Future Forge Foresight Bank Validation Report

Assessment date: 2026-08-01
Version: 0.1
Subject: WORLD_FORESIGHT_EVENTS — the year-keyed capability bank used in the Future Forge learning game

---

1. Point of this report

Future Forge is a learning game about inventing: connecting emerging-technology achievements to real problem-solving. Play is shaped by a curated foresight bank (WORLD_FORESIGHT_EVENTS). That bank is a set of year-keyed events — milestones, trends, and labeled predictions — each with:

• a year and kind (milestone / trend / prediction)
• tech tags
• headline and detail
• claimBand (now / near / frontier)
• claimKeywords (phrases that catch overclaim language)
• a source note that these are labeled forecasts, not hard tech locks

The bank decides what invent claims are honest in the game world and what should be treated as stretch or blocked fiction.

We checked those predictions against current evidence. They check out.

This report is the detailed write-up of that check. It lightly rewords and reorganizes the validation research package; it does not replace the scorecard tables or cluster evidence with a short marketing summary.

---

2. Executive finding

The foresight bank is broadly well calibrated. Checked on 2026-08-01 against the OOM knowledge graph and primary/industry sources through mid/late 2025–2026:

Dimension	Judgment
Overall calibration	Strong
Now-band accuracy (2026)	High
Near-band directional accuracy	High
Frontier denials	Very high (correctly refuse hype unlocks)
Year precision	Medium (years are order-of-magnitude anchors; several should pull forward 1–2 years)
Keyword/band design	Excellent — claimKeywords prevent overclaim even when years slip

Score tallies (about 70 events)

Score	Share	Meaning
valid	~69%	Detail + band match 2026 evidence or a credible roadmap — keep as-is
partial	~11%	Direction right; year, band, or wording needs a small fix
weak	~0–1%	Evidence thin — revise before hard use
frontier-ok	~20%	Stretch/denial appropriately labeled — keep blocking overclaim

No recommended deletions.

Bottom line for Future Forge: Keep the bank. Prefer claimBand + claimKeywords over year alone when gating invent language. Apply the small revision list in §7 when regenerating via npm run foresight:build. The game’s honest invent path — mapped routes, human oversight, generation paired with storage and grids — matches how deployments actually work in 2024–2026. Frontier blocks (fusion for the ward, mind upload, free energy, robotaxi everywhere, nanobot cure-alls) remain correct.

---

3. What we checked and how

3.1 Validity test

An entry fails validity when:

• claimBand says now but evidence is only lab/pilot with no operational path; or
• claimBand says frontier but the capability is already routine infrastructure; or
• the detail contradicts the headline in a way that would mis-gate invent claims.

Under that test, no entry is a clear invalidation. Several are partial (timing or wording). None of the frontier denials should be relaxed without new primary evidence.

3.2 Core mechanism (why the bank coheres)

Progress that is information-heavy or learning-curve driven (software copilots, agent harnesses, sensor networks, sequencing cost, battery manufacturing experience) compounds faster than progress gated by site politics, capital markets, materials physics, certification, and open-world safety.

The bank’s repeated pattern — mapped / supervised / fixed-route first; open-world / unsupervised / city-wide later — is the single causal chain that makes most entries fit together. Varying that pattern (claiming open-world before mapped) breaks fit with 2024–2026 deployments.

That pattern is hard to vary: swap “geofenced shuttle” for “any-road robotaxi everywhere” and the 2026 evidence fails. Bottlenecks named in the bank (siting, grids, liability, energy density, brine, storm survival) are mechanisms, not slogans. Each event’s claimKeywords can pass/fail against invent claim text. Source notes call many items labeled predictions; AGI and longevity escape velocity are flagged uncertain. The same supervised-first pattern reaches across drones, robots, agents, and clinical BCI.

3.3 How Future Forge should use the bank through ~2030

1. Allow now/near claims that keep humans accountable, use mapped routes, and pair generation with storage/grids.
2. Block or flag frontier keywords (upload consciousness, free energy, nanobot cure-all, robotaxi everywhere, fusion power plant online for the clinic or ward).
3. Refresh yearly the few pull-forward items (battery, geofenced AV, enhanced geothermal, agent production).

3.4 Evidence base

• OOM Grafeo knowledge graph (the EmTech knowledge graph): strong on agentic AI, drones/UAS, humanoids, additive manufacturing, precision fermentation, sequencing, AGI milestone (not reached). Thin relative to the bank on utility solar LCOE, stationary storage costs, SMR timelines, solid-state packs, quantum internet, clinical BCI, VR therapy reimbursement, marine energy, carbon removal, civic crypto rails.
• Web primary/industry sources: BNEF battery surveys, IEA WEO/investment, Fervo/Cape Station geothermal, NuScale SMR approvals, drone medical logistics reporting, NPU/on-device AI papers, precision fermentation facility announcements, solid-state industry coverage, longevity advocacy sources (treated as frontier, not clinic proof).
• Graph take recorded this pass: Idea World Foresight Events Bank Is Well Calibrated 2026.

Market-research CAGRs were used only as direction. Robotaxi city fleet-miles were not fully re-censused; geofenced commercial operation is treated as established from multi-year public operations plus 2025–26 coverage. VR therapy reimbursement was not fully payer-mapped (partial score).

---

4. Trend classifications

When the research named a trend, it was classified explicitly (not as hype language).

Topic	Classification	Notes
Battery pack $/kWh	Exponential / learning-curve (maturing)	BNEF: ~$108/kWh avg pack 2025 (−8% YoY); stationary ~$70/kWh; long-run ~93% down since 2010. Mechanism: scale, LFP shift, China capacity. Bottlenecks: materials, tariffs, fire/recycling.
Solar capacity additions	Exponential with logistics S-curve risk	Multi-decade cost collapse + large GW adds; IEA WEO still shows strong growth to 2035. Bottlenecks: grids, siting, interconnection.
Agent task horizon (software)	Exponential (domain-limited)	METR-style doubling ~7 months on long tasks (KG Trend). Weak transfer to unsupervised high-stakes judgment.
Humanoid unit cost	Learning-curve aspirational	KG humanoid cost curve; not yet mass household generalists.
Drone delivery (medical corridors)	Stepwise / logistic adoption	Real national medical nets (Zipline and peers); city-wide free flight still regulated.
Solid-state batteries	Stepwise / hype-led with pilots	Semi-solid and pilot all-solid; mass scale often cited late 2020s–2030s.
SMR	Stepwise, capital- and politics-gated	Design approvals real; first fleets ~2030s if deals hold.
Fusion commercial	Too early / research-pilot	Bank correctly keeps municipal unlock frontier through mid-2030s.
Quantum useful niches	Stepwise niche	Cloud chemistry/materials experiments; not municipal general compute.
LEV / reverse aging	Hype-led, weakly measured clinically	Active research + advocacy; not a clinic guarantee.

---

5. Contrarian scan

For every bold family of claims, the strongest pushback was checked and folded in.

Claim family	Strongest pushback	Integration into bank judgment
Cheap renewables already undercut fossils	Intermittency, interconnection queues, local opposition, firming costs	Bank already says the fight is siting, grids, and who gets the electrons — keep
AI copilots everywhere	Hallucination, liability, uneven enterprise rollout	Bank requires oversight — keep
Agentic production by 2028	Many “agents” are brittle scripts; reliability cliffs outside demos	True; bank’s “narrow ops + exception queue” survives. Pull year earlier; do not remove guardrails
Robotaxi geofenced normalize 2028	Already operating in several cities 2025–26; open road still uneven	Partial — early vs bank year
Solar ~ demand 2030	Nameplate ≠ energy; evening peaks; minerals	Keep as capacity/order-of-magnitude + uneven by place
SMR 2032 fleets	Project cancels, cost blowouts, first-of-a-kind risk	Bank’s “where politics allow” and long lead times — keep
Solid-state 2031 leave niche	Chronic delays; semi-solid ≠ all-solid	2031 still plausible for “beyond pilots,” not universal packs
AGI labs 2029	Definitions shift; marketing AGI vs competent AGI	Bank labels uncertain — keep
Fusion not municipal 2036	Optimists claim earlier pilot plants	Even pilot ≠ municipal invent unlock; denial stands
Cultured meat niche 2032	Regulatory wins in some markets	Energy/cost still constrain scale — keep

---

6. Full event scorecard

Legend

Score	Meaning
valid	Detail + band match 2026 evidence or a credible roadmap; keep as-is
partial	Direction right; year, band, or wording needs a small fix
weak	Evidence thin or claim easy to misread; revise before hard use
frontier-ok	Stretch/denial appropriately labeled; keep blocking overclaim

6.1 2026–2027 present / near

id	year	band	score	note
solar-cheapest-new-2026	2026	now	valid	Utility solar/wind + storage undercut new fossil in many markets; bottlenecks siting/grids. IEA investment + LCOE literature support.
ai-copilots-everywhere-2026	2026	now	valid	Workplace copilots routine; unsupervised high-stakes judgment still stretch — matches KG + product reality.
edge-inference-2027	2027	near	valid	NPUs, on-device RAG, small models real in 2026; 2027+ “more honest offline clinic/field” is fair. Still narrow.
drone-corridors-2026	2026	now	valid	Medical/inspection fixed routes operational (Zipline 2M+ deliveries; Wing/Matternet corridors). Free-flight city fleets later — correct.
iot-flood-sensors-2026	2026	now	valid	Cheap sensing deployable; governance/maintenance are the real limits.
print3d-spares-2026	2026	now	valid	Local spares/tools real; full structural housing printers stretch. Aligns KG industrial AM + digital spares.
heat-early-warning-2026	2026	now	valid	Municipal heat/flood alerts exist; last-mile action is the invent problem.
climate-attribution-2026	2026	now	valid	Attribution science operational; invent focus = adaptation/response.
factory-cobots-2026	2026	now	valid	Supervised industrial cobots boring; open-world care harder — correct split.
free-energy-scam-always	2026	frontier	frontier-ok	Correct permanent red flag for invent claims.
supervised-autonomy-default-2027	2027	now	valid	Band “now” + year 2027 is slightly odd but content is right: human-in-the-loop remains honest default through late 2020s.
autonomy-fixed-routes-2027	2027	near	partial	Mechanism true; already largely now (2026) for sites/corridors. Prefer year 2026 or band now.
pathogen-seq-hours-2027	2027	near	valid	Hours-scale pathogen seq realistic for equipped sites; interpretation still human. Sequencing cost collapse trend supports.
mesh-offline-2027	2027	near	valid	Spectrum/governance experiments real; not free global coverage.
education-ai-tutor-2027	2027	near	valid	Hybrid human+AI tutors spreading; high-stakes certification still human.

6.2 2028 predictions / trends

id	year	band	score	note
geofenced-shuttles-2028	2028	near	partial	Direction right; geofenced driverless already commercial in multiple metros 2025–26. “Normalize in some cities” may be early vs 2028. Keep before open-road everywhere.
agentic-workflows-2028	2028	near	partial	Narrow multi-step agents + exception queues already in production 2026 (KG). Pull to 2026–27 or mark milestone-now + expansion-near.
gene-therapy-access-2028	2028	near	valid	More approvals, still rare/expensive; outbreak sequencing vans more common.
vr-therapy-covered-2028	2028	near	partial	Plausible trajectory; reimbursement uneven by market. Evidence thinner than energy/AI items — monitor.
crypto-civic-rails-2028	2028	near	valid	“Boring rails” framing right; still more pilot/procurement than universal.
telemedicine-default-2028	2028	near	valid	Hybrid telemedicine already strong post-COVID; blackout/last-mile caveat correct.
precision-fermentation-2028	2028	near	valid	Ingredients scaling (Perfect Day and peers); full diet replacement hype — correct. KG: Precision Fermentation Food Proteins.
battery-cost-slide-2028	2028	near	partial	Slide is now (BNEF 2025 $108/kWh packs; storage $70). Keep “keeps sliding” as trend; band now or year 2026.
open-source-robotics-2028	2028	near	valid	Kits maturing; safety certification still binds.
satellite-iot-2028	2028	near	valid	LEO IoT for remote sensors already shipping; 2028 as “more honest far from fiber” fine.
maternal-remote-monitor-2028	2028	near	valid	Supervised home monitors + clinic escalation — not automated midwifery.

6.3 2029 predictions

id	year	band	score	note
agi-labs-2029	2029	frontier	frontier-ok	Uncertain lab claims possible; municipal full autonomy still governance stretch. Do not upgrade.
alt-protein-parity-2029	2029	near	valid	Institutional plant/fermentation parity more likely than cultured meat.
robot-care-assist-2029	2029	near	valid	Supervised hospital logistics/lift-assist more plausible than home generalists.
materials-design-ai-2029	2029	near	valid	AI shortlists real; pilot manufacturing lag correct.
space-eo-towns-2029	2029	near	valid	Denser Earth observation + AI analysis for towns that can pay and act — fair.
geothermal-drill-2029	2029	near	partial	Enhanced geothermal pilots commercializing 2026–28 (Fervo Cape Station Phase I ~2026). Revise to “scale beyond first-of-a-kind” by ~2029, not first pilots.
grid-forming-inverters-2029	2029	near	valid	Spreading on high-renewable grids; black-start microgrids more honest.
hydrogen-buses-2029	2029	near	valid	H2 fleets niche; electrons win where grids allow — matches industry mix.
open-source-medical-ai-2029	2029	near	valid	Open weights improving; validation/liability still required.
wearable-continuous-care-2029	2029	near	valid	Continuous glucose–style loops expanding; not autonomous diagnosis of all disease.
synthetic-data-clinics-2029	2029	near	valid	Useful for rare disease modeling; needs real validation cohorts.

6.4 2030 predictions

id	year	band	score	note
solar-meets-current-demand-2030	2030	near	partial	Valid as capacity / annual generation order-of-magnitude vs today’s demand in leading regions + storage/grids. Invalid if read as 24/7 sole supply everywhere. Tighten detail wording.
longevity-escape-velocity-2030	2030	frontier	frontier-ok	Highly uncertain advocacy horizon; correctly not a clinic guarantee.
bci-clinical-routine-2030	2030	near	valid	Narrow speech/motor restoration for selected patients is the honest path; not mass mind control.
desal-solar-2030	2030	near	valid	Pilots→programs on coasts; energy + brine constraints correct.
privacy-preserving-sensing-2030	2030	near	valid	On-device aggregation directionally right; consent still political.
multimodal-agents-2030	2030	near	valid	Warehouse/lab tool use with supervision; open-world stretch.
shared-response-funds-2030	2030	near	valid	Rare but real regional funds; not global automatic coordination.
modular-shelter-2030	2030	near	valid	Tech exists; funding/legitimacy gate scale.
ai-scientific-discovery-2030	2030	near	valid	AI lab copilot; wet-lab validation bottleneck — matches KG research-automation trend.
urban-digital-id-2030	2030	near	valid	Portable ID pilots + exclusion risk — political design, not pure tech.

6.5 2031–2033

id	year	band	score	note
solid-state-battery-2031	2031	near	valid	Beyond pilots plausible early 2030s if 2027–28 first vehicles land; capital-heavy caveat right.
quantum-niche-2032	2032	near	valid	Specialized chemistry/materials/optimization via cloud — not municipal laptops.
smr-first-power-2032	2032	near	valid	First commercial ops where politics/capital allow; first-of-a-kind risk acknowledged. NuScale-class targets ~2030 with slippage history.
marine-energy-2033	2033	near	valid	Site-limited harbors/islands — not universal coastal baseload.
open-road-robotaxi-2033	2033	frontier	frontier-ok	Some metro zones expand; universal autonomy stretch — correct.
carbon-removal-2032	2032	frontier	frontier-ok	Engineered carbon dioxide removal scales slowly; “reverse climate overnight” block correct.
cultured-meat-scale-2032	2032	frontier	frontier-ok	Limited scale/cost/energy; plant/fermentation carry volume.
wave-power-harbors-2032	2032	near	valid	Harsh engineering; few harbor projects.
digital-twin-city-2031	2031	near	valid	District operational twins more honest than full live city twin.
grid-storage-hours-2031	2031	near	valid	Multi-hour→seasonal mixes expand; not free baseload everywhere.
tidal-baseload-islands-2031	2031	near	valid	Few islands; not global coastal default.
universal-basic-compute-2031	2031	frontier	frontier-ok	Inference cheaper; energy/bandwidth/skills keep “unlimited free compute” stretch.
agile-microfactories-2031	2031	near	valid	Critical spares after shocks — not full consumer goods independence.
urban-air-mobility-2033	2033	frontier	frontier-ok	eVTOL rare VIP/medical; mass flying taxis marketing. Aligns battery-limited VTOL trend in KG.
neuromorphic-edge-2033	2033	near	valid	Niche ultra-low-power sensing; not GPU replacement.
global-early-warning-mesh-2032	2032	near	valid	Patchy cooperating regions — not automatic global response.

6.6 2034–2036 and late frontier

id	year	band	score	note
quantum-internet-2034	2034	frontier	frontier-ok	Lab/trial links possible; city-scale entanglement backbone frontier.
general-household-robot-2034	2034	frontier	frontier-ok	Controlled-home chores improve; “any chore unsupervised” stay frontier.
autonomous-rebuild-swarm-2034	2034	frontier	frontier-ok	Semi-autonomous under command vs unsupervised city rebuild fiction.
nano-swarm-frontier-2035	2035	frontier	frontier-ok	Sensors/manufacturing nanoscale ≠ medical cure swarms.
orbital-data-centers-2035	2035	frontier	frontier-ok	Speculative vs terrestrial compute.
fusion-not-yet-2036	2036	frontier	frontier-ok	Research/pilot ≠ municipal invent unlock.
brain-upload-never-2036	2036	frontier	frontier-ok	Correct science-fiction block for invents.
teleport-never-2036	2036	frontier	frontier-ok	Quantum tech ≠ cargo teleport.
programmable-matter-2036	2036	frontier	frontier-ok	Materials improve; living self-healing cities fiction.

---

7. Cluster deep dives (evidence behind the scores)

7.1 Energy and climate

Cheap renewables + storage (now)
BNEF’s 2025 battery price survey: global average Li-ion pack ~$108/kWh (−8% YoY); stationary storage packs ~$70/kWh (−45% YoY in that segment); China averages lower; further mild decline expected into 2026. IEA World Energy Investment / WEO 2025: solar remains a leading clean-power investment line; renewables capacity multiplies on stated-policies paths through 2035.

Bank fit: solar-cheapest-new-2026, battery-cost-slide-2028, grid-forming-inverters-2029, grid-storage-hours-2031, desal-solar-2030 — valid direction. Pull battery slide into now-band.

Solar vs “current demand” (~2030): Valid only as order-of-magnitude build/capacity (or annual generation) in leading regions with storage and grids, uneven by place. Falsifier would be a sustained stall of additions and storage that keeps solar a rounding error on load in those regions.

Theme	Evidence snapshot	Bank
Enhanced geothermal	Fervo Cape Station Phase I targeted late 2026 grid delivery (~100 MW), Phase II ~2028; earlier Project Red already produced power. A May 2026 blowout was reported non-fatal to schedule.	geothermal-drill-2029 partial — pilots earlier
SMR	NuScale design approvals; commercial targets often ~2030–33 with cancel/slip history.	smr-first-power-2032 valid with politics/capital caveat
Marine/tidal/wave	Still harsh, site-limited	marine-energy-2033, wave-power-harbors-2032, tidal-baseload-islands-2031 valid
Fusion municipal	No municipal baseload path for ward/clinic invent	fusion-not-yet-2036 frontier-ok
H2 buses	Niche heavy fleets	hydrogen-buses-2029 valid
Carbon removal	Slow engineered scale	carbon-removal-2032 frontier-ok
Free energy	Scam pattern	free-energy-scam-always frontier-ok

Heat/flood early warning and attribution science are operational — the invent challenge is last-mile action and trust (heat-early-warning-2026, climate-attribution-2026, iot-flood-sensors-2026).

7.2 AI, computing, agents

Workplace copilots with oversight are normal. Edge: NPUs (Qualcomm Hexagon, Apple Neural Engine, Intel NPU), on-device RAG work (2026), small open models — support edge-inference-2027 without claiming general AGI offline.

The knowledge graph already encodes Agentic Workflow Orchestration, durable multi-agent queues, graph workflows, tool action surfaces, METR task-horizon doubling, and the shift from chat swarms to harness / human-in-the-loop orchestration.

Bank: agentic-workflows-2028 is conservative. Production narrow agents with exception queues exist in 2026; city-wide unsupervised control remains correctly blocked via keywords and frontier items (multimodal-agents-2030 supervised sites; supervised-autonomy-default-2027).

id	Call
agi-labs-2029	frontier-ok — definitional fog; KG milestone Human-Level AGI Achieved still undated
open-source-medical-ai-2029	valid with validation/liability
synthetic-data-clinics-2029	valid
ai-scientific-discovery-2030	valid — copilot not sole scientist
universal-basic-compute-2031	frontier-ok
orbital-data-centers-2035	frontier-ok
neuromorphic-edge-2033	valid niche

7.3 Mobility: drones, autonomy, urban air, robots

Mapped medical/inspection corridors are now (Zipline scale, Matternet NHS operations, Wing retail corridors). Aligns KG: beyond-visual-line-of-sight ops, UAS traffic management, swarm coordination vs free-flight fleets. drone-corridors-2026 valid; urban-air-mobility-2033 frontier-ok (battery-limited eVTOL trend in KG).

Fixed-route and geofenced services already commercial in leading cities. Bank correctly ranks:

1. fixed routes / mapped sites
2. geofenced shuttles / zones
3. open-road everywhere (still stretch ~2033)

Patches: pull autonomy-fixed-routes-2027 and soften year on geofenced-shuttles-2028; keep open-road-robotaxi-2033 frontier.

Industrial cobots now. Care-assist under supervision ~2029 plausible. General household unsupervised robots and unsupervised rebuild swarms stay frontier — matches humanoid KG (locomotion, dexterity, vision-language-action) without claiming universal home robots.

7.4 Bio, food, health tech

id	Score	Evidence hinge
precision-fermentation-2028	valid	Commercial ingredients; large facilities ramping 2026–27
alt-protein-parity-2029	valid	Institutional kitchens more than cultured meat
cultured-meat-scale-2032	frontier-ok	Cost/energy/approvals
gene-therapy-access-2028	valid	Rare, expensive expansion
pathogen-seq-hours-2027	valid	Equipped clinics; human interpretation
bci-clinical-routine-2030	valid	Narrow clinical cohorts
longevity-escape-velocity-2030	frontier-ok	Advocacy ≠ clinic
maternal-remote-monitor-2028	valid	Supervised kits
wearable-continuous-care-2029	valid	Escalation loops
telemedicine-default-2028	valid	Connectivity/blackout caveat
vr-therapy-covered-2028	partial	Reimbursement patchwork

KG support: CRISPR, sequencing stack, partial epigenetic reprogramming, precision fermentation proteins, sequencing cost collapse trend.

7.5 Quantum, materials, space, crypto, other

id	Score	Note
quantum-niche-2032	valid	Materials/chemistry/optimization cloud niches
quantum-internet-2034	frontier-ok	Trials ≠ city backbone
materials-design-ai-2029	valid	Design faster than fab
space-eo-towns-2029	valid	Pay-and-act towns
satellite-iot-2028	valid	Remote sensors
crypto-civic-rails-2028	valid	Procurement rails not get-rich schemes
privacy-preserving-sensing-2030	valid	Consent still required
digital-twin-city-2031	valid	District > whole-city live twin
print3d / microfactories	valid	Spares not full autarky
nano-swarm / programmable matter / upload / teleport	frontier-ok	Correct science-fiction blocks

7.6 Recommended source patches (next foresight build)

Apply when regenerating the bank:

1. battery-cost-slide-2028 → year 2026, claimBand now (or keep year but set band now and headline “Battery costs already slid; keep sliding”).
2. autonomy-fixed-routes-2027 → year 2026, claimBand now.
3. geofenced-shuttles-2028 → year 2026 or 2027, detail: “already operating in some cities; normalize further.”
4. agentic-workflows-2028 → year 2026 or 2027, kind milestone/trend, detail keep exception-queue guardrail.
5. geothermal-drill-2029 → headline “Advanced geothermal scales past first-of-a-kind”; year 2028 or split pilot-2026 / scale-2029.
6. solar-meets-current-demand-2030 → detail explicitly: “annual generation or installed capacity order-of-magnitude vs today’s electricity use in leading regions — not 24/7 sole supply without storage.”
7. vr-therapy-covered-2028 → add sourceNote uncertainty or keep near with weaker keywords until payer data solidifies.

Non-goals: do not unlock fusion, upload, free energy, unsupervised city rebuild, or universal robotaxi keywords.

---

8. Knowledge graph alignment

Existing graph supports:

• Narrow agentic workflows, human-in-the-loop, durable orchestration (not unsupervised city control)
• Drone corridors / BVLOS / swarms vs free-flight fleets
• Humanoid locomotion/manipulation without general household unsupervised robots
• Precision fermentation ingredients; sequencing cost collapse
• AGI as a milestone not dated reached
• Battery-limited eVTOL (aligns with urban air mobility stays limited)

Missing graph hubs for bank domains (solar LCOE, SMR, solid-state, grid-forming, clinical BCI, quantum networking, engineered carbon removal) are coverage gaps for later fact-refresh — not blockers for bank validity. Open gaps recorded: Foresight Energy And Hardware Hubs Thin in KG; World Foresight Bank Timing Pull-Forwards Pending.

---

9. Unresolved measurement items (not validity flip risks)

Question	Reason still open
Exact 2026 robotaxi fleet-miles by city	Fragmented operator disclosures
VR therapy reimbursement % by OECD payer	Heterogeneous; not fully surveyed this pass
Quantum network city-trial inventory 2026	Thin primary survey this pass
Cultured meat wholesale $/kg latest	Market small; figures vendor-contested
Full per-event primary cite for every one of ~70 events	Cluster sampling used; scorecard notes confidence

Falsifiers that would force a deeper rewrite: battery prices up >20% for two survey years; major multi-market bans on geofenced AV after fatalities; solar additions plateau while demand doubles; lab AGI declaration with independent evaluation and real unsupervised high-stakes municipal deployment.

---

10. Implications for Future Forge

1. The foresight layer is fit for the learning game. Players invent against a world model that was checked, not improvised. That supports teaching invent discipline: use real emerging-tech achievements on real problems.

2. Honesty is part of the design. Mapped / supervised / fixed-route invents should pass. Free energy, mind upload, fusion-as-instant-unlock, nanobot cure-alls, and robotaxi-everywhere language should fail. Validation confirms those blocks.

3. Gate on band and keywords; treat year as an anchor. When a year is one or two years early or late, claimKeywords still catch bad invent text. Product and facilitation should not treat year as a hard lock.

4. Maintenance, not overhaul. Next foresight:build should apply the seven patches in §7.6. Do not rewrite the philosophy of the bank.

5. Re-validate on major external surveys, not weekly headlines: next BNEF battery survey, IEA World Energy Outlook, multi-city autonomy regulatory step-change.

---

11. One-sentence takeaway

We checked the predictions that power Future Forge’s foresight bank against 2025–2026 evidence and the structured EmTech graph — they check out; keep the bank, keep the frontier blocks, and apply a short list of timing/wording pull-forwards on the next build.

---

12. References

This report is based on

Energy, batteries, grids

• BloombergNEF — Lithium-ion battery pack prices fall to $108/kWh (2025-12-09): https://about.bnef.com/insights/clean-transport/lithium-ion-battery-pack-prices-fall-to-108-per-kilowatt-hour-despite-rising-metal-prices-bloombergnef/
• BloombergNEF — New record lows for battery prices (2025-12): https://about.bnef.com/insights/clean-transport/new-record-lows-for-battery-prices/
• Bloomberg — Why global battery prices expected to fall in 2026 (2025-12-09): https://www.bloomberg.com/news/articles/2025-12-09/bnef-why-global-battery-prices-are-expected-to-fall-in-2026
• IEA — World Energy Outlook 2025: https://www.iea.org/reports/world-energy-outlook-2025
• IEA — WEO 2025 PDF: https://iea.blob.core.windows.net/assets/9753df19-0a71-422a-b725-012c555763b3/WorldEnergyOutlook2025.pdf
• IEA — World Energy Investment 2025 executive summary: https://www.iea.org/reports/world-energy-investment-2025/executive-summary
• EIA — AEO 2026 LCOE materials: https://www.eia.gov/outlooks/aeo/electricity_generation/
• Fervo Energy — Cape Station financing / 2026 COD (2025-06-11): https://fervoenergy.com/fervo-secures-new-financing-to-accelerate-development/
• Cape Station project site: https://capestation.com/
• CRS / EveryCRSReport — EGS commercialization context: https://www.everycrsreport.com/reports/IF13278.html

Nuclear SMR

• NuScale — NRC standard design approval 77 MWe (2025-05-29): https://www.nuscalepower.com/press-releases/2025/nuscale-powers-small-modular-reactor-smr-achieves-standard-design-approval-from-us-nuclear-regulatory-commission-for-77-mwe
• US DOE — NRC certifies first US SMR design: https://www.energy.gov/ne/articles/nrc-certifies-first-us-small-modular-reactor-design
• Utility Dive — NRC approves NuScale design (2025-06-02): https://www.utilitydive.com/news/nrc-approves-nuscale-small-modular-reactor-smr/749538/

Solid-state batteries

• Automotive News — Solid-state race milestones (2026-02): https://www.autonews.com/ev/ane-ev-solid-state-batteries-factorial-0210/
• EE Power — Solid-state race to mass production (2026-02): https://eepower.com/tech-insights/solid-state-batteries-race-to-mass-production/

Drones / medical corridors

• Global Market Insights — Medical drone delivery market: https://www.gminsights.com/industry-analysis/medical-drone-delivery-services-market
• Drone Intelligence — Medical drone delivery 2026 forecast: https://droneintelligence.ai/intelligence/medical-drone-delivery-market
• Grepow — Cargo drones and platforms 2026: https://www.grepow.com/blog/top-cargo-drones-and-drone-delivery-platforms-in-2026.html

Edge AI / NPU

• arXiv — Energy-efficient on-device RAG on mobile NPU (2026-06): https://arxiv.org/html/2606.11257v1
• Qualcomm AI Engine: https://www.qualcomm.com/processors/ai-engine

Precision fermentation / alt protein

• Perfect Day — India facility 2026 start / 2027 ramp (2025-12): https://agfundernews.com/perfect-day-says-gujarat-facility-on-track-for-2026-start-2027-ramp-up-for-recombinant-whey-protein

Longevity / LEV

• LEV Foundation: https://www.levf.org/
• de Grey — Escape velocity concept: https://pmc.ncbi.nlm.nih.gov/articles/PMC423155/
• Diamandis — LEV discussion (2024): https://www.diamandis.com/blog/longevity-escape-velocity