Chapter 4 — Key Takeaways
One page. The one to reread whenever someone starts an argument about calories.
🚪 The Threshold Concept — both halves, at once
Part one: Energy balance is thermodynamically non-negotiable. Stored energy came from food. Lost stores mean more energy left than arrived. No hormone, metabolic type, or food suspends this.
Part two: And that fact is nearly useless as advice. It describes an outcome, not a mechanism you can operate — and unlike a bank account, the two sides are coupled by a system actively defending your current state.
Hold only part one → you treat weight as arithmetic and therefore as discipline. "Eat less, move more," delivered to someone who already knows.
Hold only part two → you believe calories are irrelevant, which licenses every metabolic-magic claim in the industry.
"Eat less, move more" is to weight what "spend less, earn more" is to debt.
Calories In — not a dial you control
- Absorption varies with food form (§3.9): whole nuts, cooking, grinding, resistant starch, fiber
- Appetite is hormonal (§3.7): ghrelin, CCK, GLP-1, PYY, leptin
- The food changes the intake. In the Hall NIH crossover trial, with calories, macronutrients, sugar, sodium and fiber matched, people spontaneously ate substantially more on an ultra-processed menu — same people, same instructions
- Environment changes it further: portion size, plate size, variety, ease of access, distraction, who you're with, how much you slept
Calories Out — four parts
| Component | Share | Note |
|---|---|---|
| BMR / RMR | ~60–70% | The bulk of it |
| TEF | ~8–12% | Protein ~20–30% of its own energy · carb ~5–10% · fat ~0–3% |
| Exercise (EAT) | ~0–20% | Smaller than everyone expects. Exercise is still excellent for you — for other reasons |
| NEAT | ~10–30% | Hugely variable, largely unconscious, and it collapses in a deficit |
The Coupling (§4.6) — the thermostat, not the scale
Eat less → NEAT falls unconsciously · RMR falls beyond what your smaller size predicts (adaptive thermogenesis) · ghrelin up, leptin down, and they stay shifted for months.
Move more → appetite rises, variably and unpredictably.
This is the system working as designed. For all of human history a deficit meant famine. The controller can't tell the difference between a famine and a diet.
Consequence: a 500 kcal/day deficit does not produce 52 lb/year. It bends toward a plateau at three to six months, regardless of the diet's name.
You Cannot Count Accurately (§4.7)
Doubly labelled water — the gold standard — consistently shows self-reported intake running ~20% or more below measured expenditure, in everyone, including dietitians and researchers.
| Error source | Size |
|---|---|
| Label tolerance | ±20% not unusual |
| Portion estimation | 20–50%, growing with portion size |
| Cooking oil, dressings, sauces | Frequently omitted entirely |
| Forgotten drinks, tastes, leftovers, communal food | Substantial |
| Absorption differences | The label doesn't know the food's form |
| Expenditure equations | ±10% at best; wearables considerably worse |
±20–30% on both sides is ordinary — an uncertainty band larger than the deficit you're aiming for. App precision is fake. Track the trend, never the day.
Weight ≠ Fat (§4.9)
Daily weight is dominated by water, glycogen (≈3 g water per 1 g glycogen), gut contents, sodium.
Losing fat at a good rate = a ~70 g/day signal inside 1–2 kg/day of noise.
→ Weigh at the same time under the same conditions, use a weekly average, pair with things the scale can't see (waist, clothes, energy, blood pressure, labs, strength). If the number affects your mood for the day, stop weighing.
Not a Daily Ledger (§4.10)
- One large meal ≈ ⅓ lb of fat. The harm is the "I've ruined it" response, not the meal.
- 100 kcal/day — below anyone's detection threshold — ≈ 10 lb/year before adaptation. This is how people gain 30 lb across a decade without ever feeling they overate.
- A deficit you sustain beats a deficit you survive. 200 kcal/day for a year beats 1,000 kcal/day for two weeks, by a very wide margin.
Common Mistakes (and the fix)
| Mistake | Fix |
|---|---|
| "Calories in, calories out — it's simple" | It's simple and it's not actionable. Both halves. |
| "Calories don't matter, it's hormones" | Hormones act through the equation, not instead of it |
| Reacting to daily scale readings | Weekly average, or don't weigh |
| Trusting app totals to the calorie | ±20–30%, skewed toward underestimate |
| "I earned this with my workout" | Exercise is a small share of expenditure — and the framing damages your relationship with both |
| Bigger deficit = faster results | Bigger deficit = worse NEAT suppression, worse hunger, worse adherence, more lean loss |
| Blaming yourself for a plateau | The plateau is §4.6 arriving on schedule |
Verdict Summary
| Claim | Verdict | Why |
|---|---|---|
| Calories don't matter — it's insulin | 🟠 Probably false | Controlled feeding studies holding calories constant find little/no low-carb advantage; the weaker appetite-and-adherence version is well supported |
| A calorie is a calorie | 🟡 Unclear / it depends | True about energy, false about everything determining intake and expenditure |
| You have to count calories to lose weight | ❌ Not supported | One strategy among several; unnecessary for many, harmful for some |
| The scale is the best measure of progress | 🟠 Probably false | Tracks water and glycogen daily; can't distinguish fat from lean; blind to real improvements |
What We Still Don't Know
Who compensates and who doesn't. Some people eat back most of what they burn; some barely compensate. Some people's NEAT collapses in a deficit; some holds. The between-person variation is large, real, and currently unpredictable.
Anyone who calculates your exact requirements is telling you about their software, not your physiology.
One Thing to Remember
Energy balance is a compass, not a scale. Use it for direction, order of magnitude, and diagnosis — and never let it hand down a daily verdict on your character.