radically different eating rate, satiety, and subsequent intake. This is why the refined/intact axis predicts real outcomes and the simple/complex axis doesn't — and it's a mechanism you can verify on yourself this week without reading a single study.

🔄 Check your understanding. Rank these four by how quickly they'd raise blood glucose, and give your reason: boiled lentils · white rice · a banana · white bread.

Answer

Roughly: white bread ≈ white rice > banana > lentils.

The reasoning is what matters. White bread and white rice are refined starch with the structure already dismantled — amylase reaches them immediately. Banana is mostly sugars but comes with fiber and intact cell structure, and its response varies with ripeness (a greener banana contains more resistant starch, §7.5). Lentils are the slowest by a wide margin — high fiber, high protein, and an intact cell structure that resists digestion.

Notice that the two "complex carbohydrates" occupy the top and bottom of the ranking, and the "simple" one sits in the middle. That's the simple/complex framework failing in a single line.

📊 Diagram (described). Picture a single wheat grain, magnified, in cross-section. It has three parts, arranged like a seed because that is exactly what it is.

On the outside, the bran — a tough multi-layered husk, mostly fiber, carrying most of the grain's B vitamins, minerals, and antioxidants. Inside that, occupying the great bulk of the grain, the endosperm — a dense white starch store, the seed's packed lunch for germination, with some protein and very little else. And tucked at one end, small, the germ — the embryo itself, rich in fat, vitamin E, folate, and more minerals.

Now watch what milling does. White flour is the endosperm alone, with the bran and germ removed and discarded (or sold separately as bran and wheatgerm, at a markup, to people trying to add them back). You have kept the starch and thrown away most of the fiber, most of the micronutrients, and all of the structure. Then you grind the endosperm to a powder, which increases the surface area your amylase can attack by an enormous factor.

That's the whole of §7.3 in one picture. Refining is not a chemical transformation — it's a subtraction and a grinding. And it explains why "enriched" flour, which has a handful of B vitamins and iron added back by law, is still not the same thing: you can restock four nutrients, but you cannot put the architecture back.

Hold the same image over a potato (skin and flesh, then mashed), an orange (segments, membranes and pith, then juiced), and an oat groat (whole, then steel-cut, then rolled, then instant). The same subtraction, running at different speeds.

🔄 Check your understanding. Steel-cut oats, rolled oats, and instant oat sachets are all "whole grain oats" and all can carry that claim legally. Why do they behave differently, and which distinction from this chapter explains it?

Answer

All three retain the bran and germ, so all three are genuinely whole grain. What differs is particle size and physical processing — the grinding half of the diagram above, not the subtraction half.

Steel-cut oats are groats chopped into a few coarse pieces; rolled oats are steamed and flattened; instant oats are pre-cooked, rolled much thinner, and often cut finer still. Each step increases the surface area available to amylase and reduces the time needed to hydrate and cook — which is exactly what makes instant oats convenient and exactly what makes their glucose response faster.

This sits on the intact-versus-refined axis, and it shows that the axis is a continuum rather than a binary. "Whole grain" tells you a subtraction hasn't happened. It tells you nothing about how finely the remainder has been ground — which is why a whole-grain instant cereal and a bowl of steel-cut oats can carry the same label and behave quite differently.

(And many instant sachets add sugar on top, which is a separate issue that the whole-grain claim also doesn't tell you about.)

7.4 Glycemic index: useful, oversold, and misused

If intact-versus-refined is about structure, glycemic index was the attempt to measure the consequence directly.

What it is. Developed in the early 1980s by David Jenkins and colleagues in Toronto, GI ranks foods by how much they raise blood glucose over two hours compared to a reference (pure glucose = 100), when eaten alone, in a quantity containing 50 g of available carbohydrate, by fasted subjects.

Glycemic load improves on it by accounting for portion: GL = (GI × grams of carbohydrate) ÷ 100. This matters enormously — watermelon has a high GI and a low GL, because a slice of watermelon is mostly water and contains very little carbohydrate.

GI was a genuinely good idea and it has real uses. It also has five problems that together mean you should not organize your eating around it.

Problem 1: nobody eats one food alone, fasted. GI is measured in isolation. Add fat, protein, acid, or fiber to a meal and the response changes substantially. The GI of white rice tells you little about the response to white rice with beans, chicken, and vegetables — which is how rice is actually eaten almost everywhere on earth.

Problem 2: the values are not stable. The same food varies with variety, ripeness, growing conditions, processing, cooking time, and cooling. A potato's GI depends on whether it's boiled, baked, mashed, or roasted, and on whether it's eaten hot or cold. Published GI values also vary between testing laboratories for the same food.

Problem 3: individual variation is large. Different people produce genuinely different glucose responses to identical foods — a finding demonstrated most memorably by work from the Weizmann Institute in which participants' responses to the same standardized meals diverged widely, with some people spiking on bread and not on ice cream and others the reverse. That's real, it's interesting, and Chapter 35 asks whether we can yet do anything useful with it.

Problem 4: low GI does not mean healthy. Ice cream, chocolate, and full-fat milk all have lower GIs than boiled potato, because fat slows gastric emptying. A GI-based hierarchy will hand you some genuinely odd rankings.

Problem 5: the outcome evidence is weaker than the enthusiasm. Trials of low-GI diets generally show modest effects on glycemic markers and inconsistent effects on body weight and hard outcomes. Where low-GI diets help, much of the benefit tracks with the fact that low-GI foods tend to be intact, high-fiber foods — which brings you back to §7.3 without needing the index at all.

🔬 Claim → Evidence → Verdict

The claim: "Glycemic index tells you which carbohydrates are healthy. Choose low-GI foods."

Where it comes from: A genuinely valuable research tool, developed for good reasons, with real clinical applications — GI and especially glycemic load remain useful in diabetes management and in research contexts.

What the evidence actually shows: As a personal food-selection system, it underperforms. Values are unstable, they're measured in a way nobody eats, individual responses vary widely, and low GI doesn't track healthfulness — several high-fat confectionery items outrank potatoes. Meanwhile, the simpler heuristic — eat carbohydrate that still has its structure — captures most of the same benefit with none of the lookup tables.

📉 Evidence quality: Rung 6 for the measurement; rung 5–6 and mixed for low-GI diets on hard outcomes.

Verdict: 🟡 Unclear / it depends. Genuinely useful in clinical diabetes care and research; a poor everyday organizing principle for a general reader. If you find yourself looking up a GI value before eating, the intact/refined question would have gotten you there faster.

7.5 Resistant starch and the fiber overlap

A brief detour, because it's where carbohydrate and Chapter 11 meet.

Resistant starch is starch that escapes digestion in the small intestine and arrives in the colon, where it's fermented like fiber (Chapter 3 §3.6), producing short-chain fatty acids and feeding the gut lining. It yields less energy than digested starch and behaves, functionally, like fiber.

There are several types, but two matter practically:

• Starch physically protected by intact structure — whole or coarsely cracked grains, whole legumes, seeds. §7.3 again.
• Retrograded starch — starch that has been cooked and then cooled. As it cools, some of it recrystallizes into a form your enzymes handle poorly.

That second one is the origin of a genuinely true internet claim: cooking and cooling rice, potatoes, or pasta increases their resistant starch content, and reheating them doesn't fully reverse it. Cold potato salad, day-old rice, and pasta salad genuinely contain more resistant starch than the freshly-cooked versions.

And the effect is real and small. Estimates of the increase are modest — a few percent of the total starch — which slightly reduces available energy and slightly blunts the glucose response. It is not a mechanism by which you can eat unlimited pasta, and the internet's enthusiasm has run considerably ahead of the numbers.

(One safety footnote: cooling cooked rice is fine; leaving it at room temperature for hours is not, because Bacillus cereus spores survive cooking and multiply in warm rice. Cool it quickly, refrigerate it, and reheat it thoroughly.)

7.6 How much carbohydrate do you actually need?

Now the question everyone wants answered, and the answer has three layers that get conflated.

Layer 1 — the strict biochemical requirement: essentially zero.

Chapter 6 established this. There are essential amino acids and essential fatty acids; there is no essential carbohydrate, because gluconeogenesis can manufacture glucose from lactate, glycerol, and amino acids, and ketones can substitute for much of the brain's demand.

Layer 2 — the RDA: 130 g/day for adults.

This number surprises people who've absorbed layer 1. The Dietary Reference Intakes set an RDA for carbohydrate of 130 g/day for adults (higher in pregnancy and lactation), based on the average minimum glucose used by the brain.

So which is it? Both, and the apparent contradiction is instructive: the RDA describes the amount that removes the need for the body to manufacture glucose. It doesn't describe an amount below which you cannot survive — it describes the amount below which your body has to start doing extra work, and paying for it in amino acids.

Layer 3 — the AMDR: 45–65% of energy.

The Acceptable Macronutrient Distribution Range is a population-level guideline, not a personal prescription, and it's wide for a reason: the evidence doesn't support a narrow optimum. For Theo at 2,200 kcal, 45–65% is 248–358 g/day — an enormous span, and both ends are defensible.

So what's the honest answer?

Notice what's missing from that table: an optimum. There isn't one, the evidence doesn't support one, and anyone who gives you a single number is giving you a preference.

🧩 Productive struggle. Four minutes before reading on.

Populations eating very high carbohydrate diets — traditional Okinawan (heavily sweet-potato-based), rural Chinese (rice-based), Kitavan (tuber-based) — have historically shown low rates of obesity and type 2 diabetes.

Populations eating very low carbohydrate diets — traditional Inuit, Maasai (largely meat/milk/blood) — have also historically shown low rates of those conditions.

Both are true. What does this tell you, and what does it not?

What I'd say

What it tells you: carbohydrate quantity is not the primary driver of metabolic disease. Humans thrive across an enormous range — from perhaps 5% to over 80% of energy — which is strong evidence against any claim that a particular macronutrient ratio is uniquely correct for our species.

What it does not tell you: almost anything about which diet you should eat. Every one of these populations differs from a modern industrialized one in a dozen other ways at once — physical activity, total energy intake, ultra-processed food (essentially absent), food variety, meal frequency, alcohol, smoking, sleep, chronic stress, and often genetics and life expectancy. This is Chapter 2's confounding problem in its purest form.

And watch for how these populations get used. Low-carbohydrate advocates cite the Inuit; high-carbohydrate advocates cite the Okinawans; each side treats the other's example as an anomaly. The honest reading is that both are evidence against the other's absolutism and neither is evidence for anyone's protocol.

One more thing the traditional diets share, and it's probably the answer: essentially no ultra-processed food. Chapter 22.

7.7 What a "blood sugar spike" actually is

Continuous glucose monitors have made this a live consumer question, so it's worth being precise.

When a healthy person eats carbohydrate, blood glucose rises and then returns to baseline. That is not a malfunction. That is the system working — the whole purpose of insulin is to move glucose out of the blood and into cells. A postprandial rise is as normal as a rise in heart rate when you climb stairs.

Typical figures in a person without diabetes: fasting glucose around 70–99 mg/dL (3.9–5.5 mmol/L), peaking after a mixed meal somewhere below about 140 mg/dL (7.8 mmol/L), returning toward baseline within two to three hours.

What is genuinely concerning is not the excursion but the failure to clear it: chronically elevated fasting glucose, an HbA1c creeping upward (Theo's is 5.9%, in the prediabetes range of 5.7–6.4%), and excursions that are large and prolonged because insulin isn't working properly. That's insulin resistance, and it's Chapter 26.

🔬 Claim → Evidence → Verdict

The claim: "Blood sugar spikes damage your body even if you don't have diabetes. You should eat to keep your glucose curve flat."

Where it comes from: Two real things. First, sustained hyperglycemia genuinely does cause damage — that's the entire pathology of diabetes, and it isn't in question. Second, some observational data associates greater post-meal glucose excursions with cardiovascular risk markers, even within ranges considered normal.

What the evidence actually shows: The leap is from "very high, sustained glucose is harmful" to "any rise is harmful," and it isn't supported. Normal postprandial excursions in metabolically healthy people are physiology, not pathology. The observational associations are subject to the usual Chapter 2 problems — people with larger excursions differ in weight, activity, and fitness. And crucially, there is no trial evidence that a healthy person who flattens their glucose curve improves any hard outcome. Meanwhile the strategies sold to flatten it — eating vegetables first, vinegar before meals, walking after eating — are individually harmless to mildly beneficial and are being marketed with a confidence the evidence doesn't carry.

📉 Evidence quality: Strong for pathological hyperglycemia. Rung 5, confounded, for excursions in healthy people. Absent for outcome benefit from flattening curves in the non-diabetic.

Verdict: 🟡 Unclear / it depends. Genuinely important if you have diabetes or prediabetes. Currently unsupported as a concern for a metabolically healthy person, and the consumer CGM market is running well ahead of the evidence. Chapter 35 takes this on directly.

⚠️ When to see a professional. If you have symptoms of high blood glucose — excessive thirst, frequent urination, unexplained weight loss, blurred vision, recurrent infections, slow healing — see a physician rather than buying a monitor. Diagnosis of diabetes and prediabetes uses HbA1c, fasting glucose, or an oral glucose tolerance test, not a consumer CGM, and a consumer device should never be used to self-diagnose or to adjust diabetes medication.

7.8 How carbohydrate became the villain

The history is short, and knowing it defuses a lot of argument.

Roughly 1980–2000: the low-fat era. Dietary guidance emphasized reducing fat, particularly saturated fat. The food industry responded, as it always does, by reformulating — and when you remove fat from a product you have to replace it with something to keep it palatable. That something was usually sugar and refined starch. Supermarkets filled with low-fat, high-sugar products marketed as healthy.

Obesity and type 2 diabetes rose through this period. What followed was a completely understandable inference and a mistaken one: we cut fat, we got fatter, therefore fat wasn't the problem — carbohydrate was.

2000s: the low-carbohydrate counter-revolution. Atkins, then South Beach, then paleo, then keto, then carnivore. Each wave arriving with a mechanism (insulin), a villain (carbohydrate), and — this is the part that made them work — a set of foods that were genuinely worth removing. Cutting carbohydrate meant cutting sugary drinks, white bread, pastries, breakfast cereal, and most ultra-processed food. People felt better because they were eating better, and attributed it to the macronutrient.

2010s–2020s: the mechanism goes mainstream. Insulin, then blood glucose, then continuous glucose monitors — each providing a measurable-looking number and a purchasable device.

💡 Aha moment. Notice the structure of the mistake, because it's Chapter 2's substitution question wearing a different hat. The low-fat era didn't fail because fat is good. It failed because of what replaced the fat. And the low-carbohydrate era succeeds, where it succeeds, largely because of what gets replaced when carbohydrate goes.

Neither era was about a macronutrient. Both were about the food that got substituted in — and whichever camp you're in, that's the uncomfortable conclusion, because it means the macronutrient argument you've been having for twenty years was about the wrong variable.

🔬 Claim → Evidence → Verdict

The claim: "Carbohydrates make you fat."

Where it comes from: Real observations. Carbohydrate intake drives insulin, insulin promotes fat storage, and people who cut carbohydrate frequently lose weight. Millions of people have experienced this personally, which is why it's so persistent.

What the evidence actually shows: Controlled feeding studies holding calories and protein constant while varying carbohydrate generally find little or no fat-loss difference (Chapter 4 §4.6; Chapter 6 §6.8). Populations eating 70–80% of energy as carbohydrate have historically shown low obesity rates. What does happen when people cut carbohydrate in the real world is that they cut sugary drinks, refined snacks, and ultra-processed food — reducing energy intake and increasing satiety per calorie. That's a real and useful effect operating through intake, not evidence of a uniquely fattening macronutrient.

📉 Evidence quality: Rung 6 controlled feeding trials on the mechanism; rung 5 population data; rung 6 outcome trials in Chapter 10.

Verdict: 🟠 Probably false as stated. Refined, energy-dense, rapidly-eaten carbohydrate makes overeating easy. That is a claim about food form, not about a molecule.

7.9 Whole grains: what the evidence actually shows

Of everything in this chapter, this is where the evidence is strongest and the online noise loudest.

The evidence for whole grains is unusually good by nutrition standards. Prospective cohorts across many countries and populations consistently associate higher whole grain intake with lower rates of cardiovascular disease, type 2 diabetes, colorectal cancer, and all-cause mortality. There's a dose-response relationship — one of Chapter 2 §2.10's markers of a finding worth taking seriously. Trial evidence on intermediate outcomes (LDL cholesterol, inflammatory markers, glycemic measures) generally points the same direction. And there's a plausible mechanism: fiber, resistant starch, micronutrients, and the food-matrix effects of §7.3.

Cohort data plus dose-response plus trials on markers plus mechanism, converging. That's the convergence criterion, met.

The honest caveats, because this is a book that states them: the cohort data carries healthy-user bias like everything else — whole grain eaters do many other healthy things. And "whole grain" as a label is applied loosely; a sweetened cereal with some whole grain flour in it is not a bowl of steel-cut oats.

🔬 Claim → Evidence → Verdict

The claim: "Whole grains are no better than white bread — the phytates block mineral absorption and the lectins damage your gut. Grains are inflammatory."

Where it comes from: Two real chemical facts. Phytate does bind minerals — iron, zinc, calcium — and reduces their absorption. Lectins are real proteins, and some are genuinely toxic: raw or undercooked kidney beans contain phytohaemagglutinin and cause real, documented food poisoning.

What the evidence actually shows: Both facts are true and neither supports the conclusion. Phytate's effect is real, modest, and substantially reduced by ordinary preparation — soaking, sprouting, fermenting (which is what sourdough does), and cooking. In populations where mineral deficiency is a genuine problem, the driver is overall dietary inadequacy, not the presence of whole grains; and whole grains supply more of those minerals to begin with, which partly offsets the reduced absorption. Phytate also has plausible benefits as an antioxidant.

Lectins are largely destroyed by cooking. Nobody eats raw kidney beans; the poisoning cases involve undercooked or slow-cooked-at-low-temperature beans. Properly cooked legumes and grains are among the most consistently health-associated foods in the entire literature — which is a difficult fact for a theory that they're damaging.

And the outcome data settles it. Whatever phytates and lectins do at a molecular level, the people eating the most whole grains and legumes have better outcomes across many populations and decades. A mechanism that predicts harm and an outcome that shows benefit means the mechanism has been over-weighted — Chapter 2's rung 1 versus rung 5–8 problem, exactly.

📉 Evidence quality: Rung 1 mechanistic claims versus rung 5 and 6 outcome evidence pointing the other way.

Verdict: ❌ Not supported. Cook your beans properly. That's the entire practical implication.

🧾 Cost check. Worth pricing, because this argument has a class dimension nobody mentions. Dried lentils run roughly $1.50–$2.50/lb and deliver, per dollar, more protein, fiber, iron, and folate than almost anything else in a supermarket. Brown rice costs a few cents more per pound than white. Rolled oats are among the cheapest breakfasts available anywhere.

Meanwhile a grain-free "paleo" bread runs $7–$10 a loaf, and almond flour is roughly ten times the price of wheat flour.

The foods being demonized are the ones a family on $180 a week can actually afford, and the replacements being sold are among the most expensive in the store. That's not an argument that the claims are wrong — it's a reason to check them very carefully before repeating them to someone whose budget can't absorb being wrong.

7.10 So what should you actually do?

Six things, in descending order of how much they matter.

1. Prioritize intact over refined. This is the whole chapter. Whole grains, legumes, whole fruit, potatoes with skin, oats that still look like oats. Not because refined carbohydrate is poison, but because intact carbohydrate is harder to overeat and brings fiber and micronutrients with it.

2. Drink fewer calories. Sugar-sweetened beverages are the one part of the carbohydrate category where the evidence is genuinely strong and genuinely negative — they deliver energy with almost no satiety return. Chapter 18 gives the numbers. If you change one thing from this chapter, change this.

3. Eat legumes. Beans, lentils, chickpeas. Cheap, filling, high-fiber, high-protein, and among the most consistently health-associated foods in the entire literature. Most people in industrialized countries eat almost none.

4. Don't fear fruit. The fructose-in-fruit argument dies on §7.3 — whole fruit comes with fiber, water, volume, and structure. Fruit intake is associated with better outcomes essentially everywhere anyone has looked. Chapter 18 handles the fructose question properly.

5. Match your intake to your activity. If you train hard, you need carbohydrate (Chapter 23). If you're sedentary, you probably need less than you're eating. Neither is a moral position.

6. Stop optimizing GI, and stop chasing a flat curve unless you have diabetes or prediabetes. The intact/refined heuristic gets you most of the benefit without a lookup table or a $200 monitor.

🍽️ On your plate — Camila's version. So what did I actually tell the Ortiz-Lindqvists about the rice?

Keep the rice. It's cheap, it's cultural, Alma eats it, and Erik can cook it at 6:40 p.m. Removing it would cost money, time, and a child's dinner in exchange for a benefit that isn't there.

Two adjustments, both small. First, when it's easy, use brown rice or mix half and half — genuinely better on fiber and micronutrients, cheap, and not a lifestyle change. Second, and more importantly: what's on the rice matters more than the rice. A serving of beans alongside adds fiber, protein, and satiety for about forty cents, and it's the single highest-yield change available to that household.

And the actual answer to why Camila is tired is that she works three twelve-hour night shifts a week and sleeps five and a half hours in daylight. No carbohydrate adjustment addresses that, and pretending one might is how a person spends three weeks on cauliflower rice and ends up feeling like they failed.

7.11 The carbohydrates nobody warned you about

There's a corner of this category that causes a disproportionate share of real-world misery, and it almost never appears in the macronutrient arguments.

Oligosaccharides — the ones your enzymes can't touch

Remember from §7.1 that oligosaccharides are chains of three to ten sugar units. Humans lack the enzymes to break down several of them — notably the galacto-oligosaccharides in beans and the fructans in wheat, onion, and garlic.

So they travel undigested to the colon, where bacteria ferment them (Chapter 3 §3.6). That fermentation produces short-chain fatty acids, which is good, and gas, which is why beans have their reputation.

These are the "O" in FODMAP — a group of fermentable carbohydrates that trigger symptoms in people with IBS and in a substantial fraction of people who believe they're reacting to gluten. This is Nico's story, and Chapter 28 owns it. What matters here is the principle: a carbohydrate that causes you symptoms is not necessarily a carbohydrate that is bad for you, and the distinction between "I react to this" and "this is unhealthy" gets collapsed constantly.

Sugar alcohols — the reason "sugar-free" has a warning label

Polyols — sorbitol, mannitol, xylitol, maltitol, erythritol — are used to sweeten "sugar-free" products. They're partially absorbed at best, provide fewer calories than sugar, and produce little glucose response.

They also arrive substantially undigested in the colon, where they do two things: draw water into the bowel osmotically and get fermented. The result, at sufficient dose, is bloating, cramping, and diarrhea — which is why sugar-free confectionery carries laxative warnings and why "I ate the whole bag of sugar-free sweets" is a genuinely bad afternoon.

Tolerance varies enormously between people and between polyols. Erythritol is the best tolerated, because it's absorbed in the small intestine and excreted largely unchanged in urine — which is also why it became the sweetener of choice in low-carbohydrate products.

🔬 Claim → Evidence → Verdict

The claim: "Erythritol causes heart attacks and strokes — there's a study."

Where it comes from: A real, well-publicized 2023 study from a Cleveland Clinic group, published in Nature Medicine, which found that higher blood erythritol concentrations were associated with increased risk of major adverse cardiovascular events, and which also reported laboratory evidence of enhanced platelet reactivity. The coverage was extensive and alarming.

What the evidence actually shows: The findings are real and the interpretation is genuinely unsettled, for three reasons. One: confounding. The cohorts studied were people at elevated cardiovascular risk, and people with obesity and diabetes are precisely the people who consume sugar substitutes — so higher erythritol may be a marker of the metabolic disease that causes the events. Chapter 2's healthy-user bias, running in reverse. Two: your body makes erythritol. It's produced endogenously from glucose via the pentose phosphate pathway, and production increases in conditions of metabolic stress — so blood erythritol is not simply a readout of what you ate. Three: the platelet findings are mechanistic and were largely produced at concentrations that may not reflect ordinary dietary exposure.

This is a case where the honest answer is genuinely we don't know yet, and where both the alarm and the dismissal have outrun the data. Further research is underway.

📉 Evidence quality: Rung 5 observational plus rung 1 mechanistic, in a high-risk population, with substantial confounding and an endogenous-production problem. Not nothing; not settled.

Verdict: 🟡 Unclear / it depends. If you use erythritol occasionally, this is not a reason to panic. If you consume large amounts daily and have cardiovascular risk factors, it's a reasonable thing to raise with your physician and to watch as evidence accumulates. Chapter 18 covers non-sugar sweeteners more broadly, and the honest summary there is similar: generally better than the sugar they replace, less thoroughly studied than anyone would like.

7.12 Finding the carbohydrate on a label

A short practical section, because the panel is where this chapter meets reality.

On a US Nutrition Facts panel, under Total Carbohydrate, you'll find indented sub-lines:

Total Carbohydrate          38 g
    Dietary Fiber            6 g
    Total Sugars            12 g
        Includes 8g Added Sugars

Total Carbohydrate includes everything — starch, sugars, fiber, and sugar alcohols. Starch is not listed separately; it's whatever's left after you subtract the others.

Dietary Fiber is the part you don't digest (Chapter 11).

Total Sugars includes both naturally occurring sugars (the lactose in milk, the fructose in fruit) and added ones. This is why plain milk and plain yogurt show sugar on the label despite nothing being added — a fact that causes a great deal of unnecessary alarm.

Added Sugars — introduced in the 2016 US label update — is the genuinely useful line, and it's the one to look at. Chapter 18 gives the targets.

"Net carbs" — a marketing term, not a regulated one

You'll see net carbs on low-carbohydrate products, calculated as total carbohydrate minus fiber minus sugar alcohols.

The logic is partly sound. Fiber genuinely isn't digested to glucose, and sugar alcohols are poorly absorbed, so subtracting them approximates the glucose-raising fraction.

The problems are that it's unregulated and the arithmetic is generous. There's no legal definition of "net carbs" in the US, so manufacturers choose their own method. Some sugar alcohols — maltitol particularly — are meaningfully absorbed and do raise blood glucose, but get subtracted in full. And the resulting number tends to make products look considerably better than they behave.

🍽️ On your plate. The five-second label triage for carbohydrate, which you'll expand in Chapter 30:

1. Read the Added Sugars line, not Total Sugars. Naturally occurring sugar in milk and fruit is not the thing to worry about.
2. Check fiber per serving. For a grain product, roughly 3 g or more per serving suggests it's genuinely closer to intact than refined. Under 2 g on a "whole grain" product tells you the claim is doing more work than the flour is.
3. Look at the first ingredient. "Whole wheat flour" first means something. "Enriched wheat flour" first, with whole grain further down, means considerably less.
4. Ignore "net carbs." It's a marketing calculation, not a measurement.

What we don't know

We don't know why individual glucose responses vary as much as they do. The Weizmann work and subsequent studies show large person-to-person differences in response to identical foods, with microbiome composition, genetics, sleep, prior meals, and physical activity all implicated. We can observe the variation reliably. We cannot yet predict it, and we don't know whether acting on it improves any outcome. Chapter 35.

We also don't have good long-term trial evidence comparing intact versus refined carbohydrate at matched energy intake with hard endpoints — the trials are short, the cohorts are confounded, and the definitive study will probably never be run for the reasons in Chapter 1 §1.2. The intact/refined principle rests on convergence, not proof, and it's worth knowing which of those you're standing on.

Spaced Review

Answer before reading on.

1. (Chapter 3) Why does 300 kcal of apple juice behave so differently from 300 kcal of apples, mechanically?

Physical form changes the rate of delivery. Juice requires no chewing, empties from the stomach fast (Ch 3 §3.3), produces little gastric distension, and provides a weak satiety-hormone signal (§3.7). Whole apples take fifteen minutes to eat, occupy volume, and carry fiber to the colon (§3.6). Chapter 3 §3.9's principle: the same molecules, differently packaged, are differently available.

2. (Chapter 6) Someone says humans need carbohydrate because the brain runs on glucose. You now have two answers — one from Chapter 6 and one from §7.6. What are they?

Chapter 6: gluconeogenesis makes glucose from lactate, glycerol and amino acids, and ketones cover much of the brain's demand — so there's no strict biochemical requirement, though the cost is paid in protein. §7.6: the RDA of 130 g/day exists and is based on brain glucose use — but it describes the amount that removes the need to manufacture glucose, not a survival threshold. Both are true; conflating them is where the argument comes from.

3. (Chapter 2) The whole grain evidence was described as unusually good. Which of Chapter 2 §2.10's markers of convincing evidence does it meet?

Convergence across designs with non-overlapping weaknesses (cohorts + trials on markers + mechanism), dose-response, and consistency across populations and countries. It does not have long-term randomized trials with hard endpoints — and it still carries healthy-user bias. Strong by nutrition standards, which is not the same as proven.

Project Checkpoint: Your Carbohydrate Audit

Component seven, and the start of Phase 2 — Analysis. Use your three-day diary from Chapter 4.

Step 1 — the total. Add up carbohydrate grams for each day and average them. Then convert to a percentage of your energy:

Carbohydrate % of energy = (grams × 4 ÷ total kcal) × 100

Theo's: 316 g × 4 = 1,264 kcal ÷ 3,160 = 40% — squarely inside the 45–65% AMDR's lower neighbourhood, and note that his problem was never the percentage.

Step 2 — the split that actually matters. Go back through your diary and sort every carbohydrate source into two columns. This is the step that produces the insight.

Estimate the rough percentage split. Most people in industrialized countries find they're 70–85% refined, and most are surprised.

Step 3 — the liquid line. Separately, total the carbohydrate you drank: juice, soda, sweetened coffee, sports drinks, alcohol mixers. Write it as a number of grams and as calories.

This one line is, for a lot of readers, the single most actionable finding of the whole audit.

Step 4 — placement. When in your day is your carbohydrate concentrated? Is any of it near activity? No judgment attached — the answer for a sedentary person is different from an athlete's and neither is wrong.

Step 5 — pick exactly one swap. Not five. One intact-for-refined substitution you would genuinely make and keep. Write it down with an honest 1–5 likelihood rating, the way you did in Chapter 4.

Good candidates: white → brown or half-and-half rice · juice → whole fruit · adding a legume to one meal a week · cereal → oats · one sugary drink → water or unsweetened.

Non-tracking alternative. Skip the grams entirely. Instead, list every carbohydrate-containing food you ate over three days and sort it into the intact/refined columns by count rather than weight. You'll get the same insight — and honestly, the ratio matters more than the total anyway.

Next checkpoint (Chapter 8): your protein audit — g/kg, distribution across meals, and sources.

Chapter Summary

Retire "simple versus complex." It's about chemistry when the thing that matters is structure — white bread and lentils are both "complex" and behave nothing alike.

Use intact versus refined instead:

Glycemic index: a good research tool and a poor everyday system — unstable values, measured in isolation, large individual variation, and low GI ≠ healthy. The intact/refined question gets you there faster.

How much do you need? Three layers: strict requirement ≈ zero (gluconeogenesis + ketones) · RDA 130 g/day (the amount that removes the need to manufacture glucose) · AMDR 45–65% of energy (a population range, not a prescription). There is no optimum, and populations have thrived from 5% to over 80%.

A "spike" in a healthy person is physiology, not pathology. What matters is failure to clear, not the excursion.

The history: the low-fat era failed because of what replaced the fat; the low-carb era succeeds, where it does, because of what gets replaced when carbohydrate goes. Neither was ever about a macronutrient.

This chapter's verdicts:

The one thing to remember: the question is never "is this a carb?" It's "how much structure is still around it?"

What's Next

Chapter 8 turns to the macronutrient nobody demonized — which is itself interesting, and which has produced a different failure mode: not fear, but a marketing free-for-all in which protein is added to water.

How much do you actually need, as opposed to the RDA? Why was the protein-combining rule wrong, and why won't it die? How wide is the anabolic window really? Does high protein hurt your kidneys or your bones? And why do older adults need more than young ones, exactly when they want it least?

Chapter 6 §6.10 already gave you the mechanism — protein turnover, a small pool, and a one-way nitrogen leak. Now we put numbers on it.