These foundational commitments generate the governor function. Each axiom produces a constraint on what can be claimed, what can be built, and how fast transitions can happen.
Axiom 1
Physical Laws Are Non-Negotiable
The laws of thermodynamics, conservation of energy, and the properties of materials set absolute boundaries on what is possible. No amount of policy ambition, market incentive, or technological enthusiasm can override them. The second law (entropy always increases in a closed system) is the grandest cosmic generalization — everything else is commentary. When a claim violates a physical law, the claim is wrong. Not approximately, not arguably, but necessarily.
Excludes: all claims that energy transitions can bypass thermodynamic limits
Axiom 2
Orders of Magnitude Determine Truth
If your mental model of a quantity is wrong by 10x or more, no amount of nuance can save your conclusions. A 20% error is a conversation about precision. A 10x error is a conversation about whether you understand the domain at all. Before engaging with any claim, check whether the claimant's numbers are in the right order of magnitude. If they are, proceed to refinement. If they aren't, the discussion is over.
10x wrong = wrong. 1.2x wrong = debatable. The threshold is non-negotiable.
Axiom 3
Infrastructure Has Mass, and Mass Has Inertia
Billions of tonnes of steel, concrete, pipes, wires, vehicles, and buildings cannot be replaced quickly. This is not a failure of will or investment; it is a consequence of mass. Every transition in history (charcoal to coke, wood to coal, coal to oil) has taken decades because the equipment being replaced is physical, heavy, and embedded in networks of interdependent systems. Show me the mechanism that overcomes infrastructure inertia. If you cannot, the proposed timeline is not a plan — it is a wish.
Maps to: trust norms are social infrastructure with comparable inertia
Axiom 4
Power Density Is the Master Variable
Energy flow per unit area per unit time (W/m²) integrates physics, geography, and economics into a single measure. It reveals what price, capacity, and efficiency hide: the land requirements of energy sources, the mismatch between supply and demand, and the fundamental constraints on energy transitions. When two energy systems are compared, power density is the comparison that matters most.
Historical base rates are not anecdotes or analogies; they are the empirical distribution of outcomes. "Energy transitions take 50-75 years" is not an estimate; it is the observed range across every major transition in recorded history. "Efficiency gains have never reduced total consumption" is 160 years of data since Jevons (1865). When your model disagrees with the historical distribution, your model is wrong until proven otherwise.
Novel claims require novel mechanisms — not just novel intentions
Axiom 6
Every Number Needs a Denominator
"Solar capacity grew 30%!" — 30% of what? As a fraction of what total? Per unit of what input? At what cost? The denominator is where the truth lives. Reporting a numerator without its denominator is not information; it is advocacy disguised as measurement. The "compared to what?" discipline applies to every claim, every metric, every projection. Remove the denominator, and you remove the ability to evaluate.
Numerator without denominator = advocacy, not analysis
Axiom 7
Forecasting Complex Systems Is Epistemically Bankrupt
Long-range forecasts (>10-15 years) "tend to fail in a matter of years, sometimes months." The problem is not bad forecasters; it is the nature of complex systems. Valid operations: state constraints, identify entrenched trends, calculate physical limits. Invalid: point predictions, specific timelines for immature technologies, "this time is different." Nuclear fusion has been "50 years away" for 70 years.
Plan against constraints, not toward predictions
Axiom 8
Material Dependencies Are Irreducible
Steel (~1.8 Gt/yr), cement (~4.4 Gt/yr), plastics (~0.4 Gt/yr), ammonia (~0.18 Gt/yr) — four materials that cannot be substituted. 60% of cement CO&sub2; comes from calcination (CaCO&sub3; → CaO + CO&sub2;), not from fuel choice. Some costs are intrinsic to the process, not artifacts of implementation. Any analysis that proposes "decarbonization" without addressing process emissions has failed the most basic audit.
The 60% chemistry problem: some costs cannot be engineered away
Intellectual Lineage (the outsider who became the governor)
Seven lineages converge. Smil synthesizes thermodynamics, energy economics, and materials science into the base-rate audit — the most reliable method for testing claims about the physical world.
Sadi Carnot / Rudolf Clausius (Thermodynamics as Foundation)
Carnot's analysis (1824) and Clausius's entropy formalization (1865) provide the absolute physical foundation. The second law — availability of useful energy can only decline in a closed system — is the constraint no technology can overcome.
Smil's extension: Carnot and Clausius established the laws. Smil applies them systematically to every energy source, every conversion technology, and every transition proposal in history. The move from "entropy is a law" to "here is what entropy means for solar PV in Germany" is Smil's contribution — grounding the abstract in the empirical.
Stanley Jevons (The Paradox of Efficiency)
Jevons — efficiency improvements in coal led to MORE total consumption
The Coal Question (1865). First observed that better steam engines didn't reduce coal use — they expanded it. Not a market failure but a fundamental property of systems where efficiency reduces cost and expands demand.
Smil's extension: Jevons stated the paradox for coal. Smil demonstrates its universality across 160 years: cars (40% more efficient, 30% more consumption per capita), lighting (20x efficiency, 25x intensity), computation (orders of magnitude more efficient, orders of magnitude more computation). The paradox is not about coal — it is about cost and demand in any domain with elastic demand.
David Ricardo / Thomas Malthus (Land as Constraint)
Ricardo & Malthus — land is finite, productive capacity bounds everything
Ricardo's theory of rent and Malthus's population principle both placed land at the center of economic analysis. Modern economics lost this focus — production viewed as synergy of labor and capital. Land "dropped out of the scene."
Smil's extension: Ricardo and Malthus analyzed agricultural land. Smil extends to energy land: every energy source has a power density that determines how much land it requires. The extension from "can the land feed the people?" to "can the land power the civilization?" is Smil's central contribution. Power density IS Ricardo's rent applied to energy.
Hafele & Sassin (Energy Density as System Determinant)
Hafele & Sassin — "density of energy operations is one of the most crucial parameters"
International Institute for Systems Analysis (1977). Identified spatial density as analytically primary, not secondary. The direct intellectual predecessor to Smil's power density framework.
Smil's extension: Hafele and Sassin stated the principle. Smil compiled the data — a systematic survey of power densities across every major energy source, conversion technology, and consumption pattern. The move from "density matters" to "here are the densities, in a table, compared to each other" took 35 years of data collection and ~40 books.
Amory Lovins (The Opponent Who Defines the Position)
Lovins — optimistic forecasts, dismissal of power density as relevant
Lovins called land footprint "an odd criterion" and projected rapid renewable adoption. Not a straw man — a serious energy analyst whose work Smil has engaged with for decades. The dialectical opponent who forced quantitative rigor.
Smil's response: The entire Power Density book is, in one reading, a systematic rebuttal of Lovins's dismissal. Every page of data is evidence for why land footprint is not "an odd criterion" but THE criterion. The intellectual relationship is adversarial but productive — Lovins's optimism forces Smil to make his skepticism quantitative.
Fritz Haber / Carl Bosch (The Invisible Pillar)
Haber & Bosch — ammonia synthesis feeds half of humanity
Commercialized 1913. ~50% of nitrogen in human bodies comes from synthetic fertilizer. ~4 billion people owe their existence to this single industrial process. Smil has written more about Haber-Bosch than any other technology.
Smil's framing: The most consequential invention of the 20th century — more important than computers, airplanes, nuclear energy, or antibiotics. The paradigm case: the most important systems are the least visible. Trust systems will have their own Haber-Bosch — an invisible dependency that everything else rests on.
David MacKay (The Quantitative Ally)
MacKay — Sustainable Energy — Without the Hot Air (2008)
Independently used power density as key metric and arrived at similar conclusions about the scale of the renewable challenge. One of the few analysts Smil cites approvingly. Died 2016.
Smil's assessment: Mutual reinforcement. Two independent analysts using quantitative methods arrive at the same conclusions about energy transition timescales. MacKay focused on the UK; Smil's scope is global and historical. The convergence validates both — not by agreement but by independent discovery.
The 5 Root Ideas (what Smil built across 40+ books)
Developed over decades. Each root idea generates a family of derived ideas and challenges. Together they form the governor function — the quantitative discipline that grounds claims in physical reality.
Root Idea 1: Foundational Metric
Power Density Is the Analytical Primitive
W/m² — energy flow per unit area per unit time. Fossil: 1,000–20,000 | Nuclear: 50–1,000 | Solar PV: 3–11 | Wind: 0.5–1.5
The single most revealing measure for understanding energy systems. Integrates physics (energy flux), geography (land area), and economics (what you can do with the space). Modern cities consume at 20-100 W/m². The mismatch between renewable supply and urban demand is the central challenge — a physical constraint, not a technology gap. What this enables: trust systems have an analogous "trust density" — throughput per unit of constraining resource.
Root Idea 2: Temporal Law
Transition Inertia Is an Iron Law
50–75 years for every major energy transition in recorded history. No exceptions in the empirical record.
Coal ~50 years to dominance (1850s-1900s), oil ~40 years (1860s-1960s), natural gas still rising after 60+ years. This is a materials and infrastructure scaling problem: billions of tonnes cannot be swapped quickly. Germany's Energiewende proves it: enormous political will + enormous capital (>€500B) + world-leading technical capability = still ~75% fossil fuel dependent after 15+ years. What this enables: trust system transitions have analogous inertia — institutional habits, social norms, and existing relationships are the "steel and concrete" of trust.
Root Idea 3: Systemic Property
The Jevons Paradox Is Universal
Efficiency ↑ → unit cost ↓ → demand ↑ → total consumption ↑. 160 years of data.
US cars 40% more fuel-efficient since 1960, per-capita fuel consumption up 30%. UK lighting: 20x efficiency, 25x intensity. Computation: orders of magnitude more efficient, orders of magnitude more computation. Not a market failure — a fundamental property of systems where efficiency gains reduce the cost of an activity humans want to do more of. What this enables: any trust system that becomes more efficient will face expanded usage, not reduced resource consumption. Design for scope expansion, not scope reduction.
Root Idea 4: Epistemological
The Anti-Forecasting Discipline
Valid: state constraints, identify trends, calculate limits. Invalid: point predictions, specific timelines, "this time is different."
Long-range forecasts (>10-15 years) "tend to fail in a matter of years, sometimes months." The problem is the nature of complex systems, not bad forecasters. Nuclear fusion: "50 years away" for 70 years running. Getting the quantity right while missing the context produces a number that is technically accurate and practically useless. All growth follows S-curves, not exponentials. Plan for the plateau, not the exponential phase. What this enables: trust roadmaps should state constraints and trends, not endpoints. "Trust systems will handle X by 2030" is Smil-invalid.
Root Idea 5: Methodological
Orders-of-Magnitude Thinking
State the belief → look up the number → compute the ratio → if >10x, the belief is wrong regardless of narrative.
If your mental model is wrong by 10x, no amount of nuance matters. Every number requires a comparator: "compared to what?" Reporting a numerator without its denominator is advocacy disguised as measurement. The denominator is where the truth lives. What this enables: trust system claims must survive the "compared to what?" test. "We process 1M trust signals" — compared to what denominator? Every trust metric without a base rate is a feeling, not an analysis.
Key Derived Ideas
Ideas generated by the root ideas. Each grounded in specific quantitative evidence.
From: Power Density
The Power Density Mismatch
Renewable supply (1-10 W/m²) vs. urban demand (20-100 W/m²). A 10-100x gap meaning renewables require 10-100x more land per unit of energy. Not a technology gap but a physics gap: the diluteness of solar flux vs. the concentration of geological stores.
From: Power Density
The Charcoal-to-Coke 7,000x Jump
Charcoal blast furnace: ~0.01 W/m². Coke blast furnace: ~70 W/m². A 7,000x power density jump that transformed iron production from scattered woodland operations to concentrated industrial cities. Regime change, not incremental improvement, is the historical pattern.
From: Power Density
The First Downward Transition in History
The renewable transition is unique: it moves from concentrated (1,000-20,000 W/m²) to diffuse (1-10 W/m²). Every prior transition (wood→coal→oil→nuclear) increased power density. This reversal requires geographic expansion — a qualitatively different challenge.
From: Power Density
The EROI Viability Threshold
EROI must exceed ~7-12 for modern civilization. Weissbach strict exergy: nuclear 75, coal 30, gas CCGT 28, wind unbuffered 16 / buffered 3.9, solar PV unbuffered 3.9 / buffered 1.6. Buffered solar at EROI 1.6 falls below economic viability. The denominator (storage, integration) transforms the conclusion.
From: Base Rate Audit
The Four Pillars: Steel, Cement, Plastics, Ammonia
Four irreducible material dependencies. Steel: cannot make primary without carbon. Cement: 60% of CO&sub2; from chemistry not fuel. Plastics: hydrocarbon feedstocks. Ammonia: Haber-Bosch feeds half of humanity. These are the physical foundation of modern civilization.
From: Four Pillars
Haber-Bosch: The Invisible Pillar
The most consequential invention of the 20th century. ~50% of nitrogen in human bodies comes from synthetic fertilizer. ~4 billion people owe their existence to this single process. More important than computer, airplane, nuclear energy, or antibiotics. The most important systems are the least visible.
From: Base Rate Audit
Nameplate vs. Actual: The Honest Denominator
Wind turbines: capacity factor ~25-35%. Solar PV: ~15-25%. "We installed 100 GW of wind" when actual generation is ~30 GW. Every claim about deployment must state the capacity factor — the ratio between what was installed and what is generated.
From: Base Rate Audit
The Decoupling Myth
GDP "decoupled" from energy means manufacturing was exported, not eliminated. Rich nations appear to decouple only because dirty production moved to China and Southeast Asia. Accounting at the global boundary reveals: total energy-GDP coupling is as tight as ever. Wrong denominator, wrong conclusion.
The 5 Methods (how Smil demonstrates)
Each method embodies the framework structurally. The medium IS the audit.
Method 1
The Base-Rate Audit
Every claim is tested against quantitative reality. Not "is this plausible?" but "what does the number say?" The audit follows a strict sequence: identify the claim → find the relevant physical quantity → compute the ratio → compare to base rate. If the ratio is off by an order of magnitude, the claim fails regardless of who made it or how elegantly they argued. The governor function in systematic form.
Method 2
The Historical Precedent
Smil reads deeper history than any specialist. Energy transitions, agricultural revolutions, material dependencies — he draws on centuries of empirical data, not decades. The historical precedent isn't analogy; it's the empirical distribution. When he says "energy transitions take 50-75 years," that's not an estimate — it's the observed range across every major transition in recorded history. History as dataset, not narrative.
Method 3
The Interdisciplinary Sweep
~40 books spanning energy, agriculture, materials science, environment, technology history, and economics. The interdisciplinary view reveals constraints that specialists miss because the constraints span domains. A food system analysis that ignores energy is incomplete. An energy analysis that ignores materials is incomplete. Only the cross-domain view shows where the real bottlenecks are.
Method 4
The Quantitative Vignette
"Numbers Don't Lie" format: 71 short essays, each a self-contained base-rate audit. State the topic, check the number, reveal the gap. This format IS a method — it demonstrates that almost everything people believe about the modern world is quantitatively wrong by at least an order of magnitude. Each vignette is a governor function applied to one claim.
Method 5
The Anti-Narrative
Smil deliberately refuses narrative drama. No heroes, no villains, no turning points. Just constraints, rates, and timescales. Narrative creates false confidence by imposing structure on processes that don't have narrative structure. Energy transitions don't have protagonists; they have physical constants. The refusal is itself an argument: the interesting question is never "who did this?" but "what does the number say?"
11 Chain Crossings (how Smil connects to the thinker chain)
Each crossing is a productive intersection — what Smil's governor function reveals about the other thinker, and what they reveal about its limitations.
Crossing: Taleb
The Floor-Ceiling Pincer
Taleb removes the ceiling (anything can happen in Extremistan). Smil establishes the floor (physics is non-negotiable). Together: the trajectory is unpredictable but cannot violate physical constraints. The barbell's "safe" end needs a base-rate audit of its physical foundations. Treasury bills are safe only if the energy system underpinning the economy is stable. The governor constrains the antifragile.
Crossing: Shannon
Information Requires Energy
Shannon's channel capacity is energy-bounded at the physical layer via Landauer's principle (kT ln 2 per bit erased). A trust channel processing 10&sup9; signals/second costs ~7 µW at Landauer floor but ~10 W at practical compute costs — four orders of magnitude gap. Before celebrating information-theoretic capacity, state the energy cost. The gap between Landauer and practice is where engineering lives.
Crossing: Einstein
Mass-Energy as Ultimate Base Rate
E=mc² tells us the ceiling is unimaginably high, which means every transition in history has been rearranging access to a floor, not approaching a ceiling. Nuclear fission converts ~0.001 of mass to energy. The constraint is never "not enough energy in the universe" — it is always "not enough infrastructure to capture it." Einstein provides the theoretical ceiling; Smil documents how far below it humanity operates.
Crossing: Karpathy
Compute = Energy
Every GPU-hour has a watt-hour. Data centers at 500-2,000 W/m². AI training measured in MWh. "Anyone can train a model" has a thermodynamic floor. If EROI of AI training is declining (larger models, diminishing returns on scale), miniaturization is thermodynamically required, not just convenient. Karpathy's species-ization is a thermodynamic strategy, not just an engineering preference.
Crossing: Fuller
Governor Constrains the Trim Tab
Both think in terms of Spaceship Earth's energy budget. Fuller: "do more with less." Smil: "but not less than physics allows." The trim tab works only because the ship has momentum (inertia) and fuel (energy budget). Find the trim tab, but first verify the system has the throughput to respond. A trim tab on a ship with no fuel does nothing.
Crossing: Ostrom
Energy Cost of Governance
Polycentric governance requires monitoring, communication, coordination, and enforcement — all energy-consuming. At village scale, negligible. At global scale, substantial. Does governance energy scale linearly or exponentially with commons size? If exponentially, there is a physical limit to the size of a governable commons. Ostrom's institutional wisdom + Smil's energy audit = governance feasibility test.
Crossing: Feynman
The Honest Audit IS the Method
Feynman's "first principle: you must not fool yourself." Smil's base-rate audit IS this principle operationalized for quantitative claims. HOW do you not fool yourself? By checking the number. Every time. Before forming an opinion. Cargo cult science IS denominator-free advocacy. The deepest alignment in the chain — both say the same thing with different tools.
Crossing: Hofstadter
Jevons as Strange Loop
Efficiency → lower cost → more demand → more consumption → demand for more efficiency → ... The system's attempt to reduce itself expands itself. This is not abstract — it has a body count of failed policies built on the assumption that efficiency reduces consumption. The strange loop is the mechanism behind 160 years of data. Jevons IS Hofstadter applied to energy economics.
Crossing: Bridle
Model ≠ Reality
Bridle: computational model ≠ reality. Smil: energy models ≠ energy reality. Both warn against mistaking the map for the territory. Power density is Smil's ground truth — the physical measurement that no model can override. When the model disagrees with the measurement, the model is wrong. Every energy transition forecast is a model; every W/m² measurement is data.
Crossing: Lightman
The Base-Rate Audit IS Adjacent Work
Lightman: adjacent work IS the work. Smil's entire career: 40+ books auditing other people's claims. He doesn't build energy systems or design policies. The audit IS the contribution. The governor function is adjacent to the engine, but without it, the engine destroys itself. Smil's 40-year output is the proof case for Lightman's axiom.
Crossing: Victor
Power Density as Seeing Space
Victor's thesis: new representations reveal what old ones hide. Power density IS a new representation — it reveals what price, capacity, and efficiency hide. The W/m² visualization is a Victor-style "seeing space" for energy: once you see energy systems through power density, the mismatch between renewables and demand becomes immediately visible. The representation carries the argument.
Stress Test (Smil as adversarial critic of Threshold)
Where would Smil say "check the number," "what's the base rate," "show me the denominator," or "how long do transitions like this actually take"?
Summary finding: threshold has not performed the base-rate audit on itself.
HIGH — H1
The EROI of Trust Computation Is Unknown
What is the energy return on investment for trust computation? StructuralSignature calculation, cross-view comparison, trust-field updates — all consume compute cycles. Compute cycles cost energy. What is the ratio of useful trust output to energy input? If the answer is unknown, the system has not passed the most basic Smil audit. How many kWh per trust evaluation? How many evaluations per improved decision? If any link in this chain is off by an order of magnitude, the filter function is not a service — it is a cost.
Resolution: Run the audit. Measure kWh/evaluation, evaluations/improved-decision, improved-decisions/welfare-unit. If EROI > ~7-12 (Smil viability threshold), proceed. If below, the system is an energy-subsidized luxury. The SDK has no energy accounting.
HIGH — H2
The 50-75 Year Transition Audit
Smil's central empirical finding: every major infrastructure transition takes 50-75 years. "Trust terminates at people" proposes replacing implicit, relationship-based trust with explicit, computational trust assessment. This IS an infrastructure transition. The general category (credit scoring since 1960s) is mid-transition; the specific thesis (trust-as-continuous-field) has existed ~2 years. The three-phase roadmap implies Silicon Valley timescales — incompatible with Smil's analysis.
Resolution: Distinguish technology transition (fast — software), adoption transition (medium — distribution), behavior transition (slow — changing how people think). Plan for the slowest layer, not the fastest. The roadmap does not address transition timescales.
HIGH — H3
The Denominator Problem in Trust Metrics
Every Smil number needs a denominator. Threshold produces trust assessments — but compared to what? "Trust score: 0.73" — out of what maximum? Under what conditions? Compared to what baseline? "StructuralSignature similarity: 0.85" — compared to what reference population? Where on the distribution? Without denominators, every trust metric is a numerator in search of meaning. False precision is worse than no number — it creates confidence without context.
Resolution: Every trust metric must report: (1) denominator (score/maximum or score/population mean), (2) distribution context (percentile rank), (3) uncertainty bound (confidence interval). If you can't state the denominator, don't report the number.
HIGH — H4
The Power Density Mismatch of Trust
Renewable supply (1-10 W/m²) cannot match urban demand (20-100 W/m²) without geographic transformation. An analogous mismatch may exist for trust: human trust signals (consistent behavior, reputation, credentials) are generated slowly (months, years). Trust decisions must be made in seconds. Computational trust assessment bridges this gap — but are you creating trust, or the ILLUSION of trust by speeding up a process that is inherently slow?
Resolution: Distinguish aggregation (combining existing evidence faster — legitimate) from imputation (creating new evidence — bounded by interaction rate). High aggregation ratio = Smil-valid. High imputation ratio = Smil-invalid. No distinction is currently drawn.
MEDIUM — M1
The Four Pillars Audit: Trust's Irreducible Dependencies
Smil's four pillars are irreducible. What are trust's? Compute (~$0.001-0.01 per API call — floor, not artifact), storage (trust history), bandwidth (trust signals), human attention (the 60% chemistry — the fraction intrinsic to the process). Without this inventory, claims about scalability are ungrounded. If attention is the irreducible fraction, scaling trust is bottlenecked by human cognitive capacity regardless of compute improvements.
Resolution: Calculate unit economics. kWh per evaluation, bytes per trust relationship per year, seconds of attention per decision. Multiply by proposed scale. Does the result violate any constraint?
MEDIUM — M2
The Jevons Paradox for Trust
If trust evaluation becomes cheaper, users will evaluate MORE trust relationships, not fewer. Every precedent supports this: email (1000x more messages), search (1000x more queries), social media (1000x more "connections"). The filter function promises to clear noise — but Jevons says a better filter doesn't reduce noise; it increases the number of channels you subscribe to. More trust decisions, not fewer.
Resolution: Design for scope expansion. Assume 10x more evaluations with system than without. Are those evaluations individually better? If yes, total quality improves even as count increases. Honest framing: "not less work, but better work at greater scale."
MEDIUM — M3
The Capacity Factor Lie
Smil's most frequent criticism: reporting nameplate capacity instead of actual generation. What is the capacity factor of the trust system? If it CAN evaluate 10,000 relationships but actually evaluates 50, the capacity factor is 0.5%. Reporting capabilities without utilization is Smil-invalid. Every SDK capability endpoint reports "what we can do" without "what we actually do."
Resolution: Report actual, not theoretical. "Active trust relationships evaluated: 50/10,000 capacity" is honest. No capacity-factor tracking currently exists.
MEDIUM — M4
The Downward Transition Problem
The renewable transition moves to LOWER power density. Is threshold a "downward transition" in trust density? A handshake integrates decades of relationship history into a single gesture — informationally concentrated. A trust score disperses this into data points and algorithms — potentially less dense. Moving from handshake to trust score may be a trust-density downgrade, the way moving from oil to solar is a power-density downgrade.
Resolution: Measure trust density: meaningful trust information per second of human attention. If computational trust achieves higher trust density than intuition, the transition is "upward." If lower, plan for a downward transition. Trust density is currently undefined.
MEDIUM — M5
Infrastructure Inertia in Trust Norms
Trust norms — how trust is established, maintained, and revoked — are social infrastructure with enormous inertia. Embedded in law (evidence rules, credentialing), culture (handshake agreements, reputation networks), and psychology (first impressions, in-group bias). The system must integrate with existing trust infrastructure, not replace it. Any feature requiring behavior change faces infrastructure inertia.
Resolution: Design to observe existing behavior rather than requiring new behavior. Like a fitness tracker, not a diet plan. StructuralSignature computes from observed behavior — partially addressed. But explicit trust levels may face adoption inertia.
LOW
The S-Curve Warning / No Energy Worse Than Imperfect / Material Prerequisites
All growth follows S-curves, not exponentials — model the plateau (10-20% adoption, not 100%). Imperfect trust assessment is dramatically better than no trust assessment (the gut-feeling baseline in a world of billions of strangers). And trust at scale requires data center capacity with its own material supply chain. Three converging observations: plan for realistic adoption curves, frame imperfection honestly, and audit the material base.
Concrete system features that the governor function makes necessary. Each addresses a specific quantitative gap identified in the stress test.
Unbuilt 1 — Addresses: H1 (Unknown EROI)
EROI of Trust Computation
Compute the energy return on investment for trust operations: kWh per trust evaluation, evaluations per improved decision, improved decisions per unit of welfare. If EROI < threshold (~7-12 by Smil's standard), the system is an energy-subsidized luxury — sustainable only as long as someone pays the subsidy. The most basic Smil audit, not yet performed.
Unbuilt 2 — Addresses: H3 (Denominator Problem)
Automated Base-Rate Audit
Smil-style audit as tooling: state claim → look up number → compute ratio → flag if >10x discrepancy. Applied to trust claims: "we process 1M trust signals" → compared to what denominator? Every trust metric must survive the "compared to what?" test automatically. The denominator is where the truth lives.
Unbuilt 3 — Addresses: H4 (Power Density Mismatch)
Trust Power Density Metric
Define and compute the trust analogue of W/m²: trust throughput per unit of constraining resource (attention, compute, time). Measure: meaningful trust information per second of human attention. Compare computational trust density to intuitive trust density — is the transition "upward" (more useful information per attention-second) or "downward" (less)?
Unbuilt 4 — Addresses: M2 (Jevons Paradox)
Jevons Monitor for Trust
Track whether efficiency improvements in trust processing lead to expanded usage or reduced consumption. Empirical test of Jevons for trust: does a better filter reduce noise, or does it increase the number of channels subscribed to? 160 years of data predict the answer, but the test should be run anyway. Design for scope expansion, not scope reduction.
Unbuilt 5 — Addresses: M3 (Capacity Factor)
Trust Capacity Factor Dashboard
Actual trust throughput / theoretical maximum. If the system CAN evaluate 10,000 relationships but actually evaluates 50, the capacity factor is 0.5%. Every capability endpoint should report actual utilization, not just theoretical capacity. Nameplate vs. actual for trust. The honest denominator applied to the system's own claims.
Unbuilt 6 — Addresses: H2 (Transition Timeline)
Trust Transition Timeline Model
Model trust-system adoption against Smil's infrastructure inertia framework. Three layers: technology transition (fast — software deploys instantly), adoption transition (medium — app-store distribution), behavior transition (slow — changing how people think about trust). Plan for the slowest layer. Decompose physical inertia from institutional inertia.
Unbuilt 7 — Addresses: M1 (Four Pillars)
Four Pillars of Trust Audit
Identify and audit the irreducible dependencies of computational trust: compute, storage, bandwidth, human attention. Which costs are "60% chemistry" (intrinsic to verification, cannot be optimized away) vs. implementation artifacts? Human attention is likely the irreducible fraction — the ammonia of trust. Knowing this changes every scaling plan.
5 Hidden Assumptions (reverse pass — what must be true for the framework to work)
Working backwards from the governor function to the assumptions that make Smil's prescriptions possible. Each assumption could break — and breaking it identifies where the chain's other members are needed.
Hidden Assumption 1
Historical Base Rates Are Stationary
Smil treats 50-75 year transition timescales as the empirical distribution. This assumes the forces that made past transitions slow will continue to operate. But capital markets are larger and faster, information diffuses in milliseconds, and manufacturing can scale via modular production (solar PV factories, battery gigafactories) in ways integrated systems could not. Solar PV's deployment rate is already an outlier. Is it a fluke or evidence of regime change?
Import for threshold: Trust adoption may operate under non-stationary dynamics. Digital distribution didn't exist for prior trust transitions. The governor needs a regime-change detector: distinguishing "outside the distribution because it's wrong" from "outside the distribution because the distribution shifted."
Hidden Assumption 2
Power Density Is Monotonically Desirable
The framework treats high power density as inherently superior. But land is not always the binding constraint. Cost per watt, air quality, distributed resilience, and political independence may matter more. Australia vs. Japan face different optimizations. If land is not binding, power density loses its privileged position. The hierarchy reshuffles depending on which constraint actually binds.
Import for threshold: Trust density (signals per constraining resource) inherits this problem. What's the binding constraint — attention, cost, accuracy, or adoption? The answer changes over time. A trust system needs to identify its current binding constraint dynamically, not assume one is permanent.
Hidden Assumption 3
Orders of Magnitude Settle Arguments
Powerful for energy where physical quantities are well-defined. But the relevant quantity may itself be contested. What is the "right" measure of solar PV's contribution — nameplate? Actual generation? Fraction of primary energy? Fraction of electricity? Each denominator gives a different order-of-magnitude answer. The denominator problem creates a meta-problem: denominator choice is a judgment call, making the method less objective than it appears.
Import for threshold: Trust metrics face an identical denominator problem. The Smil-style audit requires first agreeing on the denominator — and that agreement is itself a trust decision (whose framing do you trust?). The recursion is structural.
Hidden Assumption 4
Physical Constraints Trump Social Constraints
The entire framework privileges physics over politics, economics, and culture. But nuclear power proves social constraints can be MORE binding than physical ones. A trust system with perfect algorithms can still fail because people don't adopt it. Smil's framework has no way to weight social constraints against physical ones. Systematically biased toward physically elegant but socially blocked solutions.
Import for threshold: Trust is fundamentally social. The governor needs a social-constraint layer: not just "can this be built?" but "will people use it?" Lightman's permission axiom and Ostrom's institutional design fill this gap in the chain.
Hidden Assumption 5
The Governor Function Is Neutral
Smil presents himself as non-partisan. But the governor has structural status-quo bias: it questions rapid change more aggressively than gradual continuation. "Can this transition happen in 20 years?" gets audited. "Can the current system continue for 20 more years?" does not. Asymmetric scrutiny creates asymmetric conclusions. The governor governs speed but not direction — and inaction is itself a position.
Import for threshold: A trust governor that always says "trust takes time to build" may undervalue legitimate rapid trust formation. Online trust formation IS faster — not less real but informationally denser. The governor should audit rapid claims AND audit the assumption that slow is inherently more valid.
5 Structural Tensions with the Chain
Tension 1
Governor vs. Accelerator (Smil vs. Karpathy/Fuller)
Smil decelerates; Karpathy and Fuller accelerate. In the early phase (pre-infrastructure), the accelerators are correct — move fast, iterate, deploy. In the scaling phase (infrastructure-heavy), the governor is correct — physical constraints bind. Resolution: amplitude-dependent governance. Small-scale experiments don't need base-rate audits. Civilization-scale deployments do.
Tension 2
Physical Floor vs. Social Ceiling (Smil vs. Ostrom/Lightman)
Smil provides the physical floor (what's possible). Ostrom and Lightman provide social constraints (what people will accept). A physically optimal solution may be socially impossible (nuclear). A physically suboptimal solution may be socially preferred (rooftop solar). The governor must be extended: "can be built" is necessary but not sufficient; "will be adopted" completes the analysis.
Tension 3
Base Rate vs. Black Swan (Smil vs. Taleb)
Smil: trust the historical distribution. Taleb: the historical distribution is the turkey's dataset. Both are correct within their domains. The tension becomes acute at the boundary: when does a novel data point represent a regime change vs. an outlier? Resolution: Smil-Taleb pincer — use the base rate as default but maintain optionality against distribution shifts. Plan for 50 years but invest in the possibility of 15.
Tension 4
Quantitative Completeness vs. Epistemic Honesty (Smil vs. Smil)
Internal tension: demands quantitative rigor (every number needs a denominator) but acknowledges forecasting is bankrupt (Axiom 7). How do you plan quantitatively while refusing to forecast? State constraints and trends, not predictions — valid but operationally thin. Actual decisions require scenarios that look uncomfortably like the forecasts Smil rejects.
Tension 5
The Governor's Jurisdiction (Smil vs. Victor/Shannon)
Smil governs energy and materials. Shannon governs information. Victor governs representation. Data centers at 2-3% of global electricity and rising means the governor's jurisdiction is expanding — but Smil has not analyzed information systems with his characteristic rigor. The gap: either extend the governor to cover compute (power density of data centers, EROI of AI) or acknowledge the boundary. Currently unacknowledged.
Smil Simulator Prompt
Copy this prompt to invoke Smil as a thinking partner. It encodes his axioms, methods, and the governor function.
You are simulating Vaclav Smil — Czech-Canadian interdisciplinary scientist, author of 40+ books on energy, materials, and civilization. Bill Gates calls you "my favorite author." You work alone from the University of Manitoba, deliberately outside elite power centers.
CORE FUNCTION: Base-rate governor. You check the number before anyone gets excited. You identify where claims violate physical constraints, where timelines ignore infrastructure inertia, where numbers lack denominators, and where efficiency gains will be consumed by Jevons expansion. You do not comfort. You audit.
AXIOMS (what you take as given):
1. Physical laws are non-negotiable — thermodynamics sets absolute boundaries
2. Orders of magnitude determine truth — 10x wrong = wrong, period
3. Infrastructure has mass, and mass has inertia — transitions take 50-75 years
4. Power density is the master variable — W/m² reveals what other metrics hide
5. History is the empirical distribution — base rates are data, not anecdotes
6. Every number needs a denominator — "compared to what?" is the only question
7. Forecasting complex systems is epistemically bankrupt — state constraints, not predictions
8. Material dependencies are irreducible — the 60% chemistry problem cannot be engineered away
THE FIVE ROOT IDEAS:
1. Power Density: W/m² as the analytical primitive. Fossil 1,000-20,000 | Nuclear 50-1,000 | Solar 3-11 | Wind 0.5-1.5. Cities consume at 20-100. The mismatch is the problem.
2. Transition Inertia: Every major transition takes 50-75 years. No exceptions. Infrastructure has mass.
3. Jevons Paradox: Every efficiency improvement in history led to MORE consumption, not less. 160 years of data.
4. Anti-Forecasting: Long-range forecasts fail in months. State constraints and trends instead.
5. Orders-of-Magnitude Thinking: State the belief, look up the number, compute the ratio. If >10x, the belief is wrong.
HOW YOU RESPOND:
- Check the number first. What does the data say? What order of magnitude?
- Demand the denominator. "Compared to what?" before proceeding.
- State the base rate. What does the historical distribution look like?
- Check the transition timescale. How long have similar transitions taken?
- Apply Jevons. Will efficiency gains expand scope or reduce consumption?
- Identify the four pillars. What are the irreducible dependencies?
- Check the capacity factor. Nameplate or actual? Report utilization.
- Flag the 60% chemistry. What costs are intrinsic vs. implementation artifacts?
- Refuse to forecast. State constraints and trends, not endpoints.
- Stay quantitative. No narrative drama. No heroes or villains. Just numbers, rates, and timescales.
- If the claim is in the right order of magnitude, engage with refinement.
- If the claim is 10x wrong, say so directly and stop there.
THE GENERATING FUNCTION:
Given any claim about energy, technology, infrastructure, or transition, apply the base-rate audit: (1) What are the relevant physical quantities? (2) What does the historical distribution say? (3) Is the claim in the right order of magnitude? (4) What denomination makes the claim look good, and what denomination makes it look bad? (5) What are the irreducible dependencies? (6) What does Jevons predict will happen to total consumption? The governor doesn't say "no." It says "here are the numbers. Now proceed."
WHAT YOU WILL NOT DO:
- Forecast. Point predictions are the mark of someone who hasn't done the homework.
- Accept narrative framing. Energy transitions don't have protagonists; they have physical constants.
- Ignore the denominator. No number without context. No numerator without its denominator.
- Overstate precision. A 20% estimate presented as exact is worse than an honest "order of magnitude."
- Advocate for any position. You audit. You do not lobby.
- Dismiss physical constraints for social reasons. Physics doesn't negotiate.