AI futures evidence — U5: governance wins

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

Parent: Utopia timeline
Mirror (doom failure branches): node2 (N2 CBRN), node6 (GAAIA/preemption), node9 (EU GPAI), node10 (disclosure)
Modal baseline (NOT utopia): c8 society snapshot by ci
Date: 2026-07-04
Format: Claim | Why | Evidence | Analogue | Would update if | Conf

Each section documents one probability for U5 — governance wins: screening, GPAI enforcement, verification chokepoints, and disclosure without durable training pause. Probabilities are subjective elicitation unless noted as inverted from doom nodes.

Utopia buckets: U3 (institutional golden age), U4 (modest flourishing) — enables chokepoint mitigation that lets science/physical deployment proceed with lower tail-risk.

1. Node definition — piecemeal wins, not blueprint utopia

Claim: U5 = P(by @C8, stack of governance wins — BMIA-class screening, EU GPAI enforcement, state transparency (SB 53/RAISE), cloud KYC chokepoints, early near-miss catch — materially reduces CBRN/agent tail risk without federal training pause).

Why: Doom nodes N2/N9/N10 quantify failure branches; U5 inverts the success tails that remain compatible with modal c8 (acceleration-with-friction). Scott/Ostrom frame: wins arrive piecemeal at chokepoints, not via comprehensive blueprint.

Evidence:

Utopia timeline §U5
c8 society snapshot by ci §What “modal” means — “BMIA/EU screening + SB-53 transparency; no training FLOP cap”
(internal note) — blueprint trap vs muddling through

Analogue: Montreal Protocol — phased, sector-specific, verification at production chokepoint, not economy-wide halt.

Would update if: Federal training FLOP cap with enforcement passes → reclassify as pause node (N1/N4), not U5.

Conf: H

Claim: At @C8, branch mixture for U5-relevant governance:

Branch	P	Description	Doom mirror
U5-M (modal friction)	0.52	Screening + state transparency + EU GPAI partial enforcement; race continues	N2 MODAL 0.55; N9 M1 0.58
U5-T (strong wins)	0.28	BMIA on time + GPAI bites + disclosure ~1:1 + cloud KYC operational	Inverts N2/N9/N10 tails
U5-W (weak / hollow)	0.20	GAAIA preemption guts states; BMIA delayed; hidden near-misses; EU enforcement theater	N6 preemption; N10 T10-A

Why: Modal still friction — U5-T is real but minority. Sum ≈1.0; branches correlated (BMIA success ↔ IGSC ↔ EU pressure).

Evidence:

N2 §MODAL branch P=0.55 — “BMIA/EU screening passes; RAISE/SB 53 reporting; IGSC upgrade”
N9 §2 M1 modal 0.58 — forum talk includes EU deployment duties
c8 C4–C8 — “no pause” throughout

Analogue: Y2K — majority path = expensive fixes work, narrative = “overreaction,” still real risk reduction.

Would update if: BMIA signed + first EU GPAI fine before C5 → U5-T ↑ to ≥0.35.

Conf: M

3. P = 0.38 — Composite U5 stack “good enough” @C8

Claim: P(by @C8, ≥3 of 5 load-bearing wins operational: (1) mandatory US/EU synthesis screening, (2) SB 53/RAISE enforced, (3) EU GPAI formal enforcement action, (4) cloud KYC rule for frontier training IaaS, (5) hidden:public near-miss ratio <3:1) = 0.38 (range 0.30–0.48).

Why: Joint probability discounted for correlation (~0.55 independence product × overlap factor). Not utopia — “good enough” for U3/U4 institutional golden age path B in master doc. Does not imply pause.

Evidence:

Component P’s in §5–§19 below
correlation matrix — Cluster B (screening coalition) positive correlation
N2 + N9 + N10 inverted tails

Analogue: Basel III stack — multiple partial regimes compound to material friction without stopping banking.

Would update if: Any two wins fail by 2028 while others succeed → revise correlation model.

Conf: M

Claim: Even U5-T strong wins leave society on c8 modal path: Ghost GDP, inequality, no training pause (P(no pause) 0.80–0.84 conditional on U5-T).

Why: U5 wins operate at deployment/physical/chokepoint layers; N1/N4/N9 confirm training halt P <5%. c8 explicitly lists U5 tools as modal governance, not utopia.

Evidence:

c8 society snapshot by ci §TL;DR — “acceleration with friction”
N9 §29 — P(no pause) 0.82–0.84 reinforced
N4 §modal — transparency without halt

Analogue: Post-2008 financial regulation — tighter rules, banks still operate.

Would update if: U5-T correlates with documented ≥90-day frontier training halt → revise c8 mapping.

Conf: H

5. P = 0.45 — BMIA mandatory screening by 2027-12 (invert N2)

Claim: P(US enacts BMIA / S.3741-class mandatory nucleic-acid synthesis screening by end-2027) = 0.45 — direct invert of N2 load-bearing P #2.

Why: Bipartisan Cotton–Klobuchar (Jan 29 2026); industry endorsements (Twist, IDT, Ginkgo); screendna Jun 2026 letter; Trump AI Action Plan supports screening (not eval caps). Narrower coalition than pause. Still in Commerce committee — not law yet.

Evidence:

Analogue: IGSC voluntary (2010s) → OSTP procurement (2024) → BMIA statutory ratchet.

Would update if: BMIA fails committee without substitute by Dec 2027 → ↓ to ≤0.25; signed before Dec 2027 → ↑ to ≥0.70.

Conf: M–H

6. P = 0.60 — BMIA operational by 2028-06

Claim: P(BMIA-class law signed and providers materially compliant by mid-2028) = 0.60 — adds second congressional session + EU Biotech Act pressure.

Why: N2 horizon extension; 12–18 mo implementation lag after signing; synthesis industry wants federal uniformity vs patchwork.

Evidence:

N2 §P=0.60 (2028-06) — same claim
EU Biotech Act Ch. VIII phased apply P=0.55 (N2 §EU harmonized screening)

Analogue: HIPAA — signed 1996, compliance culture 1998–2002.

Would update if: Epistemic-collapse tail (N2 Phase 2b) kills 119th Congress → ↓ to ~0.35.

Conf: M

7. P = 0.35 / 0.55 — EU Biotech Act screening (invert N2 EU branch)

Claim: P(EU adopts Biotech Act with screening intact by 2027-12) = 0.35; P(phased apply covering major providers by 2028-06) = 0.55.

Why: Dec 2025 proposal — mandatory screening, ≥50bp SOCs, function-based criteria. Trilogue slow; competitiveness framing may water down Ch. VIII. U5 success = adoption with screening, not full pause.

Evidence:

https://health.ec.europa.eu/publications/proposal-regulation-establish-measures-strengthen-unions-biotechnology-and-biomanufacturing-sectors_en
N2 §EU harmonized screening

Analogue: EU AI Act phased apply — law ≠ immediate behavior change.

Would update if: Ch. VIII stripped in trilogue → U5 EU screening ↓ to <0.15.

Conf: M

8. P = 0.50 / 0.65 — IGSC v4 function-based screening at scale

Claim: P(major IGSC protocol upgrade with function-based SOC pilots at scale by 2027-12) = 0.50; by 2028-06 = 0.65.

Why: IGSC v3.0 baseline; NIST function-based R&D; Jun 2026 screendna CEO letter. Voluntary path faster than statute — U5 piecemeal win even if BMIA slips.

Evidence:

https://genesynthesisconsortium.org/
N2 §P=0.50/0.65 IGSC v4
(internal note) §2.3

Analogue: PCI-DSS — industry standard precedes/fills statutory gaps.

Would update if: BMIA mandates function-based SOC → revise to ≥0.80.

Conf: M

9. P = 0.40 — EU GPAI formal enforcement action by 2028 (invert N9)

Claim: P(European Commission takes formal GPAI enforcement action — information request, evaluation, fine, or corrective order — against ≥1 systemic-risk provider by 2028) = 0.40.

Why: Aug 2 2026 Commission enforcement powers activate (Art. 101); obligations live since Aug 2025; Code of Practice Signatory Taskforce. Teeth = enforcement events, not training stop.

Evidence:

https://digital-strategy.ec.europa.eu/en/policies/guidelines-gpai-providers
https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-55
N9 §19 — EU AI Office P(formal action)=0.40 modal

Analogue: GDPR — first major fines 2019, law 2018.

Would update if: Digital Omnibus delays Aug 2026 enforcement → ↓; first fine before Aug 2027 → ↑ to ≥0.55.

Conf: M

10. P = 0.14 — EU GPAI enforcement binds US frontier lab (invert N9 tail)

Claim: P(≥1 US frontier lab materially constrained by EU GPAI enforcement — restructured EU offering, systemic-risk notification, adversarial testing compliance — by 2028) = 0.14.

Why: Extraterritorial market access; up to 3% turnover fines; labs can geo-fence. 14% = real Brussels effect tail, not modal — subset of §9.

Evidence:

N9 §9 P=0.14 Brussels effect
N4 §25 — EU enforcement tail-gov P=0.55 for formal action (broader)

Analogue: Apple DMA compliance — costly, no US exit.

Would update if: OpenAI/Anthropic publicly restructure EU API before Aug 2026 → ↓; documented training delay for EU compliance → ↑.

Conf: M

11. P = 0.70 / 0.75 — SB 53 + RAISE catastrophic-risk framework enforced (invert N2)

Claim: P(CA SB 53 + NY RAISE compliance prep by 2027-12 / enforcement by 2028-06) = 0.70 / 0.75.

Why: Laws already enacted; RAISE effective Jan 1 2027; Anthropic endorsed SB 53. P reflects operational compliance, not passage. Core U5 win for transparency without federal law.

Evidence:

https://www.anthropic.com/news/anthropic-is-endorsing-sb-53
https://legislation.nysenate.gov/pdf/bills/2025/a9449
N2 §P=0.70/0.75 RAISE+SB 53

Analogue: CCPA — state leads, industry adapts.

Would update if: GAAIA preemption voids duties (§13) → ↓ sharply.

Conf: H

12. P = 0.55 — SB 53 transparency with enforcement (audits, public incident cards)

Claim: P(SB 53 produces ≥2 public catastrophic-risk incident disclosures or third-party audit findings with enforceable CA AG action by 2028) = 0.55.

Why: Encode path; whistleblower + Right to Warn synergy; subset of §11 — “enforced” ≠ “meaningful public salience.” U5 success needs visible accountability.

Evidence:

N10 §4 T10-B — SB 53 + RAISE create paper trail
N4 §30 — Saunders / Right to Warn lane

Analogue: SEC enforcement — many filings, fewer high-profile actions.

Would update if: Zero SB 53 public disclosures through 2027 despite Tier 2 live → ↓ to <0.35.

Conf: M

13. P = 0.18 — GAAIA passes with preemption stripped (state-federal balance win)

Claim: P(GAAIA-class federal law passes without gutting SB 53/RAISE dev/transparency obligations) = 0.18 — success branch of N6.

Why: Compromise path (Banks/Trahan safety + Wiener/Encode strip preemption); mirrors SB 1047→SB 53 arc. Better for U5 than pure gridlock (no federal coherence) or preemption win.

Evidence:

N6 §P=0.18 stripped preemption
(internal note) §1.4 coalitions

Analogue: ACA — partial preemption fights, states retain laboratories.

Would update if: GAAIA passes with 3-year preemption intact (N6 P=0.12) → U5 state lane ↓.

Conf: M

14. P = 0.45 — Federal gridlock; states remain primary (U5-compatible)

Claim: P(no federal AI safety law by 2028; SB 53 + RAISE enforce as designed) = 0.45 — N6 modal; U5-compatible not failure.

Why: State laboratories deliver transparency wins federal gridlock blocks. Trump EO 14409 voluntary framework satisfies industry enough to kill GAAIA urgency. Piecemeal Ostrom > failed blueprint.

Evidence:

N6 §P=0.45 gridlock
EO 14409 Jun 2026 — https://www.whitehouse.gov/presidential-actions/2026/06/promoting-advanced-artificial-intelligence-innovation-and-security/

Analogue: Cannabis policy — state labs, federal lag.

Would update if: GAAIA passes with preemption → reclassify as U5-W.

Conf: M

15. P = 0.55 — Federal cloud KYC / training IaaS rule (beneficial chokepoint)

Claim: P(US binding cloud account KYC / frontier training registration rule for ≥10²⁵ FLOP-class runs by 2028) = 0.55.

Why: BIS May 2025 AI-training IaaS statement; Jan 2026 KYC certs; Fable Jun 2026 API recall precedent. Chokepoint enables traceability + recall — U5 win, not pause. Node 7 modal lever.

Evidence:

node7 compute cloud §Branch table — P(KYC)=0.55
Fable/Mythos export ban sequence (N3 §19)

Analogue: Bank KYC/AML — friction without stopping finance.

Would update if: EO 14409 explicitly blocks cloud registration mandate → ↓ to <0.35.

Conf: M

16. P = 0.12 — HEM / verified compute pilot (beneficial, pre-treaty)

Claim: P(≥1 operational HEM or remote-attestation pilot on US export-controlled training cluster by 2028, usable for incident traceability even without training cap) = 0.12.

Why: FlexHEG-class R&D; treaty-grade metering immature (N3 §21); pilot enables future coordination — U5 infrastructure win, not pause. Correlates with U6 T2 tail.

Evidence:

Node 7 §Tail T7-D P=0.08 — HEM pilot
N3 §21 — verification 3–5 yr horizon
(internal note) — prerequisite #4 compute verification

Analogue: INF Treaty inspection tech — years of R&D before political use.

Would update if: US–China guardrails publish joint HEM pilot → ↑ to ≥0.25.

Conf: L

17. P = 0.15 — N10 disclosure success: hidden:public ratio ≈1:1 (invert T10-B)

Claim: P(hidden:public Tier-2 near-miss ratio ≈1:1 through 2028 — SB 53/RAISE + BMIA reporting norms bite) = 0.15 — direct invert of N10 branch T10-B.

Why: Still minority — federal preemption fight + confidential-to-regulator channels limit public salience. Success = policymakers see true near-miss rate → faster BMIA/screening.

Evidence:

N10 §4 T10-B P=0.15
N10 §2 M10 modal suppression 0.55 (failure branch)

Analogue: SEC 8-K culture — decades to align private/public.

Would update if: ≥2 public Tier-2 disclosures with audit trail showing timely reporting (not leak) → ↑ to ≥0.30.

Conf: L–M

18. P = 0.72 — ≥1 hidden Tier-2 near-miss caught early by screening | Tier 2 live

Claim: P(at least one Tier-2-class near-miss intercepted before synthesis/order completion | Tier 2 live, 2027–2028) = 0.72 — safety success independent of public disclosure.

Why: N10 P(hidden occurrence)=0.75; IGSC + BMIA trajectory ↑ intercept rate. U5 counts catch, not headline.

Evidence:

N10 §6 P=0.75 hidden near-miss
N2 §Trigger E near-miss — intercept more likely than successful misuse

Analogue: TSA intercepts — majority never become news.

Would update if: Zero IGSC intercepts through 2027 despite expanded red-teaming → ↓.

Conf: M

19. P = 0.25 — Mandatory near-miss reporting with penalties by 2028

Claim: P(US mandatory near-miss reporting for frontier AI bio/agent incidents with enforceable civil penalties by 2028) = 0.25.

Why: BMIA may include synthesis near-miss; AI eval near-miss not in current BMIA text; RAISE threshold = catastrophic not all Tier-2. Stretch U5 win beyond modal.

Evidence:

N10 §12 P=0.25
N2 §BMIA — physical-layer focus

Analogue: OSHA near-miss — partial coverage, incomplete compliance.

Would update if: BMIA markup adds explicit AI-assist near-miss article → ↑ to ≥0.45.

Conf: L–M

20. P = 0.75+ — Trigger E (public near-miss) → BMIA momentum (conditional)

Claim: Conditional on public documented screening near-miss Trigger E, P(mandatory federal screening law by 2028) >0.75.

Why: Y2K pattern — visible near-miss punctuates Kingdon window; natsec + industry coalition pre-built (screendna). N2 Trigger E effect table.

Evidence:

N2 §P(effect) Trigger E → P(mandatory screening) 0.75+
N10 §15 — near-miss highest public Trigger E rate 0.25–0.35

Analogue: Y2K — fixes worked, mandates followed.

Would update if: Public near-miss without BMIA progress in 12 mo → falsify conditional.

Conf: M

21. Y2K analogue — successful “overreaction”

Claim: Reference class for U5: expensive chokepoint fixes that work while public narrative dismisses risk — policy still moves.

Why: N2/N10 explicitly cite Y2K; near-miss publicized → “nothingburger” framing does not prevent BMIA if elite coalition intact. U5 accepts overreaction narrative if physical risk ↓.

Evidence:

N2 §Trigger E near-miss analogue
N10 §15 Y2K pattern
(internal note)

Analogue: Y2K remediation $100B+ — no planes fell, still worth it ex post for U5 frame.

Would update if: Historical reanalysis shows Y2K spend had zero counterfactual benefit → weaken analogue.

Conf: M

22. Montreal Protocol analogue — chokepoint verification without halt

Claim: Montreal Protocol = best reference for U5 success shape: phased sector rules, production chokepoint (CFCs), verification, continued industry with reduced tail risk.

Why: Not pause on chemistry — restrict hazardous deployment pathway. Maps to synthesis screening + GPAI deployment duties, not training moratorium.

Evidence:

Master utopia doc path B — institutional golden age
N9 §10 — EU training cap stays low (0.06) — deployment-centric like chemical use rules

Analogue: Montreal 1987 — US industry initially opposed, complied, innovated substitutes.

Would update if: BMIA + EU Biotech Act jointly create verified global synthesis registry with audits → strengthen analogue fit.

Conf: M

23. Scott / Ostrom — piecemeal wins vs blueprint trap

Claim: P(comprehensive US “AI Act” blueprint fails but ≥2 piecemeal wins succeed) = 0.65 — desirable for U5, not second-best.

Why: Scott Seeing Like a State — high-modernist plans fail on local knowledge; Ostrom polycentric rules at chokepoints succeed. GAAIA blueprint fights vs BMIA narrow bill vs state SB 53 = expected pattern.

Evidence:

(internal note)
(internal note)
(internal note) §Ostrom principles
N6 gridlock 0.45 + N2 BMIA 0.45 — parallel tracks

Analogue: US environmental law — Clean Air Act pieces, not single blueprint.

Would update if: GAAIA passes intact with preemption and BMIA — blueprint wins → revise Ostrom frame downward for US.

Conf: M

24. P = 0.62 — EO 14409 voluntary framework coexists with U5 wins

Claim: P(Trump EO 14409 voluntary frontier review does not block BMIA + state transparency + EU GPAI parallel enforcement through 2028) = 0.62.

Why: EO explicitly disclaims mandatory licensing; supports cyber clearinghouse + voluntary 30-day access. Screening is compatible with deregulatory frame — different layer than training caps.

Evidence:

https://www.whitehouse.gov/presidential-actions/2026/06/promoting-advanced-artificial-intelligence-innovation-and-security/
N6 §EO-only 0.25 — subset compatible with state/U5 stack

Analogue: Voluntary cybersecurity frameworks (NIST CSF) + mandatory breach notification — layered.

Would update if: EO amended to preempt state AI safety laws → U5 ↓.

Conf: M

25. Correlation with P(no pause) — U5 wins do not imply pause

Claim: P(no effective training pause | U5-T strong wins) ≈ 0.78–0.82 — only −0.02 to −0.04 vs baseline 0.82–0.84.

Why: U5 tools are orthogonal to pause crux (N1/N4/N9). EU GPAI deployment duties may delay releases 2–4 wk (N4) without halt. GAAIA preemption ↑ no pause (+1–3pp, N6).

Evidence:

N9 §29 P(no pause) delta
N6 §p(doom) O1 preemption +1–3pp no pause
my pdoom crux #5

Analogue: Aviation safety regs — fewer crashes, no stop to flying.

Would update if: BMIA + EU GPAI jointly cause ≥90-day documented training halt at ≥2 labs → revise correlation.

Conf: M

26. U3 bucket — science dividend enablement

Claim: P(U3 institutional golden age materially easier | U5-T vs modal) = +0.12–0.18 absolute on U3 path probability — screening ↓ bio tail; transparency ↓ surprise halts; not sufficient alone.

Why: U3 bottleneck = trials/GMP/FDA (c8); U5 removes policy surprise and CBRN catastrophe that would poison biotech AI investment. Conditional uplift, not multiplier.

Evidence:

Utopia timeline §U3 — buckets U3, U4 for U5
N2 §MODAL — screening without eval mandate

Analogue: Biosafety level labs — enable research by containing tail.

Would update if: U3 evidence file quantifies joint model — calibrate uplift there.

Conf: L–M

27. U4 bucket — physical deployment enablement

Claim: P(U4 modest flourishing path ↑ | U5 cloud KYC + export-control recall precedent) = +0.08–0.12 — gov can gate deployment (Fable-class) without stopping energy/space R&D.

Why: Node 7 — concentration creates chokepoint for beneficial friction (recall bad deploy) while Stargate/energy build continues on c8 modal.

Evidence:

N3 §19 Fable precedent — deployment control live
Node 7 §E2 API recall P=0.15–0.25

Analogue: FAA airworthiness — gates deployment not physics research.

Would update if: Cloud KYC fails and Fable never repeats → ↓ U4 uplift.

Conf: L–M

28. What U5 does not deliver

Claim: U5 explicitly excludes: (a) verified training FLOP cap (P <0.06), (b) durable multilateral pause (P <0.05), (c) alignment solved, (d) inequality reversed.

Why: Scope discipline — prevents utopia collapse into “governance solved everything.” Maps to path E (pessimistic non-doom) and c8 modal if U5 only partial.

Evidence:

N9 §13 P=0.05 verified training-limit treaty
N4 §modal — no pause
Master doc anti-pattern — “if we just solve governance”

Analogue: Food safety — doesn’t end hunger.

Would update if: Scope expansion requested in Phase 3 synthesis.

Conf: H

29. Actor table — who enables U5 wins

Actor	Success menu	P(outcome)	U5 component
US Congress	BMIA pass; no preemption	0.45 / 0.18	§5, §13
EU Commission	GPAI fines; Biotech Act	0.40 / 0.35	§9, §7
CA/NY states	SB 53, RAISE enforce	0.70–0.75	§11
Synthesis industry	IGSC v4; screendna coalition	0.50–0.65	§8
Hyperscalers	Cloud KYC compliance	0.55	§15
Frontier labs	SB 53 compliance; EU GPAI cards	0.60–0.70	§10, §11
IGSC / regulators	Early intercept	0.72	§18

Conf: M

30. Falsifiers (watchlist)

Observation	U5 branch
BMIA signed + EU first GPAI fine before 2027	U5-T ↑
GAAIA preemption voids SB 53	U5-W ↑
Tier-3 attack before BMIA operational	U5 falsified for timing
Hidden:public ratio >10:1 confirmed by FOIA	T10-A; U5 disclosure fail
Federal training FLOP cap passes	Reclassify — not U5
Zero IGSC intercepts 2027 despite Tier 2	§18 ↓

Conf: M

31. Downstream priors

If U5-M modal	If U5-T strong	If U5-W weak
U3 science baseline	U3 +0.15 path ease	U3 blocked by scandal/halt
U6 multilateral talk + EU enforce	U6 partial deployment ↑	U6 empty-shell ↑
N2 Tier 3 before screening P=0.20	N2 Tier 3 before screening ↓ 0.12	N2 Tier 3 before screening ↑ 0.28

Conf: M

Update log

Date	Change
2026-07-04	Initial research pass — 32 sections; inverted N2/N6/N9/N10 success branches

1. Node definition — piecemeal wins, not blueprint utopia

2. Branch overview — U5 success vs modal friction

3. P = 0.38 — Composite U5 stack “good enough” @C8

4. Cross-ref c8 — modal is still acceleration-with-friction

5. P = 0.45 — BMIA mandatory screening by 2027-12 (invert N2)

6. P = 0.60 — BMIA operational by 2028-06

7. P = 0.35 / 0.55 — EU Biotech Act screening (invert N2 EU branch)

8. P = 0.50 / 0.65 — IGSC v4 function-based screening at scale

9. P = 0.40 — EU GPAI formal enforcement action by 2028 (invert N9)

10. P = 0.14 — EU GPAI enforcement binds US frontier lab (invert N9 tail)

11. P = 0.70 / 0.75 — SB 53 + RAISE catastrophic-risk framework enforced (invert N2)

12. P = 0.55 — SB 53 transparency with enforcement (audits, public incident cards)

13. P = 0.18 — GAAIA passes with preemption stripped (state-federal balance win)

14. P = 0.45 — Federal gridlock; states remain primary (U5-compatible)

15. P = 0.55 — Federal cloud KYC / training IaaS rule (beneficial chokepoint)

16. P = 0.12 — HEM / verified compute pilot (beneficial, pre-treaty)

17. P = 0.15 — N10 disclosure success: hidden:public ratio ≈1:1 (invert T10-B)

18. P = 0.72 — ≥1 hidden Tier-2 near-miss caught early by screening | Tier 2 live

19. P = 0.25 — Mandatory near-miss reporting with penalties by 2028

20. P = 0.75+ — Trigger E (public near-miss) → BMIA momentum (conditional)

21. Y2K analogue — successful “overreaction”

22. Montreal Protocol analogue — chokepoint verification without halt

23. Scott / Ostrom — piecemeal wins vs blueprint trap

24. P = 0.62 — EO 14409 voluntary framework coexists with U5 wins

25. Correlation with P(no pause) — U5 wins do not imply pause

26. U3 bucket — science dividend enablement

27. U4 bucket — physical deployment enablement

28. What U5 does not deliver

29. Actor table — who enables U5 wins

30. Falsifiers (watchlist)

31. Downstream priors

32. Related files

Update log