An intaglio futures study · Issued September 2026 · Denomination: one human share
The labs are starting to earn like nations. This study asks the oldest question in political economy with new numbers — when the machines learn to earn, what does one person hold? — and follows five people's paydays into four possible 2042s.
≈ 0.015% — a lab's quarterly revenue is already roughly a fifteenth of a percent of US annual GDP
UV · PLATE SP-002042 · WATERMARK: ONE HUMAN SHARE · FIBERS: 22 SOURCED, NONE SYNTHETIC · SECURITY THREAD: THE EVIDENCE LEDGER
How to read this study
This is a futures study, not a forecast. Every claim about the future below is a constructed illustration; every claim about the present carries a link you can pull. Four borrowed instruments keep the two apart.
The numbered red serials — [S1], [S2]… — are real developments, each traceable to a source found in this session. Futures practice calls this horizon scanning: a scoping question first, a source behind every scan, breadth checked by count. When a future below seems fantastical, read its serial first. The strangest things in this study are the true ones.
"Economic share of individuals" is vague until it is a ledger. Ours has four lines, per person, per year: wages (money from employment), capital (dividends, rents, appreciation on what you own), transfers (state payments: pensions, UBI, wage insurance), and commons income (dividends from shared wealth: sovereign funds, model royalties, endowments). Every payday in this study is such a stack.
Futures practice has a rule: build one future and you have built an advertisement. So the study mints four 2042s in Jim Dator's four generic futures — growth, collapse, discipline, transformation — each with its own logic of distribution and one person's payday ledger printed from inside it. No future is the best case, the worst case, or the likely one.
Because the question is "who gets the money", the study holds an inquest on universal basic income — what sixty years of pilots and one true national case actually show — and then drafts five original instruments beyond UBI: a compute royalty, a birthright stake, a caretaker's wage, an automation insurance scheme, and a grid-for-equity trust. Each is stress-tested and ethically audited.
What could change between now and 2042 in the share of the economy that flows to individuals — as wages, transfers, capital and commons income — while AI companies capture a growing share of value, and which distributions of it would leave society in better shape?
It refuses to predict, to recommend one party's platform, or to print a single future. Where a number is constructed — every payday ledger is — it is labelled as illustration and anchored to the sourced signals beside it.
Method 01 · fdbx-horizon-scanning, step 4 · the present condition · UK Government Office for Science
Before forecasting a payday, read the current one. The build-out is historic — but almost all of it is flowing to capital and almost none of it, yet, to the person holding this note.
receipt · the machine age so far · fiscal 2025–26 · items sourced · keep this stub
| Item | Amount | Serial |
|---|---|---|
| Anthropic annualized revenue run-rate, end of July 2026 — up from ~$30B in April, with company projections of $100–120B by year end | >$65B / yr | S1 |
| OpenAI annualized revenue, February 2026 (Anthropic overtook it in Q2) | ≈$25B / yr | S2 |
| Nvidia quarterly revenue, FQ2 2027 — mostly data-center | $96B / qtr | S3 |
| Big Tech AI build-out, 2026 — about 2.2% of US GDP, four times the 2023 share | ≈$630B / yr | S4 |
| Cumulative AI capex projected 2026–2031 (Goldman Sachs baseline) | ≈$7.6T | S5 |
| Billionaire wealth, January 2026 — up 16% in one year, three times the five-year average | $18.3T | S6 |
| Employment gap, US workers aged 22–25 in the most AI-exposed occupations, vs. where it would be otherwise (Stanford, Aug 2026 update) | −19% | S7 |
| Entry-level white-collar jobs the Anthropic CEO says AI could eliminate within 1–5 years, with unemployment at 10–20% | "~50%" | S8 |
| US GDP growth officially missed because statistics don't yet see most of Nvidia's domestic income (Epoch AI) | +0.3pp | S9 |
Every amount above is a sourced signal — see the shelf in chapter 04. The point of the receipt is not that the labs are rich. It is that the income side of the machine age is arriving faster than the distribution side.
Ask any household ledger today what share of its income is commons income — dividends from wealth held in common, like a sovereign fund yield or a resource royalty. Outside one US state and a handful of petro-states, the answer rounds to zero. That line is the whole study. The machine age could print it for everyone; or print it for nobody.
Five people, five countries, five kinds of exposure to AI. They are fictional composites — no single person is depicted — but each sits where a real occupational group sits in the data. Their 2026 stacks are constructed illustrations, shaped to each country's actual income structure; the 2042 paydays in chapter 06 are built the same way. C = capital line · T = transfers line · K = commons line.
Lagos, Nigeria · support-team lead at a fintech, the layer global AI agents reach first
2026 · ₦460k/mo wage · tiny brokerage app position · family remittances flow out, not in · commons ≈ nil
Her job is precisely the work [S8] warns about — and the global support market is the first place agents prove themselves. Watch her in the collapse future.
Youngstown, Ohio · union forklift operator, distribution centre; robots arrive on his floor, not in his cab
2026 · $71k wage incl. overtime · 401(k) that wobbles · heating credits in winter · one Alaska-size commons cheque would be news
A middle-age worker in a physical-logistics job: the canary study's displacement [S7] passes him by — for now — but his children are the 22–25 cohort.
Bengaluru, India · junior developer at a services firm; the exact cohort the canary study tracks
2026 · ₹78k/mo · SIP mutual-fund ₹6k/mo · no state support to speak of · the ladder's bottom rung is vanishing [S7]
India ships the world's code; agents now write much of the junior kind. She is the person the phrase "entry-level" was invented for.
Montevideo, Uruguay · history teacher, four years from pension; his classroom AI never enters
2026 · teacher salary + family orchard rented out · pension accrual counted as future transfer · steady, exposed to tax politics, not to AI
He is the control case: the person whose share depends least on the labs and most on whether the state keeps its promises. Follow him in the discipline future.
Rotterdam, Netherlands · freelance illustrator and brand designer; her clients now generate first drafts by model
2026 · clients down to 9 from 15 · loft mortgaged, appreciating · Dutch childcare & arts allowances · already living a portfolio life
Creative freelancers are the natural constituency for post-wage income: she already earns from four lines instead of one. Follow her in the transformation future.
| Claim used as input | Type | Basis |
|---|---|---|
| Anthropic run-rate >$65B (Jul 2026); company projections $100–120B | sourced | Axios; CNBC/TechCrunch coverage of the run-rate |
| OpenAI ≈$25B annualized (Feb 2026); Anthropic overtook in Q2 | sourced | reported figures in the same run-rate coverage |
| Nvidia FQ2 2027 revenue $96.22B | sourced | Rex Shares earnings summary |
| 2026 Big Tech AI spend ≈$630B ≈ 2.2% of US GDP, 4× the 2023 share | sourced | Reuters Breakingviews, Mar 26 2026 |
| $7.6T cumulative AI capex 2026–31 (baseline) | sourced | Goldman Sachs |
| Billionaire wealth $18.3T, +16% in 2025 | sourced | Oxfam America; Oxfam International (Jan 2026 Davos report) |
| US workers 22–25 in most AI-exposed occupations ≈19% below counterfactual (Aug 2026 update; was 13% in Aug 2025) | sourced | Stanford Digital Economy Lab |
| Amodei: AI could eliminate ~50% of entry-level white-collar jobs in 1–5 yrs, unemployment 10–20% | sourced | Axios, May 28 2025; doubled down Feb 2026 (Forbes) |
| US GDP growth underestimated ≈0.3pp because Nvidia's domestic income is largely unseen by official statistics | sourced | Epoch AI |
| All five people, their jobs, their stacks and every number in their ledgers | assumption | Fictional composites; stacks are constructed illustrations shaped to each country's income structure, not survey data. |
Assumptions surfaced: that the support, junior-code and freelance layers are the first to feel agent displacement; that wage income still dominates every household above. Open decisions: which future's distribution logic wins. Handoff: the receipt's concentration theme goes to causal-layered analysis next; the five stacks become the templates the four futures re-compose.
Method 02 · fdbx-causal-layered-analysis · Inayatullah
The debate about AI wealth usually arrives as a litany — "labs are getting rich, workers are getting squeezed, we need a dividend." Causal layered analysis takes that headline apart, layer by layer, because a distribution designed at the wrong depth redistributes nothing.
The issue, in the words it usually arrives in: "AI companies are capturing enormous value; ordinary people's wages fund the build-out and the models automate their work. Shouldn't there be a cut for everyone — a UBI, a data dividend, something? Or is that naive?" Notice what is already embedded: the success metric (a transfer, arriving), the solver (whoever is owed), and the frame (the pie is baked; the only question is slices). Each layer below is deeper than the last — and each names a different problem.
| Layer | Problem as it appears | Solution this framing makes obvious | Who is expected to act | Where this framing lives |
|---|---|---|---|---|
| Litany | Lab revenues and billionaire wealth are compounding fast [S1, S6] while the first cohort displacement numbers arrive [S7, S8]. "The gap is widening; give people cash." | UBI pilots, one-off "AI dividends", data cheques, viral proposals. The tool is a payment; the tempo is a press release. | Governments, mostly; foundations and labs as benefactors. | Headlines, op-eds, conference panels, the Taiwans of the world [S10]. |
| Social causes | The AI build-out is financed by everyone's electricity prices and pension funds, but its equity sits on a few cap tables [S4, S5]. Labour's bargaining position weakens at exactly the entry level where careers compound [S7]. Fiscal systems tax work heavily and compute lightly. | Change the plumbing, not the payout: royalty bases, ownership rules, bargaining law, tax treatment of capital vs. labour, entries in the national accounts [S9]. | Legislatures, unions, central statistical offices, standard-setters. | Tax codes, cap tables, collective agreements, GDP manuals. |
| Discourse / worldview | Three incompatible stories decide what "fair" even means: work = worth (income must be earned by the sweat of it), abundance = release (the point of the machines is to free people from earning), techno-feudalism (rents to the platform-lords, whatever we call it). Each defines the problem differently, so each "wins" a different argument. | Name which worldview is steering by default. Design instruments that work under more than one story, and say honestly which story each favours. | Writers, unions, religious bodies, schools — the story-tellers; then everyone who repeats them. | Op-eds and manifestos; dinner tables; the word "deserving". |
| Myth / metaphor | The deep story: the paycheck as the social contract — you are what you earn, worth is wages, and any income without work is charity, indulgence, or theft. This metaphor is so old it feels like physics; the Alaska dividend and the Taiwan payment are read as quirks of surplus, not as competing contracts. | A new story: the share as the social contract — every person holds a stake in the common machine age the way every Alaskan holds the fund: not alms, but dividends on assets already owned in common. | Founders of funds, constitutional drafters, grandparents, the people who name things. | Banknotes and founding documents; Paine's Agrarian Justice; the phrase "taxpayer money". |
The oldest story on the ledger: income must be earned, and automation that removes the earning removes the worth. Problem: idleness. Future it points to: retraining, "dignity of work" schemes, suspicion of UBI. Loudest voices: employers, work ethic traditionalists, parts of the labour movement itself.
The machines will do the work; the point of that is freedom. Income must decouple from employment or the dividend of automation is nobody's. Problem: distribution lag — abundance arrives on capital's timetable, not labour's. Future: dividends, shorter weeks, the commons line growing. Voices: UBI advocates, some lab founders, poets of the leisure society.
The platforms and labs are not firms in a market; they are lords of a cloud we all farm, extracting rent from data, compute and dependency. Problem: ownership, not income — transfers to serfs do not end the manor. Future: antitrust, data commons, public equity stakes. Voices: the techno-feudalism school of political economy, data-union organizers, sovereignty-minded states.
"You are what you earn." The paycheck is the contract; everything else is a benefit of the employer's generosity or the state's pity. The machine age is read as an employer that fires politely.
"You hold what we all built." The infrastructure of intelligence — trained on our text, running on our grid, funded by our deposits and our electricity — is common capital. Dividends are not charity; they are the yield on an inheritance.
| Version | The question |
|---|---|
| Original | "AI labs are making more money — how do we give individuals a share (UBI, dividends, handouts) so the gap doesn't explode?" (metric: a transfer that arrives) |
| Reframed | "Individuals already co-own the inputs of the machine age — their text, their grid, their deposits, their attention. Which instruments convert that latent co-ownership into standing income lines — capital and commons, not just transfers — so that the machine age strengthens the household ledger instead of merely taxing it?" (metric: composition and resilience of the personal income stack) |
Nothing in the reframed question abandons cash transfers. It demotes them from the answer to one line of four — and asks what the other three lines could be.
A wealthy state funds a modest AI dividend; the cheque lands, helps, and quietly becomes a line item contested every budget season. The concentration underneath is untouched. Distribution as press release.
Compute joins the tax base; statistical offices count machine income [S9]; collective bargaining extends to model deployment. Nothing is called a dividend; several ledger lines quietly thicken. Distribution as plumbing.
One worldview wins by default: work-is-worth meets a world of scarce entry-level work [S7] and concludes the young are undeserving. The distribution question is settled by moral fashion, not design. The most dangerous scenario is the quiet one.
A child born in 2029 receives, at eighteen, a statement in the post. Line one: wages, whatever she makes of them. Line two, printed in green ink: Common Share — yours since birth — dividends paid each solstice. She does not remember a time before line two. Neither will her teachers. That is what a change of metaphor feels like from inside: boring, printed, permanent.
| Claim used as input | Type | Basis |
|---|---|---|
| Lab revenue compounding [S1, S2]; billionaire wealth $18.3T [S6]; entry-cohort displacement [S7, S8]; capex financed economy-wide [S4, S5]; statistics blind to machine income [S9]; Taiwan's payment [S10] | sourced | See the signal shelf (chapter 04) for each link. |
| The three worldviews and the paycheck/share metaphor as named here | assumption | Analytic framing by this study (the techno-feudalism vocabulary is a live school of political economy; the paycheck-as-contract reading is the study's synthesis). |
| That fiscal systems tax work more heavily than compute | assumption | Directionally supported by the plumbing facts above; no single source states it as such. Verify against your own tax code. |
Assumptions surfaced: that the distribution debate is stuck at litany depth in most polities. Open decisions: which worldview steers 2026–2042. Handoff: the three worldviews enter the futures triangle as candidate pulls; the reframed question becomes the brief the four futures must each answer.
Method 03 · fdbx-futures-triangle · Slaughter / futures triangle
Every future of distribution is a resolution of three forces: the images that pull us forward, the drivers that push whether we choose them or not, and the weight of what already exists. Map them before minting futures, or the futures are just mood boards.
The pushes are winning the tempo argument: revenue and displacement arrive quarterly, instruments arrive in legislative years. Every future in chapter 06 is a different resolution — growth lets the pulls ride the pushes; collapse is the pushes without the pulls; discipline is the weights used deliberately; transformation is the weights rebuilt to carry a different load. None is pure; all four carry some of every force.
| Claim used as input | Type | Basis |
|---|---|---|
| Pushes listed with serials S1–S9, S11 | sourced | Signal shelf, chapter 04. |
| The four pull-images and the five weights as named | assumption | This study's synthesis of live debates (Alaska-model dividend politics, four-day-week movement, work-ethic discourse, platform-rent critique, pilot-memory gap [S12–S15]). |
Assumptions surfaced: that the entry-cohort effect is a leading indicator, not noise. Open decisions: which pull becomes policy first. Handoff: pushes become scan-tagged drivers in the shelf; each pull seeds one four-futures chapter; weights reappear as the backcast's constraints.
Method 04 · fdbx-horizon-scanning · UK Government Office for Science Futures Toolkit
Scoping question: what could change by 2042 in the share of the economy that reaches individuals, while AI companies capture a growing share of value? Every scan below was found and linked in this session — mode: scan. PESTLE balance is counted at the foot; the gap is named, not hidden.
Anthropic's annualized revenue run-rate passed $65B at the end of July 2026 — up from roughly $30B in April — with reports of positive adjusted operating income and company projections of $100–120B annualized by year end. Why it matters: the wage-competing side of the ledger (jobs) and the capital side (lab profits) now compound on different clocks.
OpenAI's annualized revenue stood near $25B in February 2026; Anthropic passed it by Q2. Why it matters: "the AI industry" is not one balance sheet — the race concentrates revenue into one or two cap tables at a time, which is precisely where distribution instruments must attach.
Nvidia booked $96.2B of revenue in its fiscal Q2 2027 — mostly data center — beating its own guidance by ~6%. Why it matters: the picks-and-shovels layer converts the whole economy's AI demand into one equity story; whoever owns that share class owns the machine age's rent.
Big Tech's ~$630B 2026 AI spend equals roughly 2.2% of US GDP — more than four times its 2023 share. Why it matters: everyone's grid, water and pension funds are now counterparties to the labs' plans; there is a distribution case before a single model is shipped.
Goldman Sachs' baseline anticipates roughly $7.6 trillion of cumulative AI capex between 2026 and 2031. Why it matters: this is the scale at which "who captures the returns" stops being a markets story and becomes the political economy question of the decade.
Oxfam's January 2026 Davos report: billionaire wealth jumped 16% in 2025 to a record $18.3T — about three times the five-year average — with a large share unearned (inheritance, monopoly, connection). Why it matters: the capital line of the household ledger is alive and well — for very few households. AI is expected to widen the engine of this.
Stanford's Digital Economy Lab update (Aug 2026): employment of US workers aged 22–25 in the most AI-exposed occupations — software, customer service — stands about 19% below its counterfactual, up from 13% in the August 2025 first cut, while older workers in the same occupations are barely touched. Why it matters: displacement, where real, arrives as a missing first rung, not a layoff notice — the hardest kind to insure against.
Anthropic's CEO said AI could eliminate roughly half of entry-level white-collar jobs within one to five years, spiking unemployment to 10–20%, and that officials should stop "sugar-coating" it — then repeated the warning a year on. Why it matters: when the labs' own leadership prices displacement into the public debate, the distribution question stops being hypothetical; insurers and treasuries can start underwriting.
Epoch AI finds US GDP growth has been underestimated by about 0.3 percentage points because official statistics miss most of Nvidia's US-generated income. Why it matters: you cannot redistribute what you cannot measure; the plumbing of distribution starts in the statistical manuals.
Taiwan's proposed 2027 budget sets aside NT$235.7B (~US$7.4B) for a universal payment of NT$10,000 (~$314) to every person, funded by windfall revenue attributed largely to TSMC and framed publicly as an AI dividend — a step up from the NT$6,000 surplus payment of 2023. Why it matters: the first working precedent for chip-windfall-to-household distribution; small in amount, enormous in precedent.
OpenAI completed its restructuring into a public benefit corporation (Oct 28, 2025): Microsoft holds roughly 27% (~$135B), and the controlling nonprofit parent holds an equity stake reported to fund over $100B of mission work. Why it matters: the rare existing case of a non-investor public-interest claim on lab equity — a germ of the ownership instruments this study drafts in chapter 08.
OpenResearch's randomized trial — 1,000 people received $1,000/month for three years against a 2,000-person control — found modest work-hours reduction, more agency and entrepreneurial activity, better parenting quality and early food-security gains, but no durable physical-health improvement. Why it matters: the closest thing the US has to a long UBI trial, funded by an AI lab's founder — the evidence ceiling for "what cash alone does".
The Stockton SEED pilot ($500/month, 125 residents, 24 months): recipients' full-time employment rose from 28% to 40% in year one, against 32%→37% in controls; less anxiety, more stability; spending concentrated on necessities. Why it matters: the canonical counter to "cash makes people stop working" — at pilot scale.
The Alaska Permanent Fund Dividend — an annual oil-revenue cheque to every resident for four decades — shows no effect on employment and a +1.8pp (~17%) rise in part-time work in the key study. Why it matters: the only true permanent dividend in a wealthy economy; the closest existing thing to a commons line in a household ledger.
When Iran replaced subsidies with monthly cash transfers in 2011, roughly 70 million people received payments worth around 55% of a minimum-wage income — and studies found no negative effect on labor supply. Why it matters: the only nationwide, population-scale cash-transfer case; it brackets the "scale will break it" objection from above.
Finland's 2017–18 experiment (2,000 unemployed people, €560/month) found no employment effect in year one, slightly more days worked in year two — and consistently better life satisfaction, mental health and trust in the future, with less bureaucracy. Why it matters: the case that moved the debate from work incentives to wellbeing, at a cost no one noticed on the state's books.
GiveDirectly's Kenya UBI study — the world's largest and longest, with MIT/IPA researchers, ~$25M reaching 20,000+ people, a 12-year arm paying ~$0.75/adult/day through ~2028 — reports durable spending gains and resilience through the COVID shock in interim results. Why it matters: the developing-world evidence base, and the design the Global South would actually argue about.
More than 150 US cities have run guaranteed-income pilots; a lawmaker coalition counts ~30,000 Americans paid about $335M in total. Tacoma launched a city pilot in January 2026; Cook County budgeted $7.5M for FY2026. Why it matters: the distribution instruments are accumulating administrative muscle memory — procurement, payment rails, evaluation — before any federal decision.
H.R. 5830 (119th Congress) would establish a federal guaranteed-monthly-income pilot and study. Why it matters: the first formal federal vehicle; its committee fate is a cheap-to-watch indicator of whether distribution instruments stay municipal or go national.
Analyst tallies put combined hyperscaler 2026 capex near $725B, with each of the four giants spending well over $100B — AI capex now roughly 5% of GDP and climbing in investment-weighted measures. Why it matters: the fiscal capacity any dividend scheme competes with; the build-out is the budget.
PwC's central scenario projects $31.6 trillion of global data-center capex through 2050. Why it matters: the horizon at which infrastructure-for-equity swaps (chapter 08, Instrument V) become the largest ownership question states have faced since electrification.
Nvidia forecasts $3–4 trillion of data-center spend by decade's end, with annual capex run-rates around $600B. Why it matters: when the suppliers publish the demand curve, the "it might not happen" hedge thins — planning distribution for zero AI windfall stops being prudent and becomes evasion.
UV · SHELF AUDIT: 22/22 SCANS SOURCED THIS SESSION · ENVIRONMENTAL COUNT: 0 — THE GAP IS DECLARED, NOT DECORATED
| Theme (cluster) | Scans | What it means for individual shares | Type | Impact × Likelihood |
|---|---|---|---|---|
| A · The machine earns | S1–S5, S20–S22 | Revenue and capex compound on capital's clock; the pool a distribution instrument could tap is real, growing and legible. | Established trend | 5 × 5 |
| B · Concentration lands first | S6, S7, S8 | The costs arrive as missing entry rungs and compounding wealth before any dividend arrives — the political fuel and the policy clock. | Risk / opportunity | 5 × 4 |
| C · States start printing shares | S9, S10, S11 | Measurement reform, one working chip dividend, one public-interest equity stake: the instruments have precedents, not just advocates. | Emerging issue | 4 × 4 |
| D · The cash canon | S12–S17 | Six serious cash cases — pilot to nationwide — converge: little or no work reduction, wellbeing up, cash alone changes no structure. | Expected development | 4 × 5 |
| E · The movement institutionalizes | S18, S19 | Payment rails, evaluations and budget lines accumulate below federal level — capability precedes decision. | New opportunity | 3 × 4 |
PESTLE count: Economic 9 · Social 9 · Political 3 · Technological 3 · Legislative 1 · Environmental 0. The environmental gap is real and named: the power, water and land politics of the build-out enter this study only through capex [S4, S20, S21]. A second scan round should chase it.
Drivers of change (the study's steering set): ① the $7.6T capex baseline [S5]; ② entry-level displacement as leading indicator [S7, S8]; ③ the first working chip dividend [S10]; ④ measurement reform [S9]; ⑤ the permanent-dividend precedent [S14]; ⑥ institutional muscle memory below federal level [S18, S19].
What to watch next: Taiwan's 2027 budget passage; H.R. 5830's committee fate; the Kenya 12-year endline (~2028); whether any lab adopts a binding windfall-sharing charter.
All 22 scans above are sourced: found via web search in this session, each with a link the reader can open. No scan is user-supplied (the brief supplied the question, not the signals). Leads to verify are typed assumption and excluded from counts, clusters, ratings and drivers, per the toolkit. Handoff: theme A's drivers become the four futures' shared present; theme D's canon feeds the UBI inquest (chapter 07); the pull-images from the triangle (chapter 03) seed each future's internal logic.
Method 05 · fdbx-three-horizons · Curry / Hodgson
The question is not "will there be a dividend" but "how does the composition of a person's income change, and when". Three horizons let the study hold the transition — the messy middle — without pretending it is a switch.
Wages still carry 60–95% of every household stack in chapter 01. The AI build-out is macro news [S4, S5] but micro-patchwork: pilots, tax credits, one-off cheques [S18]. The entry-level rung wobbles [S7, S8]; the political system answers with retraining vouchers and press-release dividends. H1's gravity: every new instrument is debated as a payment, because the paycheck metaphor still owns the debate.
Pockets of H3 visible in H1: the Alaska cheque [S14], Taiwan's budget line [S10], the nonprofit equity stake in one lab [S11], 150 cities' payment rails [S18].
The middle where this study's fight lives. Wages soften at the exposed edges while three new lines thicken: transfers (pilot infrastructure goes permanent in some polities), capital (index access spreads; lab equity stays concentrated), and — the contested line — commons income (chip-windfall funds, model royalties, grid-for-equity trusts begin to pay). Households become portfolio managers whether or not they wanted the job.
Fenna de Vries' illustrative H2 stack (~2034) — the hybrid life arriving early for portfolio people. Constructed illustration, not a forecast.
The H2 dilemma: whoever designs the commons line in H2 locks in who owns H3. Design instruments in the transition, or the transition designs you.
Two third horizons compete inside the same pocket of the present. H3-growth: the commons line reaches 10–25% of median stacks where instruments were built in H2; the paycheck loses its monopoly on worth; birth becomes a position in the common machine age. H3-collapse: no commons line was built; capital's share compounds [S6]; individuals hold licences to platforms they own nothing of — techno-feudalism with a state stamp. Which H3 lands is decided by H2 plumbing, not by 2042 speeches.
Portfolio classification: invest H2 energy in royalty bases, measurement [S9], endowments, stakes — slow plumbing; keep pilots [S18] as capability; avoid one-off cheques as strategy — H1 instruments cosplaying as H3 architecture.
| Claim used as input | Type | Basis |
|---|---|---|
| Signals referenced by serial (S4–S18) | sourced | Signal shelf, chapter 04. |
| Fenna's H2 stack and all percentage compositions | assumption | Constructed illustration for one composite person; deliberately labelled, not projected. |
| Dating of horizons (2031/2036 boundaries) | assumption | The study's own periodisation, chosen for the 2042 horizon; treat as scaffolding. |
Handoff: the four futures (chapter 06) are four different H2→H3 resolutions; the UBI inquest (chapter 07) tests whether the H1 instrument can carry H2 weight; the instruments (chapter 08) are the H2 plumbing kit.
Method 06 · fdbx-four-futures · Dator's four generic futures
Dator's rule: any useful statement about the future should appear ridiculous to the average citizen of the present — and no future is the best case, the worst case, or the likely one. Each 2042 below answers the reframed question its own way, and prints one composite person's payday from inside itself. Every stack is a constructed illustration — anchored to the signals, shaped by hand.
Growth is the point; distribution rides growth's coattails. The windfall is big enough to share without anyone fighting over principle.
republic of india · model endowment · q3 2042
“Mini-statement: Model Endowment of India — quarterly credit ₹2,843. Your share of the machines, as legislated 2033. Next solstice credit: December.”push notification, Priya's phone, Hyderabad
The world: The capex baseline [S5] turned productive. AI value-added shows up in the national accounts — statistics were fixed early [S9] — and chip-windfall funds pay a real, if modest, universal credit in a dozen countries, Taiwan's precedent [S10] having gone continental. Wages didn't vanish; they re-mixed: agent-staffed junior tiers, human senior tiers, and a premium for verified-human work. The entry rung is rebuilt wider than it ever was — as commons-funded apprenticeship, not as the boss's charity.
Disowned by this future: the question of what happens when growth disappoints — the dividend was never stress-tested against a flat decade.
senior engineer & team lead · Hyderabad
2042 payday · ₹96,400/mo all-in: ₹53k wage · ₹19k index & ESOP growth · ₹2.8k endowment × 1 (quarterly ÷ 3) + ₹21.6k state top-up — first family in her line's history with two income lines nobody works for
In 2026 she feared the ladder's bottom was gone [S7]. In this 2042 it is back — funded by the endowment, not by her firm. She mentors four commons-paid apprentices; the cheapest talent pipeline the state ever built.
Not apocalypse — the end of a deal. The machines earned; the earning stayed upstream; everyone downstream pays licence fees to hold a position.
venture workspace licence · lagos · renewal q1 2042
“RENEWAL NOTICE — your human-operator licence (support tier) +18%: compute pass-through. Reply PAUSE to keep your queue at 40 seats. — Issued automatically.”platform message, Mariam's queue console, Lagos
The world: The revenues compounded [S1, S6]; the institutions never built a commons line. Work didn't disappear — it was licenced: humans sell "operator seats" and "voice of last resort" shifts on platforms that own the models, the queue and the metrics. Middle-class incomes downstream of the labs erode as fees pass through. States with fiscal muscle buy their citizens out of the worst; states without print nothing. The stockmarket and the street live in different countries.
Disowned by this future: its own dependence on demand — rent leaves customers with nothing to rent with. The manor imports grain.
licensed support operator · Lagos
2042 payday · ≈₦120k/mo equivalent across two operator seats + market stall: fees deducted at source · remittances now flow in from her daughter · total purchasing power ≈ a quarter of her 2026 real wage · commons line: still zero
The layer global agents reached first [S8] turned out to be hers. She was excellent at the job — the job was re-priced under her. In this 2042 nobody robbed her; a thousand small renewals did.
Enough. Deployment is negotiated, growth is slower, and the paycheck metaphor was never overthrown — it was defended, with paperwork.
local 444 · negotiated deployment board · Cleveland
“AGREEMENT 9 — pick-path module runs under audit; 60% of verified savings fund the Adjustment Pool. No displacement without 18 months' notice, retraining credit, and pool top-up. Signed, both sides.”board notice, distribution centre break room
The world: After the early-2030s displacement scares [S7, S8], labour law reorganized the machine age rather than mourning it: deployment agreements, automation adjustment pools funded from verified savings, licences on high-impact deployments. Productivity grew about half as fast as the growth future's; median security grew faster. The commons line exists but stays small and pension-flavoured — the fund pays for retraining and top-ups, not philosophy. Stability is the achievement; imagination is the price.
Disowned by this future: the people outside the agreements — informal workers, migrants, the licensed seams of the economy the boards never sit on.
retired operator, now certified deployment auditor · Youngstown
2042 payday · ≈$4,100/mo: pension $2,700 · auditor honorarium $500 · pool top-up $600 · Adjustment Fund dividend $165 · the same security he had at 52 — that was the whole point
His audit stamp is the future's quiet hero: every module he signs off has been argued over by people who will still live in this town when the module is retired. It is slow. It holds.
The metaphor changed first. Intelligence infrastructure joined land, air and sea as common capital — and the paycheck stopped being the only respectable line on the statement.
low countries model fund · solstice 2042
“Solstice credit: €712 — your share of the Low Countries Compute Trust. Provenance seal verified on 41 prints this quarter. The machines are running; so are you.”statement, Fenna's studio inbox, Rotterdam
The world: The plumbing all got built in the 2030s, chapter 08's instruments among it: model royalties fund per-capita credits; grid-for-equity trusts own pieces of every data-center; birthright stakes mature at forty; a caretaker's wage pays for the work markets never priced. The workweek is where the four-day century left it — shorter. Verified-human provenance re-priced craft, care and teaching upward. Incomes are lower than the growth future's at the top and higher at the median; nobody's line is zero; the word "deserving" lost its job.
Disowned by this future: its own fragility at the borders — a commonwealth of a dozen rich states while the Mariams of the world still renew licences. The unfinished argument, in green ink, on the last page.
illustrator & craft teacher · Rotterdam
2042 payday · ≈€3,400/mo: commissioned prints & teaching €1,400 · loft & index €800 · solstice credits €885 · stake income matured at 40, €310 · studio days: three · every line earned by something — only two by jobs
Her 2026 portfolio life [chapter 01] turned out to be the rehearsal. The provenance seal on her prints reads like a signature; the trust's credit reads like weather. Both arrive. Neither owns her.
| Future | Distribution logic | Median stack (illustrative) | Who holds the machine age | Tomás, the control, at 77 |
|---|---|---|---|---|
| Growth | Windfall funds pay while growth funds them — distribution as dividend of expansion | 55 / 20 / 8 / 17 | Endowments + private cap tables, side by side | Pension solid; commons credit arrives with the orchard money; he likes the solstice |
| Collapse | No line but the licence — distribution as whatever leaks down from the platforms | 71 / 3 / 26 / 0 | Cap tables, absolutely | Pension taxed to fund other people's stability; his students keep the licenses he never needed |
| Discipline | Negotiated protection — distribution as contractual peace, audited line by line | 12 / 18 / 66 / 4 | The agreements hold it in trust; nobody "owns" it | His pension is the system's crown jewel; he teaches its history, slowly, to people who will defend it |
| Transformation | Commons capital pays by right of birth — distribution as inheritance, renewed each solstice | 41 / 24 / 9 / 26 | Everyone, a little; nobody, entirely | His grandchildren hold stakes since birth; he calls them "the second line" and pretends he predicted it |
The control case reads the same in every future: Tomás's share was never about the labs — it was about whether the state keeps promises and whether the tax base grows polite or wild. That is the deepest finding of the exercise: AI distribution is a tax-and-ownership story wearing a technology costume.
| Claim used as input | Type | Basis |
|---|---|---|
| All signal serials cited per future (S1–S11) | sourced | Signal shelf, chapter 04. |
| Every 2042 stack, amount, institution (Model Endowment of India, Adjustment Pool, Low Countries Compute Trust) | assumption | Constructed illustrations. Institutions are invented; their mechanics are drawn from the instruments in chapter 08 and precedents S10, S11, S14. |
| The four futures' structures (growth / collapse / discipline / transformation) | sourced-method | Dator's four generic futures, as routed by fdbx-four-futures. |
Assumptions surfaced: that stack composition, not income level, is the load-bearing variable for wellbeing in a machine age. Open decisions: which future's plumbing gets built in H2. Handoff: the collapse future's licence-fees and the transformation future's instruments go to black-mirror and worrystorming in the ethics wall; the preferred future (transformation, with discipline's audit habit) is the end-state for backcasting.
Method 07 · the inquest · six sourced cases, one verdict
Universal basic income is the instrument everyone reaches for first — so this study cross-examines it before drafting anything new. Six serious cases now exist, from 125 people to 70 million. They agree about more than their advocates and enemies admit.
The claim under examination: "when AI labs earn more, give everyone an unconditional payment — a UBI — and the distribution problem is solved." The inquest asks three questions of it: what does the evidence say cash does; what can no pilot ever show; and what problem is UBI actually solving? Witnesses below are the six cases the signal shelf found [S12–S17].
| Case | Design | What happened | Serial |
|---|---|---|---|
| US · OpenResearch (2020–23) | 1,000 people, $1,000/mo for three years, against 2,000 controls; largest US cash RCT, funded by an AI founder | Work hours down modestly; agency and entrepreneurship up; parenting quality up; food security stress down early; no durable physical-health gain | S12 |
| US · Stockton SEED (2019–21) | $500/mo, 125 residents, 24 months | Full-time employment up 28%→40% (controls 32%→37%); less anxiety; money spent on necessities | S13 |
| US · Alaska (since 1982) | Permanent annual oil-fund dividend to every resident — the only standing case in a rich economy | No employment effect; part-time work +1.8pp (+17%); poverty down, especially among Indigenous Alaskans | S14 |
| Iran (2011– ) | Nationwide subsidy replacement: cash to ~70M people, ≈55% of a minimum wage | No negative labor-supply effect — the only population-scale case, and it broke the "everyone stops working" prior | S15 |
| Finland (2017–18) | €560/mo to 2,000 unemployed people, two years | Year 1: no employment effect. Year 2: slightly more days worked; wellbeing, mental health and trust up; bureaucracy down | S16 |
| Kenya (2018–~2028) | GiveDirectly's 12-year RCT: ~$0.75/adult/day long-term arm, lump-sum and short-term arms, ~20,000+ people | Interim: durable consumption gains, COVID resilience; the definitive endline is the field's next landmark | S17 |
Across six designs on three continents, hours fall modestly at most, and in three cases measured work rose. The "free money makes idlers" prior is simply not in the data. What cash buys is risk-bearing capacity: the job search, the move, the course, the two-day-old business.
Mental health, stress, parenting, trust in the future: the wellbeing effects replicate everywhere; structural effects (health endpoints, community spillovers) mostly don't. Cash is fast, real and shallow — a floor, poured in days, that holds nobody's roof up.
No pilot tests: a permanent payment's macro effects, general-equilibrium rents (landlords know what you receive), the politics of the tax that funds it, or a generation raised inside the norm. Pilots are evidence about households, not about systems. Iran [S15] and Alaska [S14] are the only near-permanent cases — one accidental, one oil-funded.
A "real" US UBI at the trial's own dose — $1,000/month for ~260 million adults — costs about $3.1 trillion a year: roughly 11% of US GDP, or — for scale — nearly five times the entire 2026 AI build-out [S4], and vastly more than the labs' current combined revenue [S1, S2]. Taiwan's pioneering dividend [S10] is honest about this: $314 a person, once — a precedent, not a living.
The conclusion is not "therefore UBI is impossible". It is: a universal cash floor is a state-sized project with state-sized money, and no AI windfall on any sourced horizon funds one alone. UBI is the floor. A house needs capital and commons lines — which is what chapter 08 builds.
| Dial | Low setting | High setting | What the evidence hints |
|---|---|---|---|
| Universality | Means-tested top-ups (cheap, shaming, high dropout) | Everyone, no questions (costly, dignified, simple) | The bureaucracy relief [S16] is itself a benefit; universality buys legitimacy, at price |
| Level | Symbolic dividend (Alaska's ~$1–2k, Taiwan's $314) | Living floor ($1,000/mo, trial dose) | Symbolic level changes politics, not ledgers; floor level changes ledgers, and must be funded like a pension |
| Funding | Windfall skim (resource/compute royalties) | General taxation (income, VAT, land) | Windfall funding is palatable but volatile and small; general funding is durable but must win every budget |
| Conditionality | None (the U in UBI) | Participation- or verification-based | Unconditionality is what makes the evidence clean — and the politics hard; chapter 08's caretaker's wage is the honest compromise |
UBI is a floor, not the house. The canon shows cash lifts wellbeing, enables work, and offends no economy — at any scale yet tested. It also shows cash alone answers poverty while leaving ownership untouched: in the Rent Age [chapter 06, collapse], a UBI would simply be a subsidy to the landlords of the licence economy. The study's position: keep building the floor — the pilots are working, fund them [S18, S19] — and spend the design energy of the 2030s on the three lines cash cannot build: capital, commons, and the re-pricing of human work.
| Claim used as input | Type | Basis |
|---|---|---|
| All six case rows | sourced | S12–S17 in the signal shelf (chapter 04), each with links. |
| $3.1T/yr cost of a $1,000/mo adult US UBI ≈ 11% of GDP; 5× the 2026 build-out | assumption | Constructed illustration: 260M adults × $12k = $3.12T; US GDP ≈ $29T (2026 est.); build-out $630B [S4]. Order-of-magnitude arithmetic, not a fiscal plan. |
| "Cash is fast, real and shallow"; "subsidy to the landlords" argument | assumption | The study's synthesis of the canon above; the landlord pass-through concern is a standard general-equilibrium critique, not sourced to a specific study here. |
Assumptions surfaced: that the trial dose ($1,000/mo) is the politically imagined dose. Open decisions: the four dials. Handoff: the dials become the design constraints the five instruments must respect; the "floor not house" verdict is the brief chapter 08 answers.
Method 08 · the study's own drafting table · five instruments, beyond UBI
Original concepts drafted for this study — each converting a latent, already-common asset into a standing income line. None is a handout; every one is an ownership or re-pricing mechanism with real precedents. Each carries its funding, its stress-test, and its first ninety days.
Mechanism: every foundation-model training run or deployed model above a public compute threshold pays a small royalty — a percentage of revenue attributable to that model class — into a national Model Endowment. The endowment is invested like a sovereign fund; its yield pays a per-capita credit, Alaska-style, each solstice.
Funding: the labs' gross model revenue; a 1–5% take. The ready-made hook: safety regimes already log large training runs at compute thresholds — the tax collector's ledger exists; only the rate is missing.
Precedents, real: the Alaska Permanent Fund Dividend [S14]; Taiwan's TSMC-funded universal payment [S10]; mineral royalties, the oldest fiscal instrument on earth. Novelty: the royalty attaches to compute and model classes, not to land or oil — following the asset that actually moves.
Stress-test: jurisdictions race to zero; labs re-domicile. Counter: royalties ride the market (deployment country), not the lab's address — and the compute registry makes under-reporting a safety violation, not a tax dodge.
First 90 days: legislate the registry→treasury pipe; set a 1% pilot rate on the largest model class; publish the endowment's charter before the first credit lands.
Mechanism: at eighteen, every citizen receives a locked, diversified Personal Fund with an AI-era mandate: broad index, machine-age infrastructure, and a slice of the nation's Model Endowment units. It cannot be cashed out before forty — but it can be borrowed against at named milestones: education, first home, a business, a caregiving decade.
Funding: an endowment seeded at birth (progressive inheritance + a slice of the Compute Royalty's principal), compounding for eighteen years before it pays anyone.
Precedents, real: the UK's Child Trust Fund (2002–2011); "baby bonds" proposals in the US; Sweden's AP funds; Singapore's CPF. Novelty: the borrowing rail — liquidity without liquidation, so the stake buys a life at twenty-five without selling the machine age at twenty-five.
Stress-test: markets fall on maturation cohorts; the stake becomes a scapegoat. Counter: lifetime averaging and an explicit solvency rule written into the charter, so the bailout argument never gets its case.
First 90 days: charter the fund's mandate; start the ledger with the state's own contribution to newborns; publish the milestone-borrowing rule.
Mechanism: a public wage for verified care and commons work — elder companionship, childcare co-ops, repair cafés, mentoring, local first response, open-source maintenance — attested by the communities that receive it, paid by the state at a local living-wage rate. Not means-tested, not unconditional: participation-priced.
Funding: general taxation, at pension-scale seriousness — because that is exactly what it is: a pension for work the market never invoiced.
Precedents, real: Atkinson's participation income; care-income experiments; open-source fellowship funding. Novelty: the attestation rail — communities verify with reputation and sampling instead of case-workers with clipboards — and the explicit AI-era logic: pay deliberately for the work machines can't fake, while the market re-prices it upward [growth future's apprenticeships are this instrument, staffed].
Stress-test: attestation capture — busywork theatre, or capture by loud minorities. Counter: randomized audit sampling published openly; per-attestant caps; the scheme's own statistics on its website.
First 90 days: pick three work classes (eldercare hours, repair cafés, code review of public digital infrastructure); pay ten neighbourhoods; publish the attestation ledger.
Mechanism: no high-impact AI deployment without a filed Adjustment Schedule: an escrow of a share of verified savings (the study's futures use 60%) that funds, for affected roles — wage insurance at 80% for up to 24 months, retraining credits, portable benefits, and a top-up to the public adjustment fund. It is insurance priced to the deployment, not a punitive "robot tax": no displacement, no draw; savings that never materialise release the escrow.
Funding: the automation's own savings — the cleanest funding base in this chapter, and the reason firms co-sign it.
Precedents, real: the US Trade Adjustment Assistance (funded by everyone, protecting few); Danish flexicurity; union deployment agreements — which this instrument generalises to every firm, not just unionised ones. Novelty: the escrow attaches at deployment filing, turning displacement from an externality into a prepaid liability.
Stress-test: "verified savings" becomes an accounting war. Counter: the auditor corps — the Discipline future's best invention — with published formulas and a bilateral appeals board; a standard savings-measurement method, like accounting standards, not litigation.
First 90 days: draft the schedule template with one employer and one union; file it voluntarily for one module; publish the escrow's first quarterly statement.
Mechanism: when a data-center asks a state for the one thing only the state can give — a gigawatt of grid, water, land, and the right to interconnect — the permit carries a price in non-voting preferred equity, held by a public Compute Trust that pays a per-capita dividend. States have always negotiated these deals with tax breaks; this instrument prices the scarce input in ownership instead of abatements.
Funding: nothing new is taxed — the exchange happens at permitting, before ground is broken. Scale: against a $7.6T build-out baseline [S5] and $31.6T through 2050 [S21], a mid-single-digit equity share on public-grid deals is sovereign-fund scale by the 2040s.
Precedents, real: sovereign wealth funds (Norway's oil-to-equity conversion is the grandparent); Taiwan's chip-windfall dividend [S10]; the nonprofit equity stake in one lab [S11] — public-interest claims on AI capital already exist; this makes them systematic and local. Novelty: trading the one input states control absolutely — interconnection — for the instrument states never seem to get: equity.
Stress-test: the counterfactual flight — "build elsewhere". Counter: grid interconnection is geography; compute is chasing power, not tax codes [S4, S20]. The bargaining lever is real for exactly as long as the build-out lasts — which is the argument for starting now.
First 90 days: one utility, one campus: publish the model permit with an equity schedule attached; let the next bidder react to it.
The royalty's arithmetic is small, public and checkable — which is its virtue. Pool is a model-class revenue base; the take is the royalty rate; certificates are the people paid. The mint prints your result as a specimen note. All figures constructed illustrations; the machine just multiplies.
—
SPECIMEN — printed by your dials, not by a treasury. The engraving is regenerated per visit; the arithmetic never is.
One person's stack, instrument-era 2042 (the growth future's composition). Constructed illustration.
They are a portfolio, not a menu: a polity that builds only the comfortable ones (credits without insurance) gets the growth future's fragility; one that builds only the defensive ones (insurance without ownership) gets the Discipline future's ceiling. The transformation future is the full stack — which is why its people's ledgers have four lines and no shame.
| Claim used as input | Type | Basis |
|---|---|---|
| Precedents cited per instrument (Alaska [S14], Taiwan [S10], PBC stake [S11], capex baselines [S4, S5, S21]) | sourced | Signal shelf, chapter 04. |
| Compute-threshold registries as an existing tax hook; Child Trust Fund; baby bonds; participation income; TAA; flexicurity; sovereign-fund mechanics | assumption | Institutional knowledge from public record, not re-sourced this session — each named so it can be checked. |
| Mint outputs | assumption | Deterministic arithmetic on user-set dials; no hidden assumptions beyond the three inputs. |
Assumptions surfaced: that compute registries will exist at national scale (safety regimes are building them); that deployment geography follows power, not tax. Open decisions: rates, mandates, and which instrument goes first. Handoff: each instrument's stress-test goes to the ethics wall's black-mirror and worrystorming bays; the preferred portfolio goes to backcasting.
Method 09 · the edbx ethics wall · ethical-design kit, four bays
Drafting five instruments is the fun part; designing for the people they fail is the work. The wall runs the edbx kit in four bays — Whose share? · What could go wrong? · How do we decide? · What do we promise? — each lens labelled with the method that produced it.
Every instrument here is designed at the scale of a competent state. CIDER walks the walls: informal workers attested by nobody; migrants whose stakes don't travel; the states without compute to tax or grid to trade; disabled people whose "participation" doesn't fit attestation templates; the next generation, who inherit the funds and the frozen political bargains both. Design rule adopted: every instrument ships with a border clause — what a visitor, a migrant and an informal worker hold, written down, not improvised.
Take the frame the pilots can't test [chapter 07]: a universal credit is also revenue. Landlords, platforms and fee-takers price it in; the commons line can quietly become a subsidy to whoever owns the necessities. The lens demands each instrument publish its pass-through estimate — who is positioned to capture the credit — and a monitor for it. What the housing literature knows, the machine age will re-learn at compute speed.
The five people in this study are composites; the instruments will touch people who are not. Humane design's rule for sensitive systems applies: the attestation rails (Instrument III), the adjustment boards (IV) and the trust charters (I, V) must be co-designed with the cohorts they name — gig operators, care workers, the under-25 cohort the canary study tracks [S7] — with real agendas, not consultation theatre.
2042, plausible route: the Compute Royalty succeeds. The Endowment holds so much lab equity that the state's solvency now tracks the labs' stock price. Every dividend letter carries, beneath the fold, a gentle reminder of the sector's importance; oversight hearings acquire a new deference. The commons line was built to free households from dependence on the labs — and made the state their shareholder. Guard written into the charter: mandatory index-diversification away from the royalty-paying sector within N years, published annually.
The Birthright Stake matures at forty; the borrowing rail works. Within a decade: lenders specialise in "stake-secured" credit, fees bloom, and a third of each cohort reaches forty with nothing — having sold their machine age at twenty-four, at pawn rates, in a bad year. The stake meant to equalize becomes a callable asset for whoever holds the liquidity. Guard: borrowing only from the fund itself, at published rates; losses mutualised; no third-party pledges, ever, as constitutional text.
Caretaker's Wage, two budget crises in: attestations become performance. Grandchildren log companionship hours from the parking lot; churches and block clubs become payroll machines; the visible poor are recast as "participants" and the exhausted as "non-compliant". The instrument that re-priced care re-prices it again, in bureaucratese. Guard: sampling audits by the recipients' own federation, not the payer; zero sanctions for low hours — only for false ones.
The unspoken worries, surfaced per method: "What if the windfall never comes?" — then the commons line is small and the insurance is unused; both degrade gracefully, which is the design point. "What if it comes too fast?" — the Sandbag Clause's escrow is the dam. "What if a rival state free-rides?" — instruments ride markets, not addresses; the free-rider still hosts the compute. "What if we build the plumbing and the machines underdeliver?" — a sovereign fund with a modest yield and an insurance scheme nobody needs is still a better state than the one that skipped them.
The instruments are not ethically neutral; each spends one value to buy another. The Royalty buys solidarity with momentum (a take on the engine). The Stake buys liberty-with-security and pays in patience (forty years). The Caretaker's Wage buys solidarity and pays in bureaucracy-risk. The Sandbag Clause buys security and pays in momentum (slower deployment). The Trust buys solidarity with bargaining leverage that decays as the build-out ends. The decision rule adopted: over a forty-year ledger, price momentum last — a fast machine age inside a brittle society has one historical ending, and it is not abundance.
The scheme the study ends up endorsing, made explicit so it can be attacked: (1) a floor — no person's stack may have a zero line where a plausible instrument exists; (2) ownership before transfer — prefer mechanisms that leave the person owning something (stake, trust units) over flows that stop when politics blinks; (3) transitions pre-paid — whoever books the savings escrows the harm; (4) legibility — every instrument publishes its ledger, its pass-throughs and its failures. A scheme is honest when its failure modes are named in its own charter.
1. Publish the pass-through monitor for every credit, or don't ship the credit. 2. Any equity the public holds, index-diversified on a published schedule. 3. Attestation sanctions only for falsehood, never for scarcity. 4. Escrow before deployment; no retrospective sandbags. 5. Every instrument's dashboard carries its failure statistics next to its successes. Each pledge is checkable; that is what makes it a pledge.
A real version of Instrument IV is a two-party instrument: the deploying firm signs the Adjustment Schedule; the affected workers' representatives co-sign; the escrow agent is neither. The contract's penalty clause is the honest part — if savings are booked and the escrow is short, the deployment pauses. Ethics that cannot pause a deployment is commentary.
The growth critic: "You are taxing the ladder you climbed." Answer: the Royalty is 1–5% of revenue, not prohibition; the growth future funds it too. The libertarian critic: "Five new state machines." Answer: four of the five are ownership and insurance rails — the small-government tradition's own tools, pointed at the actual concentration. The justice critic: "Dividends for the rich world while the Mariams renew licences." Answer: the unfinished argument of the transformation future, stated in its own margin — the study does not resolve it; it refuses to hide it.
| Claim used as input | Type | Basis |
|---|---|---|
| All dark twins, guard-clauses and critiques | assumption | Constructed by the named edbx methods as forward-looking checks on this study's own instruments; no external claim is asserted. |
| Pass-through concern (credits captured by landlords/fee-takers) | assumption | A standard finding of the cash-transfer and housing literatures; named as a monitor requirement rather than asserted as a measured fact here. |
Handoff: the guards and pledges travel with the instruments into the backcast as constraints (nothing in the 90-day plan may violate a pledge), and Bay IV's rejections are kept visible at the receipts (chapter 11) — a study that buries its critics has already failed its own Bay I.
Method 10 · fdbx-backcasting · working back from the preferred future
The preferred future is specific: the transformation future with the discipline future's audit habit — four lines in every stack, guards and pledges from the ethics wall attached. Backcasting walks from there to Tuesday, scores each step for control, and keeps an avoid-mode watch on the future we'd rather not fund.
Endowments pay per-capita credits in twenty-plus polities; stakes mature at forty with the no-third-party-pledge clause intact; caretaker's wage is pension-scale; adjustment schedules are as ordinary as insurance certificates; compute trusts hold real equity in the machine age's siting boom [S21 scale]. Median stacks: 55/20/8/17. The pledges [Bay IV] hold, audited in public.
Cross-border mutual recognition of compute registries closes the race-to-zero; endowment yields smooth in a bad market year — the first live test of the charter's solvency rule; second-generation grid-equity deals standard in three power-rich regions.
First birthright stakes mature; national pilots [S18, S19 lineage] convert to standing floors in early-adopter states; measurement reform [S9's logic] lands in the big statistical offices — machine income is visible, therefore taxable, therefore shareable.
At least five chip-windfall funds pay recurring credits (Taiwan's precedent [S10] by then unremarkable); one major polity files its first Adjustment Schedule under a general rule; the caretaker's wage runs in three countries' neighbourhoods with published attestation ledgers.
Kenya's 12-year endline reads [S17]; the first grid-for-equity term sheet is signed somewhere with cheap power and a serious utility; the newborn-ledger pilot holds its first cohort of five-year-old accounts.
The table below. Every action scores high on control and is chosen to change what is thinkable by the next budget season — plumbing first, philosophy later.
| Action | Done by whom | Control | Leverage |
|---|---|---|---|
| Draft the Compute Royalty whitepaper: registry→treasury pipe, 1% pilot rate on the largest model class, endowment charter published first | a serious finance ministry; a think tank as second author | 5 / 5 | 4 / 5 |
| File one voluntary Adjustment Schedule with a willing employer and union — the template that becomes the rule | one sector's labour relations board | 4 / 5 | 5 / 5 |
| Publish a model grid-for-equity permit with an equity schedule attached; let bidders react to it | one utility + one state energy office | 4 / 5 | 5 / 5 |
| Pay a caretaker's wage in three neighbourhoods, attestation ledger public from day one | one city, one payment rail from the GBI movement [S18] | 5 / 5 | 3 / 5 |
| Charter the Birthright Stake; open newborn-ledger accounts for one birth cohort | one sovereign fund's board, one treasury | 4 / 5 | 4 / 5 |
| Spec the pass-through monitor — who captures each credit — before any credit ships [Bay I] | the statistics office that fixes S9's blind spot | 5 / 5 | 4 / 5 |
The collapse future backcast in avoid-mode gives tripwires, not predictions: watch-variables — licence/fee share of gig income; the depth of the entry rung [S7, quarterly]; credit pass-through into rents; cap-table concentration of model revenue [S1, S2 each quarter].
Pre-committed responses: entry-rung gap >25% for two quarters → Sandbag Clause moves from template to bill; licence-fee share rising while real pay falls → caretaker's wage expands before protest, not after; any polity printing a one-off "AI dividend" while its registry→treasury pipe is unbuilt → say so in public, kindly, with the arithmetic [chapter 07].
The ledger stays open. Write your own lines — they're stored in this browser only, like a pocket copy of the register.
Which line of your own stack do you actually control?
Which instrument would you legislate first — and at what rate?
Whose share did this study still fail to price?
| Claim used as input | Type | Basis |
|---|---|---|
| End-state and milestones reference instruments, precedents and signals as cited (S7–S10, S17–S21) | sourced | Signal shelf, chapter 04. |
| Milestone dates and adoption counts (five funds by 2031, twenty-plus by 2042) | assumption | The study's own scaffolding — a design target, not a projection. |
| Control and leverage scores | assumption | The drafting table's judgement per method; the scale is ordinal, the purpose is to force honesty about what "we" can actually do. |
Handoff: the 90-day entries are the study's answer to "so what do I do" — the receipts (chapter 11) close the loop with the full evidence table. Nothing in the plan violates a wall pledge; that check is the ethics wall's last job.
The receipts · consolidated evidence ledger & method provenance
A futures study earns its keep in this table. Three types only: sourced — found and linked in this session; record — well-known public institutional history, named so it can be checked, not re-verified this session; assumption — the study's own construction, labelled in place.
| Serial | Signal | PESTLE | Source (opened this session) |
|---|---|---|---|
| S1 | Anthropic run-rate >$65B; projections $100–120B (Jul 2026) | Economic | Axios; CNBC/TechCrunch coverage |
| S2 | OpenAI ≈$25B annualized (Feb 2026); overtaken in Q2 | Economic | run-rate coverage as S1 |
| S3 | Nvidia FQ2 2027 revenue $96.22B, data-center-led | Econ/Tech | Rex Shares |
| S4 | Big Tech 2026 AI capex ≈$630B ≈ 2.2% of US GDP | Economic | Reuters Breakingviews |
| S5 | $7.6T cumulative AI capex 2026–31, baseline | Economic | Goldman Sachs |
| S6 | Billionaire wealth $18.3T, +16% in 2025 | Econ/Social | Oxfam International, Oxfam America |
| S7 | US 22–25yo in AI-exposed occupations ≈19% below counterfactual | Social/Econ | Stanford Digital Economy Lab |
| S8 | Amodei: ~50% of entry-level white-collar roles, 1–5 yrs, 10–20% unemployment | Social/Pol | Axios, Forbes |
| S9 | US GDP underestimated ≈0.3pp; machine income unseen | Tech/Econ | Epoch AI |
| S10 | Taiwan 2027 budget: NT$10,000 universal "AI dividend", NT$235.7B, TSMC-driven revenue | Pol/Econ | BIEN |
| S11 | OpenAI → PBC; Microsoft ~27% (~$135B); nonprofit controls, >$100B mission resources | Pol/Econ | CNBC, OpenAI statement |
| S12 | OpenResearch: $1,000/mo × 3 yrs, 1,000 vs 2,000; work down modestly, agency up, no lasting health gain | Social/Econ | OpenResearch, NBER WP 32784 |
| S13 | Stockton SEED: full-time employment 28%→40% vs 32%→37% control | Social/Econ | Stockton Demonstration |
| S14 | Alaska PFD: no employment effect; part-time +1.8pp (+17%) | Social/Econ | Jones & Marinescu, AEA |
| S15 | Iran 2011: ~70M people paid; no negative labor-supply effect | Social/Econ | Salehi-Isfahani, JDE |
| S16 | Finland 2017–18: €560/mo × 2,000; yr-2 days worked up slightly; wellbeing up | Social/Econ | Gov't report, STM |
| S17 | GiveDirectly Kenya: 12-yr RCT to ~2028, ~20k+ people, $0.75/day arm; interim durable gains | Social/Econ | GiveDirectly |
| S18 | 150+ US city pilots; ~30,000 paid ~$335M; Tacoma 2026; Cook County $7.5M | Political | Smart Cities Dive |
| S19 | H.R. 5830: federal guaranteed-income pilot bill (Oct 2025) | Legislative | 119th Congress, via Smart Cities Dive |
| S20 | Hyperscaler 2026 capex ≈$725B; >$100B each; AI capex ~5% GDP (investment-weighted) | Economic | BetaFinch, WSJ |
| S21 | $31.6T global data-center capex through 2050, central scenario | Economic | PwC |
| S22 | Nvidia: $3–4T data-center spend by 2030; ~$600B/yr capex run-rate | Tech/Econ | Nvidia via mlq.ai |
horizon-scanning (GO-Science toolkit) — chapter 01 receipt + the 22-signal shelf, scoping question, PESTLE balance, drivers; causal-layered-analysis (Inayatullah) — chapter 02's four depths and the reframed question; futures-triangle (Slaughter) — pulls/pushes/weights; three-horizons (Curry/Hodgson) — the stack's transition, pockets, portfolio; four-futures (Dator) — the four 2042s and paydays; backcasting — the walk home, control scores, 90-day plan, avoid mode.
cider and anotherlens and humane-design-guide — Bay I's exclusion pass, borrowed frame and co-design rule; black-mirror-brainstorming, worrystorming, stf-et — Bay II's dark twins and quiet fears; values-levers + normative-design-scheme — Bay III's conflict and ordering; pledge-works, ethical-contract, anti-heroes — Bay IV's pledges, signatures and critics.
UV · COUNTERFEIT CHECK: EVERY FORECAST IN THIS STUDY FAILS THE PREDICTION TEST BY DESIGN · SPECIMEN · NO CURRENCY VALUE