← Back to all writing

AI futures evidence — U2: distribution & political economy

July 5, 2026

Futures index · 中文 · Main forecast

Each section: Claim · Why · Evidence · Analogue · Would update if · Conf (H/M/L).


Parent: Utopia timeline · my putopia
Modal contrast (NOT utopia): c8 society snapshot by ci
Doom mirror: node1 (C4 labor shock — failure branch)
Date: 2026-07-04
Settings: Hybrid time (C); modal + success-tail branches; objective: high P(concentration) is plausible
Purpose: Every probability and timing claim for shared AI gains (U3/U4 buckets), with evidence, analogues, falsifiers, and anti-utopia signals (Path D/E).


TL;DR

Node U2 asks: P(AI productivity gains are broadly shared) via SWF, compute dividend, tax/competition mix, augmentation policy — not merely created. Modal (~55–65%): Path D — Ghost GDP dynamics, compute-owner bifurcation, patchwork labor rules, no federal UBI/SWF by 2032. Success window opens @C4 (Block-class cuts, Canaries −16%, Challenger 101k AI-cited YTD) — same trigger as doom Node 1, but success branch = redistribution coalition + institutional absorption, not pause. Primary competitor to U3/U4: concentrated harm (P 0.68–0.78 that top-decile capture ≥60% of AI-attributed surplus through C8). Top cruxes: (1) Ghost GDP without transfer enactment; (2) so-so automation > machine usefulness; (3) SWF rhetoric >> SWF law; (4) Global South rule-taker status; (5) UBI solves income not Jahoda latent functions.

Bucket mapping: U3/U4 primary; U2 symbiosis blocked if distribution fails even when alignment succeeds (Path E).


Hybrid timing windows (@C4–C8)

T = 2026-09 → 2027-03 — C4 capability window (distribution lag)

Claim: Junior white-collar automation becomes economically binding before redistribution institutions respond — creating a 12–24 mo policy window where concentrated capture can lock in.

Why: Hybrid rule (C): capability fast; politics/economy +30%. Block, Canaries, Challenger already live; SWF/UBI bills introduced but not enacted.

Evidence:

Analogue: Post-WWII delay — productivity surged 1940s; median wage catch-up required GI Bill + union density + tax through 1950s–60s.

Would update if: Federal enacted compute-dividend or SWF law before 2027-Q2 (pulls success branch forward).

Conf: H (capability); M (lag duration)


T = 2027-02 → 2027-09 (peak ~2027 Q2) — C4 policy window (success-branch salience)

Claim: Peak political salience for labor/distribution frames (~2027 Q2) opens the best 18–36 mo window for SWF, augmentation mandates, tax rebalance — mirroring Node 1 salience but inverted outcome menu.

Why: Node 1 P(HELP hearings by peak) 0.75; Newsom EO N-6-26 (May 2026) study-only; Sanders SWF bill Jun 2026 — agenda set, not law. 79% voters want worker-protection plan (Sanders op-ed, Blue Rose poll cited).

Evidence:

  • CA EO N-6-26: severance, transition support, universal basic capital concepts, WARN review — 180-day recommendations due ~Nov 2026 (gov.ca.gov)
  • SB 951 (CA): AI displacement notice 60-day — pending (leginfo)
  • Node 1 P(economic anxiety ≫ x-risk) 0.85 — distribution fights on jobs terrain

Analogue: Montreal Protocol 1987 — vivid harm + narrow chokepoint → binding treaty in policy window; here chokepoint = tax/incorporation, not CFC chemistry.

Would update if: Zero state/federal distribution bills introduced after second Block-class event → window closed (normalized absorption).

Conf: M


T = 2028 H2 – 2029 — C8 public AGI / remote-worker threshold

Claim: At C8, Ghost GDP debate goes mainstream; transfer program enactment remains slow (P 0.25–0.35 any G7 material redistribution law by 2029).

Why: c8 §C8 — GDP up, wage share down; political pressure ↑, enactment lag per institutional cross-cut.

Evidence:

  • c8 society snapshot by ci §C8 — Ghost GDP mainstream [SPEC]; transfer enactment slow [EST]
  • ai_meltdown_ghost_gdp_crisis.md — intelligence displacement spiral narrative (Van Geelen/Shah Feb 2026)
  • OpenAI “Industrial Policy for the Intelligence Age” (Apr 2026) — public wealth fund proposal, not statute (TechCrunch Jul 2026)

Analogue: Post-2008 — bailouts fast, household recovery decade.

Would update if: ≥2 G7 countries enact dividend-scale transfers (>5% median income equivalent) by 2029.

Conf: M


T = 2029–2031 — C9 fiscal patchwork vs coherent settlement

Claim: C9 modal = patchwork (state UI, tax credits, voluntary corporate funds) not post-labor settlement (P coherent federal settlement 0.12–0.18 by 2031).

Why: Historical workforce programs poor uptake; Acemoglu/Johnson anti-UBI; federal gridlock baseline from Node 6.

Evidence:

  • c8 §C9 — fiscal response patchwork [EST]
  • UBI_post_work_meaning_AI.md — OpenResearch 2024: UBI ↓ labor income ~$1,500/yr; no society-scale AI-funded test
  • happy_path_with_ai_brainstorm.md §C — staged path 2028–2032 for institutions

Analogue: ACA 2010 — partial universal mechanism, decade to stabilize; not full post-scarcity settlement.

Would update if: Alaska-scale national dividend (> $2k/yr/adult, inflation-adjusted) funded by AI/compute levy.

Conf: M


Ghost GDP vs shared gains

P = 0.58 — Ghost GDP sector-visible by C6 (2028 H1 calendar)

Claim: Corporate margins ↑ and payroll flat in AI-heavy sectors become econometrically legible before economy-wide crisis — early anti-utopia signal.

Why: So-so automation: output gain small, labor share drop large (Power_and_Progress §so-so). MIT NANDA 95% zero P&L impact today — but frontier adopters (Block, Cloudflare) already cut heads.

Evidence:

  • ai_meltdown_ghost_gdp_crisis.md §1 — machines spend $0 on discretionary goods
  • c8 society snapshot by ci §C5–C6 — profits vs payroll divergence visible [SPEC]
  • Productivity-pay gap post-1979 (EPI) — institutional precedent for decoupling

Analogue: 1920s productivity ↑, farm income collapse — sector ghost output before macro.

Would update if: AI-exposed sectors show simultaneous payroll ↑ and margin ↑ through 2028 (Autor product-expansion dominates).

Conf: M

Buckets: Anti-utopia (Path D); blocks U4 if no offset.


P = 0.42 — Ghost GDP macro-binding (unemployment ≥8% + profit record) by 2029

Claim: ~42% tail/modal-fringe: Van Geelen-style demand collapse materializes — not base case, not negligible.

Why: $13T mortgage exposure to white-collar income; circular AI capex props demand until credit stress (ai_meltdown §2).

Evidence:

  • ai_meltdown_ghost_gdp_crisis.md §1 table — scenario 10.2% unemployment, S&P −38%
  • c8 §C10 — Ghost GDP Phase 4 pairs with alignment scare for crash [SPEC]
  • Yale macro null on AI exposure × unemployment (2026) — not yet binding; Canaries lead ~1–2 yr

Analogue: 2007–08 — subprime visible 2006, systemic 2008.

Would update if: National unemployment stays <5.5% through 2029 despite C8 remote workers.

Conf: L–M

Buckets: Severe whimper; may force redistribution from crisis (upward on U4 conditional).


P = 0.72 — Modal: GDP headline growth without median income lift (@C7–C8)

Claim: ~72% that national accounts show strong AI-sector growth while median household income flat absent new transfer law.

Why: Top 1% wealth share 31.7% (Fed Q3 2025, historic high); AI rents accrue to compute/data owners (research_capitalism_consumerism_inequality.md §1.2).

Evidence:

  • research_capitalism_consumerism_inequality.md — top 1% = bottom 90% wealth (2025)
  • c8 §C7 — median flat without redistribution [SPEC]
  • World Inequality Report 2026 — global top 10% = 53% income

Analogue: 1980–2020 computer revolution — productivity ↑, median wage stagnation.

Would update if: Labor share of income rises ≥2 pp in US 2027–2029 while AI capex doubles.

Conf: M–H

Buckets: Primary anti-utopia; caps U4; blocks U3 without science dividend and distribution.


P = 0.35 — Shared-gains narrative wins in media/policy (@C4–C6)

Claim: “Workers must share in AI gains” becomes bipartisan talk-track — not same as enactment.

Why: Newsom EO framing; Sanders 79% poll; Trump “American public as partner” (TechCrunch); Altman equity rhetoric.

Evidence:

  • CA EO — “Californians — not just big tech — benefit”
  • Sanders S.4825 press release — $7T fund, $1k+/person dividend at 5% yield (sanders.senate.gov)
  • happy_path_with_ai_brainstorm.md §E — competing visions explicit

Analogue: Healthcare “public option” 2009–10 — won narrative, lost law.

Would update if: No major candidate mentions AI dividend in 2028 cycle.

Conf: M

Buckets: Necessary not sufficient for U3/U4.


Acemoglu: machine usefulness vs so-so automation

P = 0.65 — So-so automation dominates modal deployment (@C4–C7)

Claim: ~65% that frontier deployment prioritizes cost-cutting substitution over machine-usefulness augmentation through C7.

Why: Tax bias labor ~25% vs capital ~5% (Acemoglu-Manera-Restrepo 2020); firm incentives favor automation over new task creation; Block/Cloudflare are headcount stories.

Evidence:

  • Power_and_Progress_Acemoglu_Johnson.md §so-so automation — half of labor share decline 1980–2016
  • research_ai_economy_article/thread1_robot_tax.md §1.3 — +4.02% employment from optimal tax reform counterfactual
  • Acemoglu 2024: AI TFP +0.05%/yr — small cake, large distributional fight

Analogue: Self-checkout rollout — adopted for labor cost, not customer experience.

Would update if: METR-exposed firms show net hiring in augmented roles 2027–2028 (Autor task-complementarity wins).

Conf: M

Buckets: U4 ceiling; U3 needs policy flip.


P = 0.28 — Pro-worker AI policy package adopted in ≥1 G7 by 2030

Claim: ~28% that Acemoglu-Autor-Johnson “Building Pro-Worker AI” menu (tax rebalance, antitrust, public health/ed AI investment) passes material law in US or EU or Japan.

Why: NBER WP 34854 (Mar 2026) — nine directions; Brookings summary — politically harder than technical.

Evidence:

  • NBER WP 34854 — pro-worker = new task-creating tech, not expertise-leveling
  • Acemoglu_vs_Autor_AI_Labor_Economics.md — race between automation and new tasks not automatic
  • EU AI Act — human oversight for HR AI (Aug 2026) is partial pro-worker, not tax rebalance

Analogue: Post-WWII GI Bill + R&D tax credit — bundle took war-window + decade.

Would update if: US payroll/capital tax symmetry bill passes committee with bipartisan score.

Conf: L–M

Buckets: U3/U4 upward driver.


P = 0.45 — Augmentation wins in product categories but not labor share

Claim: Copilot-class tools raise power-user productivity without raising median wage — machine usefulness for some.

Why: Brynjolfsson “power user premium”; c8 C4 — $100–500/mo agent stacks for elites; median sees copilot not replacement until C6.

Evidence:

  • c8 society snapshot by ci §C4 daily life
  • Autor ATM paradox — employment ↑ in restructured roles, not wages automatic
  • Anthropic >80% internal code AI-written — lab productivity, not economy-wide wage

Analogue: Excel — analyst productivity ↑, headcount ↓ in bookkeeping.

Would update if: Randomized firm-level augmentation mandates show wage pass-through >30% of productivity gain.

Conf: M

Buckets: U4 partial (modest flourishing for skilled); not U3 broad.


P = 0.55 — Acemoglu anti-UBI stance shapes Democratic menu

Claim: ~55% that major Dem distribution packages emphasize services/jobs over unconditional cash through 2030 — slowing UBI path.

Why: Power and Progress Ch.11 — UBI privatizes meaning; prefers public education, unions, antitrust.

Evidence:

  • Power_and_Progress_Acemoglu_Johnson.md §policy — explicit anti-UBI
  • UBI_post_work_meaning_AI.md — UBS > UBI for Jahoda latent functions
  • Sanders SWF = dividend, not classic UBI — hybrid

Analogue: Nordic services universalism vs US cash debates.

Would update if: Biden-successor platform centers unconditional $1k+/mo federal.

Conf: M

Buckets: U7 meaning coupling; U3 institutional golden age path.


SWF models: Norway, Alaska, Altman, compute dividend

P = 0.10 — US federal AI SWF enacted at Sanders scale (≥30% equity) by 2032

Claim: ~10% S.4825-class 50% equity tax becomes law — tail-adjacent, not zero.

Why: No cosponsors (Jun 2026); industry + antitrust opposition; constitutional takings politics; Altman counters at 5%.

Evidence:

  • S.4825 press release — 50% systemically important AI firms
  • TechCrunch Jul 2026 — OpenAI 5% voluntary framing
  • UBI_post_work_meaning_AI.md §Altman — American Equity Fund 2.5% market cap + 2.5% land (2021)

Analogue: Wealth tax Warren 2019 — agenda shift, no enactment.

Would update if: ≥10 Senate cosponsors + CBO score in 2027 session.

Conf: L

Buckets: U3 tail; high leverage if true.


P = 0.22 — Voluntary / negotiated 5–10% public stake in ≥2 frontier labs by 2029

Claim: ~22% Trump-admin or successor deal: equity slice to sovereign fund without Sanders-scale compulsion.

Why: Altman-Trump discussions reported Jun–Jul 2026; “American public as partner” rhetoric; CHIPS Act precedent for conditional federal stake.

Evidence:

  • TechCrunch, ai.cm
  • OpenAI Apr 2026 industrial policy paper — fund invests in labs, returns to citizens
  • Norway GPFG — 20% of gov revenue from fund (Investopedia SWF) — benchmark, not forecast

Analogue: TARP warrants 2008 — equity participation crisis-driven.

Would update if: Explicit admin rejection of any equity stake in writing.

Conf: L–M

Buckets: U4 modest; U3 if dividends flow to households.


P = 0.15 — Norway-style resource → SWF template copied for compute/electricity by 2032

Claim: ~15% G7 state routes compute lease royalties or electricity surcharges into permanent fund (not general revenue).

Why: Oil → GPFG worked with visible exhaustible rent; AI rent is monopoly + data, harder to classify; Gulf states may lead US.

Evidence:

  • happy_path_with_ai_brainstorm.md §C step 3 — Norway/Alaska models
  • Alaska Permanent Fund ~$88.7B, ~$1–2k/yr dividend — too small for post-labor (UBI_post_work_meaning_AI.md §exp 5)
  • thread1_robot_tax §4 — Korea 2017 subsidy trim only real enacted automation levy

Analogue: Alaska 1982 — 26 years from oil discovery to meaningful dividend culture.

Would update if: UAE/Saudi announce citizen compute dividend >$5k/yr.

Conf: L

Buckets: U3 institutional.


P = 0.18 — “Compute dividend” / universal basic compute at scale (@C6–C8)

Claim: ~18% meaningful public inference entitlement (not just rhetoric) — e.g. monthly token/GPU credits ≥ professional tier.

Why: Altman 2025 “universal basic compute” pivot; EU digital sovereignty talk; harder than cash (marginal cost, abuse).

Evidence:

  • thread1 §8.3 — Altman compute-as-right vs 2021 cash fund
  • c8 §C6 — compute access = career access bifurcation [SPEC]
  • research_ai_economy_article/thread1_robot_tax.md §5 — token/compute tax academic papers 2025

Analogue: Lifeline phone program — universal floor, not abundance.

Would update if: Federal entitlement bill for inference credits passes committee.

Conf: L–M

Buckets: U4 (access); weak U3 without cash/services.


Tax paths: token / compute / electricity / Sharing the Algorithm

P = 0.16 — Federal automation-biased tax reform (payroll ↓, capital ↑) by 2030

Claim: ~16% material reduction in labor-vs-capital tax wedge per Acemoglu-Manera-Restrepo.

Why: GOP 2025–26 extends bonus depreciation; Summers coalition anti-”robot tax”; tax reform is once per decade window.

Evidence:

  • NBER w27052 — +1.14–1.96% employment from modest reform
  • NBER w34854 §policy — symmetric tax structure
  • Node 1 P(federal binding labor standards) 0.12 — same congressional bottleneck

Analogue: 1986 Tax Reform — rare bipartisan, heavy lift.

Would update if: 2027 reconciliation includes payroll tax credit funded by capital minimum tax labeled AI/automation.

Conf: L–M

Buckets: U3/U4 multiplier.


P = 0.14 — Token / compute / electricity excise enacted in US or EU by 2028

Claim: ~14% narrow levy on inference or datacenter electricity becomes law — not full robot tax.

Why: Administrability ↑ vs robot definition problem (Summers 2017); UK electricity proposals 2026 per thread1; Mark Cuban token tax rhetoric.

Evidence:

  • thread1 §8.2 — Cuban token utilization tax
  • thread1 §9 — Manifold 37% robot tax country by end 2027 (prediction market, low reliability)
  • Gasteiger 2025 — datacenters 415 TWh (~1.5% global) — visible tax base

Analogue: Carbon border adjustments — sector tax before income tax overhaul.

Would update if: EU energy package explicitly exempts AI inference through 2028.

Conf: L

Buckets: U4 funding source for partial transfers.


P = 0.07 — Bearer-Friend Sharing the Algorithm equity tax enacted (US) by 2032

Claim: ~7% one-time in-kind equity remittance per Bearer-Friend & Polcz 2025.

Why: Novel; international competitiveness objections; definitional “AI firm” fights; industry litigation.

Evidence:

  • CJTL 17(1) — four harms: copyright, unemployment, bias, concentration
  • TaxProf Harpaz review — political rate choice dominates
  • thread1 §6.6 — equity vs transaction tax fork

Analogue: Buffett rule — decades from proposal to minimal enactment (AMT patch).

Would update if: EU adopts equity remittance in AI liability directive.

Conf: L

Buckets: U3 tail governance + distribution.


P = 0.38 — State-level AI-adjacent revenue measures (CA/NY/WA) by 2028

Claim: ~38% ≥1 large state enacts compute tax, data-center surcharge, or equity-severance mandate with enforcement.

Why: CA SB 53 ecosystem; Newsom EO may seed 2027 bills; states as labs when federal gridlocked (Node 6).

Evidence:

  • Node 1 P(CA labor AI bill) 0.50
  • SB 951 displacement notice
  • Fisher Phillips — severance/equity recommendations possible Nov 2026

Analogue: CA privacy (CCPA) → national pressure.

Would update if: GAAIA preemption without carve-outs blocks CA labor (Node 1 P 0.12).

Conf: M

Buckets: U4 patchwork federalism.


Augmentation vs automation policy

P = 0.55 — EU human-in-loop HR AI rules enforced; US voluntary (@C4–C6)

Claim: Annex III employment AI (Aug 2026) bites; US stays OSHA/HELP hearings without binding augmentation mandate.

Why: EU AI Act Art. 26(7) worker notification; US executive EOs study-only; culture-war coalition = vetting not mandates (Node 1 P 0.62).

Evidence:

Analogue: GDPR vs US sectoral privacy — Brussels effect partial.

Would update if: Federal “human final decision” statute for firing/hiring.

Conf: M

Buckets: U4 (EU); U3 if US copies.


P = 0.25 — US augmentation mandate (meaningful % tasks human-performed) by 2030

Claim: ~25% law requiring minimum human share of work units in regulated sectors — hard to define, industry fights.

Why: Hawley/Warner reporting bills easier than production quotas; Acemoglu prefers direction-shifting R&D not quotas.

Evidence:

  • Sanders moratorium conditions include anti-displacement policies (S.4214) — not augmentation %
  • happy_path_with_ai_brainstorm.md §D — “leave space” hard under market pressure
  • EU prohibits emotion recognition workplace (Feb 2025) — ban not mandate

Analogue: German co-determination — works council veto, not task %.

Would update if: AFL-CIO publishes model augmentation bill adopted by 3 states.

Conf: L

Buckets: U2 symbiosis enabler.


P = 0.70 — Market penalty for pure automation without PR offset (@C4–C5)

Claim: ~70% firms face reputational/litigation risk if AI layoffs lack transition narrative — not same as keeping jobs.

Why: Node 1 P(labs reskilling PR) 0.85; Block backlash transient; stock P(rewards layoffs) 0.65.

Evidence:

  • Node 1 actor table — frontier labs reskilling partnerships
  • Newsom EO — severance/equity review
  • research_capitalism_consumerism_inequality.md — status anxiety drives overwork, not solidarity

Analogue: ESG window dressing — talk > walk.

Would update if: Sustained consumer boycott moves >3% revenue at AI-layoff brands.

Conf: M

Buckets: Weak U4 signal.


C4 labor shock as success-branch policy window

P = 0.68 — C4 window produces labor rules not distribution settlement

Claim: Modal policy output = WARN tweaks, disclosure, EU oversight — not SWF/UBI (mirrors Node 1 modal).

Why: Node 1 stitch — vetting coalition wins; Sanders moratorium P 0.03; SWF P 0.10.

Evidence:

  • Node 1 P(HELP hearings) 0.75 + P(binding federal labor) 0.12
  • Utopia timeline §U2 — triggers same as N1
  • my putopia — concentrated harm primary competitor

Analogue: Y2K — massive compliance industry, no structural economic reform.

Would update if: SWF or equivalent passes during 2027 Q2 salience peak.

Conf: M–H

Buckets: Path D default; anti-U3/U4.


P = 0.32 — Cross-partisan layoff notice + retraining fund federal package by 2028

Claim: ~32% Warner-Hawley reporting + modest employer retraining levy — success-branch of Node 1 labor coalition.

Why: Bipartisan workforce framing easier than equity tax; Node 1 P(bipartisan reskilling bill) 0.30.

Evidence:

  • Web search Jun 2026 — Algorithmic Layoff Notice Act 90-day / 50+ employees (labor package modular)
  • Node 1 P(labor + populist vetting bridge) 0.30
  • EU FRIA for deployers — procedural model

Analogue: COBRA 1985 — narrow safety net after shock.

Would update if: 119th Congress adjourns with zero AI labor bills passed.

Conf: M

Buckets: U4 partial (managed friction).


P = 0.45 — Second-wave policy window @C6–C7 (Ghost GDP salience)

Claim: If C4 window misses redistribution, second chance ~2028–2029 when sector Ghost GDP undeniable — P 0.45 of serious federal commission → bill (not enactment).

Why: Institutions lag capability ~2 yr; c8 C8 mainstream debate; crisis may be needed.

Evidence:

  • c8 §C8 — transfer pressure ↑, enactment slow
  • Post-WWII — second New Deal partially failed 1944; GI Bill still passed in window
  • Montreal — science known 1974, treaty 1987 (13 yr); AI clock faster but analogous

Analogue: Dodd-Frank after Lehman, not after Bear Stearns.

Would update if: Material transfer law in 2027 (first window succeeded early).

Conf: M

Buckets: U3 conditional on second window.


UBI / UBS transition stages

P = 0.12 — Federal UBI (≥$12k/yr/adult) by 2032

Claim: ~12% unconditional federal floor at post-labor scale — below activist median, above zero.

Why: OpenResearch 2024 cooled enthusiasm; fiscal cost; Acemoglu opposition; no society-scale AI funding mechanism tested.

Evidence:

  • UBI_post_work_meaning_AI.md §III OpenResearch — −4.1 pp employment, −$1.5k labor income
  • Finland health gains income-mediated — not UBI magic
  • Alaska dividend insufficient magnitude for post-work

Analogue: Nixon/Family Assistance Plan 1969 — near-miss, abandoned.

Would update if: CBO scores funded UBI with compute/SWF revenue and passes Senate.

Conf: L

Buckets: U4 cash leg; U7 still needs institutions.


P = 0.28 — UBS expansion (healthcare/education/housing guarantees) in ≥1 G7 by 2030

Claim: ~28% major expansion of service universalism — closer to Acemoglu menu than Yang UBI.

Why: Sanders SWF spends on “decent standard of living” bundle; Nordic template; US ACA politics shows incremental path.

Evidence:

  • UBI_post_work_meaning_AI.md §IV — Jahoda: UBI solves manifest, not latent
  • happy_path_with_ai_brainstorm.md §C — stage 4 UBI or UBS
  • World Happiness Report 2025 — Nordic welfare → lower status anxiety (inequality note §3)

Analogue: NHS 1948 — service universalism post-shock.

Would update if: US Medicare-for-All or equivalent fails again in 2028 cycle with no substitute.

Conf: L–M

Buckets: U3 institutional golden age.


P = 0.40 — Staged transition (tax fix → narrow levy → pilot dividend → expand) starts by 2028

Claim: ~40% at least two stages of happy_path §C pipeline underway legislatively — not completed.

Why: Newsom EO + SB 951 + Sanders + Altman = stage 0–1 activity; enactment weak.

Evidence:

  • happy_path_with_ai_brainstorm.md §C steps 1–4
  • CA EO universal basic capital concepts
  • thread1 — policy momentum 2025–26 vs 2017 rejection

Analogue: ACA state waiver experiments 2011–13 before federal.

Would update if: Zero legislative activity 2027 after Block-scale events.

Conf: M

Buckets: U4 leading indicator.


P = 0.62 — UBI/UBS without meaning institutions fails U2 normative bar (@C8+)

Claim: Cash/services without Jahoda structures → anomie — blocks U2 even if U4 income met.

Why: Putnam bowling alone trend; retirement depression literature; Bostrom Deep Utopia status problem.

Evidence:

  • UBI_post_work_meaning_AI.md §IV — status zero-sum
  • my putopia — P(meaning collapse) crux; U7 node owns detail
  • happy_path_with_ai_brainstorm.md §D — community rebuild unspecified

Analogue: Lottery winners — income ↑, happiness flat/↓.

Would update if: National community-service corps at scale with >10% adult participation.

Conf: M

Buckets: U2 vs U4 split.


Competition policy + antitrust in AI economy

P = 0.40 — Effective antitrust constraint on AI stack (cloud × model × chips) by 2030

Claim: ~40% structural remedy or conduct order changes pricing/access materially — not just fines.

Why: Khan FTC mixed wins; Google search remedy narrowed; Trump admin may shift; 5–6 frontier labs still compete on capabilities.

Evidence:

  • research_ai_economy_article/thread3_platform_governance.md — neo-Brandeis record mixed
  • AI Now — Big Tech owns chips→cloud→models; divestiture discussed
  • Acemoglu w34854 — antitrust fosters pro-worker business models
  • happy_path_with_ai_brainstorm.md §C — competition + redistribution both needed (USV insight)

Analogue: AT&T breakup 1984 — decade litigation; partial opening.

Would update if: MSFT-Google-OpenAI triangle ruled illegal per se with divestiture.

Conf: L–M

Buckets: U3 multiplier on any transfer (prevents rent capture).


P = 0.55 — Modal: oligopoly persists (2–4 labs + 2 clouds) but price competition in API tier

Claim: Concentration stable; consumer/API prices fall — false abundance for compute renters, not owners.

Why: Tracker 0–2 mo capability gaps; open weights pressure inference price; capex arms race continues.

Evidence:

  • c8 §C7 — oligopolistic clusters
  • thread8 — agent coordination monopoly tail [SPECULATION]
  • H100 rent −60–80% (ai_meltdown §4) — price ↓, fixed cost concentration ↑

Analogue: Airline oligopoly — low fares, bankrupt cycles, no worker power.

Would update if: HHI for frontier training market falls below 1500 sustained.

Conf: M

Buckets: Path D; U4 if transfers decouple from monopoly rents.


P = 0.30 — Algorithmic collusion enforced in AI labor markets (hiring/wages) by 2029

Claim: ~30% major enforcement action on AI-coordinated wage suppression or surge pricing — worker-side antitrust.

Why: DOJ 2023 Amazon Project Nessie precedent; Stanford algorithmic collusion 2024; HR AI concentration.

Evidence:

  • thread3 — Nessie $1.4B internal estimate
  • EU AI Act profiling = high-risk automatic
  • research_ai_economy_article/thread3_platform_governance.md §algorithmic pricing

Analogue: Silicon Valley no-poach settlement 2014 — late fix.

Would update if: Explicit antitrust exemption for AI hiring tools.

Conf: L–M

Buckets: U4 labor protection leg.


Global South / inequality (118 countries)

P = 0.74 — 118 countries remain rule-takers on AI governance + distribution through C8

Claim: ~74% majority of UN members lack binding seat at GPAI/Bletchley-style clubs and lack compute dividend sovereignty.

Why: UN Advisory Body statistic; thread6 international coordination; capex concentrated US/Gulf/EU/China.

Evidence:

  • research_ai_economy_article/thread6_international_coordination.md — 118 countries
  • website/src/content/writing/ai-governance-landscape.md
  • World Bank 2025 — South Asia white-collar postings −20% post-ChatGPT (thread1 §9)

Analogue: WTO Uruguay Round — rules set by core; periphery adapts.

Would update if: Global South coalition wins binding compute transfer treaty.

Conf: M–H

Buckets: U3 global version fails; U4 OECD-only.


P = 0.65 — C8 remote worker undercuts Global South BPO before rich-country trades

Claim: AI remote labor arbitrage hits Philippines/India call centers before US plumbers — asymmetric harm.

Why: c8 §C8 inequality bullet; digital tasks tradable first.

Evidence:

  • c8 §C8 — developing-world BPO disrupted [SPEC]
  • thread1 — South Asia −20% postings
  • Ghost GDP — consumption collapse US-centric narrative; export economies hit via demand

Analogue: Offshoring 1990s in reverse — onshore AI replaces offshore humans.

Would update if: India/EU mandate local human quota for client-facing services.

Conf: M

Buckets: Anti-utopia global; undermines U3 narrative.


P = 0.20 — Meaningful multilateral compute/transfer fund for Global South by 2032

Claim: ~20% IAEA-analog fund with >$10B/yr real flows — not pledges.

Why: N9 success branches weak; US-China race dominates; ODA flat.

Evidence:

  • Utopia timeline §U6 — binding treaty P low
  • happy_path_with_ai_brainstorm.md Q10 — 118 countries inclusion unsolved
  • Montreal succeeded with narrow chemical; compute is strategic

Analogue: COVAX — partial, late, inequitable.

Would update if: UN AI office administers verified compute credits to LDCs.

Conf: L

Buckets: U3 necessary for global golden age.


Path families B, D, E

P = 0.58 — Path D — Modal friction (c8 default) through C8

Claim: ~58% branch mass: acceleration + inequality + institutional lag — no utopia bucket fully satisfied.

Why: c8 TL;DR; my putopia doom modal ~50–60% non-extinction overlaps Path D.

Evidence:

Analogue: 1995–2025 internet — enormous consumer surplus, inequality ↑, no post-labor settlement.

Would update if: P(shared gains) components all land high in same 24 mo window.

Conf: M

Buckets: (none) — contrast anchor.


P = 0.16 — Path B — Institutional golden age (U3/U4) without full U1

Claim: ~16% science dividend + governance + distribution wins without ASI — Hanson slow institutions path.

Why: Requires conjunction: N5/N9 partial success + U2 tax/SWF + machine usefulness — each alone low.

Evidence:

  • Utopia timeline — B maps N1,N2,N5,N9 success
  • my putopia — U3 external shorthand ~2050
  • Post-WWII analogue below — conditional on distribution

Analogue: 1947–1973 golden age — not singularity, shared wage growth.

Would update if: P(concentration) drops below 0.45 and P(UBS/SWF) > 0.35 simultaneously.

Conf: L–M

Buckets: U3 primary; U4 overlap.


P = 0.27 — Path E — Aligned-but-unequal (blocks U2/U3)

Claim: ~27% alignment sufficient for no extinction, but gains concentrate — physical symbiosis without distribution.

Why: my putopia branch table; Amodei machines of loving grace if ownership unchanged.

Evidence:

  • my putopia — aligned but unequal; P(concentration) high crux
  • OpenAI charter general labor substitute + 5% not 50% stake
  • Bostrom — control ≠ welfare equality

Analogue: Green Revolution — food ↑, landowner capture.

Would update if: Mandatory profit-sharing at AI labs constitutional in US.

Conf: M

Buckets: Anti-U2/U3; whimper channel.


P = 0.19 — Path E2 — Aligned-but-bored (U2 normative empty)

Claim: ~19% material abundance + alignment, meaning institutions fail — U7 crux.

Why: Jahoda/Bostrom; separate node but blocks U2 bucket.

Evidence:

  • my putopia — aligned but bored
  • UBI_post_work_meaning_AI.md §V — proposition 3–4
  • Graeber bullshit jobs — removing bad jobs ≠ filling meaning

Analogue: Gulf rentier youth underemployment — income without purpose.

Would update if: Mass participatory institutions scale (see U7 file).

Conf: L–M

Buckets: U2 normative failure.


Historical analogues (success-branch priors)

P(upward | analogue) = 0.55 — Post-WWII GI Bill + union + tax absorption

Claim: Institutions can convert productivity shock into broad wage growth — took ~15 yr and unique geopolitical coalition.

Why: Acemoglu Power and Progress §18–19th c UK phase 2; EPI productivity-pay sync 1948–79.

Evidence:

  • Power_and_Progress_Acemoglu_Johnson.md §institutions
  • research_capitalism_consumerism_inequality.md §1.3 — post-1979 break
  • my putopia — Post-WWII prior for U3 not U1

Analogue: GI Bill — 7.8M served; college access transformed but excluded many Black veterans in practice.

Would update if: 2027–30 policy bundle matches three of {union density ↑, top marginal tax ↑, public edu investment ↑, SWF}.

Conf: M

Buckets: U3 upward prior.


P(upward | analogue) = 0.48 — Montreal Protocol successful mitigation

Claim: Narrow chokepoint + vivid harm + industry substitution path → binding global rule — maps to CBRN screening more than distribution, but shows mitigation works.

Why: my putopia Y2K/Montreal upward on governance; U2 chokepoint = tax/incorporation harder than CFC chemistry.

Evidence:

  • Node 2 BMIA screening coalition — partial Montreal for bio
  • Montreal 1987 — 13 yr from Rowland-Molina 1974
  • happy_path_with_ai_brainstorm.md — international coordination hard

Analogue: Cap-and-trade after visible ozone hole.

Would update if: Global compute cartel treaty with citizen dividend — no movement by 2028.

Conf: M

Buckets: U5/U6 > U2; still lifts U3 if paired with transfers.


P(downward | analogue) = 0.60 — 1980–2020 computer revolution without redistribution

Claim: Same digital wave, different institutions → inequality — base rate for AI without U2 wins.

Why: Union decline 25%→10%; capital gains cuts; platform winner-take-most.

Evidence:

  • Power_and_Progress_Acemoglu_Johnson.md §1980s
  • research_capitalism_consumerism_inequality.md
  • Fed top 1% wealth record 2025

Analogue: Default for Path D.

Would update if: Labor share inflects up 3+ pp.

Conf: H

Buckets: Concentration prior 0.68–0.78 anchored here.


Concentration crux (explicit — objective)

P = 0.73 — Top decile captures ≥60% of AI-attributed surplus through C8 (modal)

Claim: Primary competitor to U3/U4 — not p(doom), but p(utopia) ceiling.

Why: Rents to compute, data, frontier equity; Ghost GDP mechanism; Acemoglu corp power quote; circular financing beneficiaries concentrated.

Evidence:

  • Sections Ghost GDP + SWF low enactment + 1980s analogue
  • thread1 §9 — top 2000 firms profits ~$4.9T (2025)
  • my putopia — gains concentrate crux [FILL]working estimate 0.73 (range 0.68–0.78)

Analogue: Oil majors 1970s — rent capture before Norway GPFG matured.

Would update if: Labor share + SWF dividends + wage growth jointly offset ≥40% of AI productivity gain to bottom 50%.

Conf: M

Buckets: Anti-U3/U4; Path E enabler.


P = 0.22 — Material broad sharing (bottom 50% gets ≥40% of AI surplus) by 2032

Claim: Success metric for Node U2 — conjunction of {tax/SWF, antitrust, augmentation, services}.

Why: Product of partial probabilities; requires beating 1980s base rate.

Evidence:

  • Inverted from concentration P 0.73 — not independent; conditional on policy window wins
  • Path B P 0.16 — similar order of magnitude
  • happy_path_with_ai_brainstorm.md — honest difficulty assessment

Analogue: New Deal subset success — not full post-scarcity.

Would update if: CBO distributional score shows bottom quintile income +20% from AI policies alone.

Conf: L–M

Buckets: U3/U4 joint success criterion.


Bucket mapping summary (@C4–C8)

CiU2 distribution state (modal)U3/U4 impactAnti-utopia signal
C4Policy window opens; study bills; no SWFU4 early frictions onlyBlock cuts; no dividend law
C5EU HR AI rules; US hearingsU4 if EU Brussels effectSector Ghost GDP visible
C6State surcharges debatedU4 compute-access floor?Median income flat
C7Patchwork retrainingU3 needs science + shareTwo-tier owner/user society
C8Transfer debate mainstream; enactment slowU3 blocked if concentration holdsRemote worker global arbitrage

Top cruxes (research priority)

#CruxWorking PFalsifierBuckets
1Concentration ≥60% surplus top decile0.73 (0.68–0.78)Labor share +2 ppU3/U4 ceiling
2So-so automation > machine usefulness0.65Net hiring in exposed firmsU4
3C4 window → labor rules not SWF0.68Federal SWF by 2027 Q4U3 tail
4Ghost GDP macro-binding by 20290.42UE <5.5% at C8U4 crisis path
5118 countries rule-takers0.74Binding South compute treatyGlobal U3
6Broad sharing by 20320.22Bottom 50% ≥40% surplusU3/U4 win
7Path D modal friction0.58Conjunction of wins(anti)
8Meaning without work (U7 overlap)0.62 fail U2 barMass civic institutionsU2

Correlation notes (do not multiply blindly)

  • Positive: SWF passage ↔ tax reform ↔ antitrust success (Path B bundle) — ρ ≈ 0.45–0.55
  • Negative: Ghost GDP macro crisis ↔ voluntary 5% equity (crisis may force deals) — ρ ≈ −0.25
  • Shared with doom: C4 labor shock trigger — same event, branching response (pause vs distribute) — see correlation_matrix_positive.md (pending)
  • Overlap discount: P(U3) + P(U4) + P(U2) not additive — my putopia §overlap

Update log

DateChange
2026-07-04Initial research pass: 42 evidence sections; web 2024–26 (Newsom EO, EU AI Act, Sanders SWF, Altman 5%, Acemoglu w34854, Bearer-Friend); repo anchors

Key P estimates (quick reference)

EstimateRangeHeadline
Concentration (top decile ≥60% surplus)0.68–0.78Modal anti-utopia
Path D modal friction0.55–0.62c8 default
Broad sharing by 20320.18–0.26U3/U4 win
Path B institutional golden age0.14–0.18Conjunction tail
Path E aligned-but-unequal0.24–0.30Alignment without share
US SWF enacted (Sanders-scale)0.08–0.12Low
Voluntary 5–10% public stake0.18–0.26Altman track
So-so automation dominates0.60–0.70Acemoglu prior
Ghost GDP sector-visible C60.52–0.64Early warning
C4 window → rules not settlement0.64–0.72Node 1 mirror

Section count: 42 evidence sections.