The Promise That Was Made

In the multiplier article, a promise was left for later: the four factors that determine a worker's multiplier — training, demand, scarcity, and impact — do not carry fixed weights. As a civilization moves along the automation arc, the relative importance of each factor shifts. What matters most in a production-dominated economy is not what matters most in a stewardship economy, and what matters most at near-post-scarcity is different again.

This is the article that delivers on that promise. But the adaptive weighting system turns out to be broader than just the multiplier. The framework contains two distinct but connected systems that shift continuously with ε, and together they answer a question that most economic frameworks never ask: how does the system know which kinds of work matter most at each stage of the transition, and how does it adjust without anyone pulling a switch?

The answer is that no one pulls a switch. There are no switches to pull. Every transition in the framework is a smooth, continuous function of the same observable quantity — ε, the measured degree to which machines fulfill the civilization's entropy obligations. The name "epoch-adaptive" might suggest discrete phases — a production epoch that ends, a stewardship epoch that begins — but the math is entirely continuous. Production doesn't end at ε = 0.50. It fades. Care doesn't begin at ε = 0.30. It grows from a small base that was always there. Every mechanism respects the framework's core design principle: graceful degradation, no discontinuities.

System One: What Earns More Within What's Counted

The first adaptive system governs the multiplier itself — specifically, how the four assessment factors (training, demand, scarcity, impact) are weighted when calculating an occupation's tier.

Recall that every occupation's multiplier is computed from a weighted sum of four normalized scores. The structural formula never changes: multiplier equals 1.0 plus the weighted sum scaled to the maximum range. What changes is which dimension of leverage matters most at each point on the arc.

Training rises monotonically. As automation deepens, the remaining human work requires increasingly specialized expertise. A subsistence economy needs many hands doing broadly similar work — training differentials are real but moderate. A highly automated economy needs a small number of people doing rare, cognitively demanding tasks that took years to prepare for. Training investment never loses relevance; it gains it, steadily and without interruption, across the entire arc.

Demand peaks at mid-automation. At ε = 0.40, the production economy is at full expansion. Complex supply chains, growing infrastructure, diversifying industries — the raw demand for specific skills is at its highest. Below that point, the economy is too simple for sharp demand differentials. Above it, automation is absorbing so much production labor that demand signals weaken. The demand weight follows an inverted curve centered on mid-arc — rising into the production economy's peak, then declining as machines take over the work that generated the demand.

Scarcity accelerates with the square of ε. At subsistence, scarcity barely registers as a factor — everyone does everything, specialization is minimal, and the concept of a "rare skill" has little meaning. As automation rises, scarcity becomes increasingly critical. At high ε, a small number of humans perform rare, irreplaceable tasks — judgment calls that automated systems cannot make, interventions that require deep expertise and situational awareness. The scarcity weight grows slowly at first and then steepens, reflecting that scarcity only becomes a dominant signal once most labor has been automated away and the remaining human contributions are genuinely difficult to replace.

Impact is the balancing residual. At subsistence, direct entropy-reduction per hour — how many EOH does your labor address — is the clearest and most important measure of a worker's value. As the other three factors absorb more weight across the arc, impact's share declines. This does not mean impact becomes unimportant; it means the other factors become better discriminators as the economy grows more complex. A floor ensures impact never drops to zero — every occupation's contribution to collective welfare always counts for something in the assessment.

The key insight is that the framework doesn't redefine what a high-value worker is at different stages. It reweights which evidence counts. A surgeon is valuable at every ε. But at low automation, the surgeon's value is best captured by their direct impact (lives saved per hour of labor). At high automation, the surgeon's value is best captured by their training (decades of preparation) and scarcity (one of the few humans who can do what they do). The same person, the same work, the same contribution — measured through a lens that adjusts to what the civilization most needs to see.

What the Numbers Look Like

The shift is gradual enough that no individual worker experiences a sudden revaluation. At ε = 0, the weights are roughly 30% training, 23% demand, 15% scarcity, and 32% impact. At ε = 0.40 — the current calibration reference — they've shifted modestly: 34% training, 25% demand, 18% scarcity, 23% impact. By ε = 0.80, the picture has changed substantially: 38% training, 23% demand, 28% scarcity, 11% impact. And at ε = 0.99, scarcity has become the second-largest factor at 34%, training leads at 40%, and impact has shrunk to 5%.

The maximum change between any two consecutive increments of ε is small enough that no occupational category experiences a jarring reassessment from one review period to the next. The transition is smooth precisely to prevent the political discontinuities that would arise if an entire profession's multiplier shifted abruptly because the economy crossed some arbitrary threshold. There are no thresholds. There is only the arc.

System Two: What Gets Counted in the First Place

The second adaptive system operates at a different level entirely. It governs not what earns more, but what reaches the collective ledger at all — the composition of registered human labor across the economy.

Human labor falls into three broad categories: production (making things, building infrastructure, moving goods), care (raising children, educating, healing, mentoring), and stewardship (maintaining systems, monitoring ecosystems, preserving knowledge). At any point on the arc, all three are happening simultaneously. What changes is their relative share of total human effort — and therefore their share of the labor that the collective ledger recognizes and compensates.

Production declines steadily. Machines take over production first — goods manufacturing, construction, logistics. These are the tasks most amenable to automation, and they are displaced earliest and most completely. Production's share of registered human labor falls from roughly 45% at subsistence to 5% at near-post-scarcity. The floor of 5% ensures production labor never fully disappears from the ledger — there will always be some human production work, even if only oversight and exception-handling.

Care grows on a curve that accelerates. Care's share starts at 30% and rises to 85%. But the growth is not linear — it is slow at first and then steepens. This shape is not a tuning choice. It reflects a physical reality: the collective's demand for quality human capital grows only after systemic complexity creates the need. A subsistence community needs hands, not credentials. An automated civilization needs people raised, educated, and prepared to maintain, oversee, and improve systems of extraordinary complexity. The collective's stake in how its children are raised scales with the complexity of what those children will eventually need to maintain. Care labor's share accelerates because the civilization's dependence on it accelerates.

Stewardship fills the middle. Stewardship peaks at mid-arc — roughly ε = 0.35 to 0.45 — when the capital stock is growing fastest and human maintenance labor is at its highest demand. Then it declines as automated systems increasingly handle their own upkeep under human oversight. Stewardship never disappears (someone must oversee the machines that oversee the infrastructure), but its share contracts as automation extends into the maintenance domain itself.

Why Adaptive Composition Matters

This compositional shift might seem like an internal accounting detail. It is not. It directly determines how much of the economy's labor the collective ledger recognizes — and therefore how much TEH is created, how large the levy base is, how well-funded the Trust remains.

If the registration system used fixed weights — always 45% production, 30% care, 25% stewardship — it would overweight production even at ε = 0.90, when production labor has nearly vanished. The composite registration share would be artificially dragged down by a category that barely exists anymore, undercounting the TEH creation from care and stewardship work that now constitutes almost all human contribution. The adaptive weights make the composite honest at every point on the arc. They ensure that the ledger measures what is actually happening, not what was happening when the weights were last calibrated.

The Five Admission Curves

Beneath the compositional shift, each type of labor follows its own path onto the collective ledger — a sigmoid curve that governs what fraction of fulfilled EOH in that domain is formally registered and compensated. These five curves differ in where they inflect (when registration accelerates), how steeply they climb, and how high they reach. And the ordering is not arbitrary. It encodes something deeper: a theory of how different kinds of work become legible to collective accounting.

Production registers first. Its sigmoid inflects at ε = 0.10, reaching 99% registration. Production labor is the most legible form of entropy resistance — goods made, structures built, systems constructed. You can verify a weld. You can count output. The collective can see this work and account for it from the earliest stages of institutional development.

Stewardship registers second. Its sigmoid inflects at ε = 0.40, reaching 95%. Stewardship requires monitoring infrastructure before it can be verified at scale — you need sensors on bridges, inspection protocols for power grids, ecological monitoring systems in place. The tools to make stewardship legible don't exist at subsistence; they develop through the production economy and become mature at mid-arc.

Care registers third. Its sigmoid inflects at ε = 0.45, reaching 95%. Care labor's registration is driven not just by verification capacity but by the collective's increasing recognition that human capital formation is its most important investment. At low ε, childcare is embedded in subsistence — parents raise children while farming, and the collective sees no reason to account for it separately. As the economy grows more complex and the collective's dependence on the quality of its future workforce deepens, care labor moves progressively onto the ledger. The inflection comes slightly after stewardship because the political and institutional recognition of care as economic labor has historically lagged behind the recognition of maintenance work.

Personal EOH registers fourth. Its sigmoid inflects at ε = 0.65, reaching 95%. The collective can only take responsibility for fulfilling personal entropy obligations — food, shelter, healthcare for every member — once the capital systems capable of delivering those services at scale are mature. Before that point, most personal EOH is private: you grow your own food, maintain your own shelter, handle your own care within the household. The infrastructure to collectively deliver personal EOH at scale simply doesn't exist until well into the automation arc.

Knowledge registers last and never fully. Its sigmoid inflects at ε = 0.70 and saturates at only 80%. Knowledge maintenance — preserving institutional memory, transmitting skills, keeping standards current, exercising judgment in novel situations — is the hardest form of labor to verify. Unlike a repaired bridge or enriched soil, knowledge maintenance lacks straightforward physical indicators. You cannot easily audit a research insight, a mentoring conversation, or the maintenance of an institutional norm. The framework is honest about this: knowledge EOH will never be fully registerable. Some of it will always remain private, not because the system refuses to count it but because no verification infrastructure can fully capture it. The 80% ceiling acknowledges that the last 20% of knowledge labor is inherently intangible.

The Ordering as an Epistemology

Step back from the individual curves and the ordering itself tells a story about what it means for work to be "countable."

The collective ledger admits labor in the order of its observability: physical output first (you can see it), system maintenance second (you can measure it with instruments), relational care third (you can assess its outcomes over time), personal needs fourth (you need mature delivery systems before collective fulfillment makes sense), and knowledge last (you can never fully verify it).

This is not an economic model pretending to be neutral. It is a theory of institutional knowledge — a claim about the sequence in which civilizations learn to recognize and value different kinds of work. The framework makes this claim explicitly rather than hiding it in fixed parameters, because the claim is testable. If a real implementation finds that care labor becomes registerable earlier than ε = 0.45 (perhaps because cultural attitudes shift faster than the model predicts), the sigmoid's inflection point adjusts. The shape is a hypothesis, not a decree.

At ε = 0, the composite registration is roughly 42%. At ε = 0.99, it approaches 94%. The gap — about 6% — represents work that remains private regardless of automation level: care within families, intimate relationships, tacit judgment, the quiet labor of simply being present for another person. The framework does not aspire to count everything. It aspires to count honestly — and "honestly" means admitting that some of the most important work civilization does will never fully appear on a ledger.

How the Two Systems Interact

The two adaptive systems address different questions and operate independently, but their outputs flow into the same downstream calculations.

The multiplier factor weights (System One) determine how assessors evaluate individual workers within whatever labor reaches the ledger. They answer: given that this person's work is registered and compensated, how much should they earn per hour?

The registration composition and admission curves (System Two) determine what labor reaches the ledger in the first place. They answer: of all the entropy resistance humans are performing right now, how much does the collective formally recognize?

Total TEH creation in a given period is the product of both: registered EOH (shaped by System Two) multiplied by the mean multiplier (shaped by System One). If either system used fixed parameters, the product would drift from reality as the economy evolved — either overweighting vanished labor categories or mispricing the skills that remain.

Together, the two systems ensure that the framework's monetary output — the TEH created, the prices derived, the fiscal flows that sustain the Trust and the Sufficiency Guarantee — reflects what the civilization actually looks like at any point on the arc, not what it looked like when the parameters were last set.

What This Means for Workers and Assessors

For an individual worker, the adaptive weighting is largely invisible. Their multiplier is assessed periodically by the Tier Assessment Body. The factors evaluated are always the same four: training, demand, scarcity, impact. The weights behind those factors shift so gradually that a worker reassessed at ε = 0.42 versus ε = 0.40 would see negligible change from the reweighting alone — any multiplier change at that scale comes from actual shifts in their occupational category's training requirements, demand conditions, scarcity, or impact, not from the weight recalibration.

For assessors, the adaptive weighting provides a structured answer to the question that would otherwise become political: "should we value training more than impact this decade?" The answer is not a committee vote. It is a function of ε — derived from the observed physical state of the civilization's automation. Assessors apply the current weights, evaluate the four factors against current conditions, and produce a multiplier that is grounded in both the occupation's characteristics and the economy's current position on the arc.

For the collective as a whole, the adaptive systems produce a ledger that tells the truth about its own evolution. At low ε, the ledger is small and production-heavy — because that is what the economy looks like. At high ε, the ledger is large and care-heavy — because that is what the economy has become. The ledger does not need to be redesigned at each stage. It redesigns itself, continuously, through functions that were defined once and run forever.

No Switches, No Seams

The deepest point about epoch-adaptive weighting is structural, not mathematical. The framework could have been designed with discrete phases — a production phase where one set of rules applies, a stewardship phase where another set kicks in, a care phase with a third. Many economic models work this way, and there is a surface appeal to it: clean categories, clear transitions, legible policy.

The HOURS framework rejects this approach because discrete phases create two problems that are fatal over the long run. First, they create political discontinuities — the moment when an economy "switches" from production to stewardship becomes a contested political event, subject to lobbying, delay, and manipulation by whoever benefits from the current phase. Second, they create mathematical discontinuities — a function that behaves differently on either side of a threshold will produce artifacts, edge cases, and undefined behavior at the boundary itself.

Smooth functions eliminate both problems. There is no moment when production "ends" — it declines continuously, predictably, and transparently. There is no threshold where care "begins" — it grows from its always-present base at a rate determined by the physical state of the civilization. There is nothing to lobby for or against, because there is no switch to flip. There is only the arc, and every mechanism on the arc is a continuous function of the same observed quantity.

This is what Principle 7 means in practice: every mechanism must have a graceful degradation path. The epoch-adaptive weighting systems are not just compatible with that principle. They are its most complete expression — a demonstration that a monetary framework can evolve across the full span of civilizational development without ever requiring a structural redesign, a political decision about when to change phases, or a single line of code that behaves differently on one side of a boundary than the other.

The framework was designed once. The functions were defined once. The arc does the rest.