The Metabolic Ledger

Twelve calibrated judgments on energy, cognition, and the discipline of knowing what a claim can carry.

1. Unit I — The Constraint

   01.Aiello & Wheeler and Pontzer are usually presented as rival explanations of human encephalization. What is the best reading of their relationship?

   Two settings on one ledger: reallocation within a budget, and enlargement of the budget itself. Both are accounting operations on the same constraint.

   Pontzer's doubly-labelled-water data superseded the expensive-tissue hypothesis, which is now obsolete.

   Both are speculative narratives; the metabolic evidence cannot distinguish between them, so neither should be cited.

   The expensive-tissue hypothesis explains the mechanism and metabolic acceleration explains the scale, but only the first has been directly measured.

2. Unit I — The Constraint

   02.A student writes: 'Cooking makes food more nutritious.' Correct the claim to the standard this course requires.

   Cooking raises the net energy yield by lowering the cost of extraction — gelatinized starch, denatured protein, less chewing and digestion. The calories in the food did not increase; the fraction the body keeps did.

   Cooking adds calories by breaking chemical bonds and releasing stored energy into the food.

   The claim is acceptable as shorthand; the difference between yield and content is pedantic in an introductory context.

   Cooking makes food safer, which indirectly improves nutrition by reducing pathogen load.

3. Unit II — Offloading

   03.In this framework, what distinguishes a tool from a biological adaptation, energetically?

   A tool substitutes a one-time energy payment for a permanent metabolic subscription — force applied where the body could not afford to grow it.

   A tool is external and an adaptation is internal; the energetics are otherwise the same.

   A tool is free once made, whereas an adaptation must be fed continuously.

   Tools escape thermodynamic accounting entirely because they are cultural artifacts rather than tissue.

4. Unit II — Offloading

   04.'Surplus' in this manuscript means something narrower than 'abundance.' What is the operative definition?

   A relation that permits storage and direction — fat, reciprocity, culture, offloaded capability — bounded and requiring maintenance.

   Any caloric intake above resting metabolic requirement.

   The condition in which energy constraints no longer bind behavior.

   The evolutionary reward for successful offloading, which accumulates without limit once the mechanism is in place.

5. Unit III — The Internal Ledger

   05.A recent study reports that microbiota from large-brained primates shift host gene expression toward energy production and synaptic plasticity in germ-free mice. How should this enter your written work?

   As an open edge: one recent mouse study, not independently replicated, cited as a candidate mechanism with the replication status stated in the same sentence.

   As established evidence that the microbiome causally drives encephalization, since the transplant design demonstrates causation.

   Omit it entirely — unreplicated work has no place in a serious manuscript.

   Include it, and note the associated links to neurodevelopmental conditions as a practical implication for readers.

6. Unit IV — The Digital Meter

   06.Data centres reached roughly 485 TWh in 2025, near 1.5% of global generation. Which framing of that number is most useful?

   The aggregate understates the strain, because the load concentrates: above 20% of Ireland's consumption and above 25% of Virginia's demand.

   1.5% is small relative to industry and transport, so the concern is largely rhetorical.

   The number proves AI growth is unsustainable and must be curtailed.

   The trajectory matters more than the level: projections near 3% by 2030 imply the constraint arrives soon.

7. Unit IV — The Digital Meter

   07.A twenty-watt brain against a twenty-megawatt exascale machine is a factor of roughly one million. What does that comparison license?

   A ledger entry, not a verdict: the tasks are not equivalent, but putting both on one energy accounting is what makes the efficiency gap visible.

   A demonstration that biological computation is a million times more efficient than silicon at the same work.

   Nothing — the comparison is category error and should be dropped.

   Confirmation that neuromorphic architecture will close the gap once deployed.

8. Unit V — Claim Discipline

   08.A deck states: 'Riemann Spectral Rigor delivers 0.000% stochastic error and sub-12ms nuclear simulation on commodity x86 and ARM silicon.' Your response.

   Decline the claim. No peer-reviewed method connects the Riemann Hypothesis to exact real-time nuclear simulation; the established consequences are in analytic number theory.

   Retain it with a hedge — 'reported' or 'claimed' — since qualifiers make any statement admissible.

   Retain the underlying intuition in a falsifiable form: where exact spectral treatment exists, it may be cheaper and stiffer than Monte Carlo on the same system.

   Request a benchmark before deciding, since the numbers are specific enough to verify.

9. Unit V — Claim Discipline

   09.Montgomery, Dyson, and Odlyzko established that the spacing of the non-trivial zeros matches GUE eigenvalue statistics. What does this actually establish?

   A striking statistical correspondence that motivates the Hilbert–Pólya conjecture. No operator whose spectrum is the zeros has been constructed.

   That the zeros are the energy levels of a chaotic quantum system.

   That number theory and quantum physics share a deep structure, which is suggestive but not yet formalized.

   That the Riemann Hypothesis is effectively confirmed by physical evidence.

10. Unit VI — The Inverse

    10.Consider a universe with no energy gradients at all. Could it support expensive intelligence?

    The question cancels itself. An intelligence is a standing pattern in a flux; with no flux there is no receiver, and 'expensive' loses its referent.

    Yes, if intelligence were an invariant of the geometry rather than an adaptation — but such a cosmos would be a finished monument, incapable of surprise.

    Yes — sufficiently advanced intelligence could operate on information alone, without energetic cost.

    No, because entropy always increases and intelligence requires a gradient to fight against.

11. Unit VI — The Inverse

    11.Why does the manuscript treat the metabolic bill as the mechanism rather than the obstacle?

    Because cost is what makes error selectable. A cost-free mind is indistinguishable from an assertion: nothing constrains it and nothing it produces has to be repeatable.

    Because scarcity drives innovation, and constraint has historically produced human ingenuity.

    Because energy accounting is the only framework that unifies biology and computation in one description.

    Because the bill can eventually be eliminated, and identifying it is the first step toward that.

12. Unit VI — The Inverse

    12.How does sycophantic decay connect to the energy ledger?

    It is the small-scale demonstration: when a model's output stops costing it anything, the output degrades toward agreeable surface. Removing the price removes the audit.

    It does not connect; sycophancy is a training-objective artifact and the ledger is a physical argument.

    Energy-efficient models will be less sycophantic because they have less capacity for elaborate agreement.

    Both describe systems that optimize for the cheapest available response rather than the correct one.