09 — Energy, Permitting & Climate

STAGE 1 — EVIDENCE BRIEF

Referee panel: Prof. Olivia Brandt (#57, statistics, C) · Prof. June Takahashi (#17, econometrics, C) · Domain specialist Prof. Helen Ashby (#82, energy systems, C)

Ground rules from the panel. Every number below was checked against the source linked, with the data year noted. Where we used a secondary source (a trade-press summary, a Wikipedia summary of court records), we say so. One figure we could not retrieve the primary text for is flagged [unverified this session]. Professors citing numbers not on this list will be checked in Stage 4.

---

A. Interconnection and transmission

1. Queue size (end-2025). About 8,200 projects are actively seeking interconnection: 1,312 GW of generation and ~749 GW of storage. The total active queue is down 10% from the year before. — LBNL, Queued Up: 2026 Edition (data through end-2025). https://emp.lbl.gov/publications/queued-2026-edition-characteristics
2. Queue mix and direction. Solar 773 GW (−19% y/y), storage 749 GW (−16%), wind 220 GW (−19%), natural gas 253 GW (+86%). — same source.
3. Late-stage backlog. 549 GW already has a draft or executed interconnection agreement but isn't operating yet (solar 256, storage 161, wind 76, gas 45). — same source.
4. Wait time. The median time from interconnection request to commercial operation was over 5 years for projects built in 2025. — same source.
5. Attrition. Of the capacity that requested interconnection from 2000–2020, 13% was operating by end-2025, 75% had withdrawn, and 10% was still active. — same source.
6. High-voltage (345 kV+) transmission built. Annual averages: 1,781 miles (2010–14), 942 (2015–19), 536 (2020–24). 2024 was first reported as 322 miles, revised to 888 (334 miles at 345 kV plus 554 at 500 kV). 2023 was first reported as 55 miles, revised to ~450. 2013 had nearly 4,000 miles. — Grid Strategies/ACEG, Fewer New Miles (July 2025, with addendum). https://gridstrategiesllc.com/wp-content/uploads/ACEG_Grid-Strategies_Fewer-New-Miles-2025_vF.pdf
7. Need. The same report says ~5,000 miles/yr of high-capacity regional transmission is needed. It bases this on DOE's National Transmission Planning Study, which calls for transmission capacity to grow 2.1–3.3× by 2050. — same source. (Panel note: this is an advocacy group's translation of a modeled scenario. It's a need estimate, not a measured shortfall.)
8. Citing FERC's 2024 State of the Market report, ACEG says ~1,000 miles of new facilities (all voltages, load-growth-driven) went into service in 2024. — ACEG 2025 Transmission Planning & Development Report Card (Feb 2026). https://cleanenergygrid.org/wp-content/uploads/2026/03/ACEG_2025-Transmission-Planning-and-Development-Report-Card15.pdf

B. NEPA: timelines, length, litigation, law

1. EIS timelines, from CEQ's January 2025 report (1,903 EISs finalized 2010–2024). Median notice of intent (NOI) → final EIS: 2.8 years overall, 2.2 years in 2024. NOI → record of decision (ROD): 2021–24 median 2.5 years, mean 3.8. 2010–18 median 3.5, mean 4.5. Share completed in ≤2 years: 41% in 2024 vs 24% in 2019. CEQ itself says the median is the better measure because the distribution is right-skewed. https://nepa.gov/sites/default/files/documents/CEQ_EIS_Timeline_Report_2025-1-13.pdf
2. EIS length. EISs from 2013–2018 (n=656) averaged 575 pages. Earlier CEQ guidance said they "should normally be less than 150 pages." — House Natural Resources Committee hearing memo (Sept 10, 2025), citing CEQ. https://naturalresources.house.gov/UploadedFiles/Hearing_Memo_--_FC_Leg_Hrg_on_3_NEPA_bills_09.10.25.pdf
3. Litigation. The Breakthrough Institute reviewed 387 NEPA appellate cases (2013–2022). Agencies prevailed in ~80%. It took an average of 4.2 years from document publication to appellate resolution. NGOs filed 72% of challenges, and 10 organizations filed 35% of cases, winning 26% of them. Energy projects were 29% of cases and were delayed 3.9 years on average, though agencies won 71% of those. https://thebreakthrough.org/issues/energy/understanding-nepa-litigation
4. Clean-energy NEPA timelines. Per Resources for the Future (RFF), solar EISs averaged 27 months and wind 45 months. NEPA-reviewed solar was only ~10% of solar capacity added 2010–23 (wind: 3.7%). After NEPA was finished, 9 of 24 solar and 7 of 14 wind projects still needed 4+ years to reach operation. https://www.rff.org/publications/reports/how-long-does-it-take-national-environmental-policy-act-timelines-and-outcomes-for-clean-energy-projects/
5. Seven County Infrastructure Coalition v. Eagle County (May 29, 2025). Justice Kavanaugh wrote for five justices. Sotomayor, Kagan and Jackson concurred in the judgment, and Gorsuch was recused, so the judgment was 8–0 with a 5–3 split on reasoning. Courts must give agencies "substantial deference" on the scope of an EIS. An EIS "need not address the effects of separate projects" upstream or downstream. https://supreme.justia.com/cases/federal/us/605/23-975/
6. CEQ rescission. CEQ published an interim final rule removing all of its NEPA implementing regulations on Feb 25, 2025, effective April 11, 2025. https://www.federalregister.gov/documents/2025/02/25/2025-03014/removal-of-national-environmental-policy-act-implementing-regulations

C. Permitting legislation

1. Energy Permitting Reform Act (Manchin–Barrasso). Cleared Senate Energy and Natural Resources (ENR) 15–4 on July 31, 2024. It included a 150-day deadline to file judicial challenges and transmission provisions. It died at the end of the 118th Congress (Dec 2024). https://en.wikipedia.org/wiki/Energy_Permitting_Reform_Act_of_2024 (secondary)
2. SPEED Act (H.R. 4776). Passed the House 221–196 on Dec 18, 2025. Provisions:

• a 150-day statute of limitations (vs. the six-year general default);
• "substantial deference" to agencies;
• remand without vacatur as the remedy;
• standing limited to those who submitted substantive comments;
• an amendment that preserves the administration's permit cancellations made between Jan 20, 2025 and enactment. Clean-energy developers objected to that amendment.

https://bipartisanpolicy.org/issue-brief/whats-in-the-speed-act/

1. Senate status (Sept 23, 2026). Senate talks are stalled. Democrats (Heinrich, Schatz and others) want safeguards and clean-energy benefits. Separately, the Ratepayer Protection Act on data-center cost allocation passed the House 417–3 but is blocked in the Senate. https://dailycaller.com/2026/09/23/house-senate-permitting-reform-congress-speed-act-susan-collins-jon-husted/ (secondary)

D. Demand, reliability, prices

1. Consumption (EIA STEO, Aug 2026). 4,195 billion kWh in 2025 → 4,268 in 2026 → 4,391 in 2027 (a record each year). Commercial sales hit a record 1,493 BkWh in 2025. After Texas paused data centers in August 2026, EIA cut Texas's 2027 demand growth forecast from 14% to 6%. https://energynow.com/2026/08/us-power-use-to-beat-record-highs-in-2026-and-2027-as-ai-use-surges-eia-says-5/ (wire summary of EIA)
2. Data centers (LBNL, Dec 2024). 176 TWh = 4.4% of US electricity in 2023. Projected 325–580 TWh = 6.7–12% by 2028. https://newscenter.lbl.gov/2025/01/15/berkeley-lab-report-evaluates-increase-in-electricity-demand-from-data-centers/
3. NERC 2025 Long-Term Reliability Assessment (Jan 2026). 10-year summer peak demand grows +224 GW (24%), a forecast 69% higher than last year's. Winter peak grows +246 GW. NERC flags MISO, PJM, ERCOT and parts of the Pacific Northwest as high-risk. https://www.utilitydive.com/news/nerc-10-year-peak-demand-forecast-jumps-24-on-new-data-center-loads/810955/
4. Retail prices (EIA, July 2026 data). The all-sector average was 14.99¢/kWh, up 4.4% y/y. Residential prices rose 4.9%. Hawaii rose 25.4% and Connecticut fell 12.4%. https://www.eia.gov/electricity/monthly/update/end-use.php
5. Price drivers (AAF, Oct 2025, using EIA data). The all-sector average rose from 13.11¢ (Jan 2025) to 14.38¢ (Jul 2025), partly seasonal. AAF names transmission and distribution (T&D) spending and generation costs as the main residential drivers. https://www.americanactionforum.org/insight/how-much-are-electricity-prices-rising-and-why/
6. PJM capacity market. Annual capacity costs went from about $2.2B (2024 auction) to >$16B (2025). In Dec 2025, PJM's auction failed to procure its full requirement for the first time. https://en.wikipedia.org/wiki/PJM_Interconnection (secondary; figures trace to PJM auction reports)

E. Tax credits, emissions, damages

1. OBBBA phase-down. Wind and solar projects claiming 45Y/48E must begin construction by July 4, 2026 or be placed in service by Dec 31, 2027. The 25D residential credit ended Dec 31, 2025. New foreign-entity-of-concern (FEOC) restrictions apply from 2026. https://seia.org/research-resources/clean-energy-provisions-big-beautiful-bill/
2. Budget effect (CBO, 2025–2034, via Peterson Foundation). Savings: clean vehicle credits $190B, clean electricity investment credit $166B, residential clean energy $77B. Energy-related total $496B. https://www.pgpf.org/article/energy-tax-policy-under-the-obbba/
3. Emissions (Rhodium, Jan 2026). US greenhouse gases rose +2.4% in 2025, the first rise in three years, and are now 18% below 2005. By sector:

• power +3.8%, with coal generation up 13% as Henry Hub gas prices rose 58%;
• buildings +6.8%, driven by a cold winter;
• transport +0.1%.

Real GDP grew 1.9%. Rhodium's 2035 outlook fell to 26–35% below 2005, from its earlier 38–56%. https://rhg.com/research/us-greenhouse-gas-emissions-2025/

1. Social cost of carbon.

• The Obama-era Interagency Working Group (IWG) used $51/t at a 3% discount rate.
• EPA (Nov 2023) used $190/t at 2%.
• Executive Order 14154 disbanded the IWG on Jan 20, 2025.
• OMB memo M-25-27 (May 5, 2025) told agencies to stop counting climate damages except where a statute requires it.
• A federal court found procedural violations in DOE's July 2025 climate report.

https://eelp.law.harvard.edu/tracker/the-social-cost-of-carbon/

1. Rennert et al. (Nature, 2022). Mean $185/t (5–95% range: $44–$413) at a near-term 2% rate. The same model gives $80/t at 3%. https://www.nature.com/articles/s41586-022-05224-9
2. Retraction. Kotz, Levermann & Wenz (Nature, 2024), which projected a 19% global income loss by 2049 and $38T in annual damages, was retracted Dec 3, 2025, following Matters Arising critiques. https://www.nature.com/articles/s41586-024-07219-0

F. Carbon pricing

1. Canada. The consumer carbon price was set to zero effective April 1, 2025 (Carney directive, March 14, 2025). It had been C$80/t, scheduled to reach C$170 by 2030. The Parliamentary Budget Officer (PBO) found in 2019 that 80% of households would get more in rebates than they paid in direct and indirect costs. The industrial Output-Based Pricing System (OBPS) remains. https://en.wikipedia.org/wiki/Carbon_pricing_in_Canada (secondary)
2. EU ETS. Emissions in covered sectors fell 47% from 2005 to 2023. Bayer & Aklin (PNAS, 2020) attribute a ~3.8% cut in 2008–16 to the ETS itself. Other estimates range 7–11.5%. The ETS2 extension to buildings and transport slipped to 2028. CBAM's definitive phase starts in 2026. https://en.wikipedia.org/wiki/EU_Emissions_Trading_System (secondary)
3. British Columbia. Murray & Rivers (Energy Policy, 2015) review estimates of a 5–15% emissions reduction with negligible aggregate economic effects. [unverified this session: the abstract could not be retrieved] https://doi.org/10.1016/j.enpol.2015.08.011

G. Nuclear

1. Vogtle 3 & 4. $36.8B total vs. a $4.4B original estimate. The project took 15 years and the units began operating Aug 2023 and Apr 2024. https://georgiarecorder.com/2024/05/31/biden-administration-georgia-officials-applaud-debut-of-plant-vogtle-expansion/
2. NuScale / UAMPS Carbon Free Power Project. The target price rose from $58/MWh (2021) to $89/MWh (2023, after IRA subsidies), and cost rose from $3.6B to $9.3B. The project was cancelled in Nov 2023. NRC approved NuScale's 77 MWe design in May 2025. https://en.wikipedia.org/wiki/NuScale_Power (secondary)
3. ADVANCE Act. Signed July 9, 2024 after passing the House 393–13 and the Senate 88–2. It cut advanced-reactor fees and changed NRC's mission to "not unnecessarily limit" nuclear. https://en.wikipedia.org/wiki/ADVANCE_Act (secondary)
4. May 23, 2025 executive orders. Targets: 400 GW by 2050, 10 large reactors under construction by 2030, and an 18-month licensing deadline (an executive directive, not a statute). https://www.energy.gov/ne/articles/9-key-takeaways-president-trumps-executive-orders-nuclear-energy

H. Insurance

1. Keys & Mulder (NBER w32579). Using 74M escrow premium payments (2014–24), they find that repricing of catastrophe risk in global reinsurance markets made premiums rise faster in riskier places. In the most exposed ZIP codes, home price growth was >$40k lower. https://www.nber.org/papers/w32579
2. Florida. Citizens (the state insurer of last resort) had ~1.3M policies in June 2023 and <800k by June 2025. After the 2023 litigation reform (HB 837), the regulator reported average premiums down 0.7% in Q4 2024. https://en.wikipedia.org/wiki/Citizens_Property_Insurance_Corporation (secondary)
3. California. The FAIR Plan grew from 124k policies (2020) to 663k (Mar 2026), with exposure of $768B (Jun 2026). It levied a $1B assessment on insurers in Feb 2025 after the LA fires. Seven of the 12 largest insurers restricted new policies. https://en.wikipedia.org/wiki/California_FAIR_Plan (secondary)

I. Gas, siting conflicts, construction costs

1. LNG exports (EIA STEO, Sept 2026). 15.1 Bcf/d in 2025 → 17.4 in 2026 → 18.6 in 2027. Henry Hub averaged $3.53/MMBtu in 2025, with $3.43 forecast for 2026. https://www.eia.gov/outlooks/steo/report/natgas.php
2. Thacker Pass. BLM issued its ROD Jan 15, 2021. The 9th Circuit denied an injunction (July 2023), and the tribes' suit was dismissed with prejudice (Dec 2023). Judge Du found the evidence did not show a massacre within the project area. DOE took a 5% equity stake in Oct 2025. https://en.wikipedia.org/wiki/Thacker_Pass_lithium_mine (secondary)
3. Dakota Access. The D.C. Circuit (2021) required a full EIS and the easement was vacated. The Supreme Court denied cert in Feb 2022. The pipeline kept operating. https://en.wikipedia.org/wiki/Dakota_Access_Pipeline (secondary)
4. Energy Transfer v. Greenpeace. A $667M jury verdict (Mar 2025) was reduced by the judge to $345M and is on appeal. https://en.wikipedia.org/wiki/Energy_Transfer_v._Greenpeace (secondary)
5. Transit Costs Project (NYU). The US is the 6th most expensive country for rapid rail. Second Avenue Subway Phase 1 cost 8–12× a low-cost baseline. Soft costs are 21% in New York vs. 5–10% abroad. Labor is 40–60% of hard costs vs. 19–30%. https://transitcosts.com/wp-content/uploads/TCP_Final_Report.pdf

---

CONTESTED EVIDENCE (where the literature really disagrees)

(i) Is NEPA the binding constraint on clean-energy build-out?

• Yes-ish: CEQ's own median NOI→ROD was 2.5–3.5 years (#9), EISs averaged 575 pages (#10), and energy litigation added 3.9 years even when agencies won (#11).
• No-ish: RFF (#12) finds NEPA-reviewed projects are a small share of solar and wind, and much of the delay comes after the ROD. LBNL (#4–5) shows the grid-connection queue alone takes over 5 years and has 75% attrition. Agencies win ~80% of appeals (#11).
• Both can be true. NEPA is a large tail risk for big linear projects (transmission, pipelines) and a minor factor for private-land solar.

(ii) How big is data-center load?

• LBNL's own 2028 range spans nearly a factor of 2 (325–580 TWh, #19).
• NERC raised its 10-year forecast 69% in one year (#20). EIA cut Texas's 2027 growth from 14% to 6% after one policy change (#18).
• Forecasts are volatile and utilities have incentives to overstate them. Neither the "bubble" camp nor the "shortage" camp can claim the data yet.

(iii) Why are retail prices rising?

• AAF (#22) points to T&D capex and generation costs.
• Rhodium (#26) documents a 58% Henry Hub gas price rise in 2025.
• PJM capacity costs rose about 7× (#23).
• No study we found cleanly decomposes the 2025–26 increase into data-center, gas, T&D, wildfire-liability and renewables-integration shares. Anyone who assigns a single cause is overreaching.

(iv) Does carbon pricing work?

• EU ETS covered emissions fell 47%, but the causal share attributed to the ETS ranges from ~3.8% to 7–11.5% (#31).
• The BC review evidence (#32) is modest but positive.
• Canada's repeal (#30) shows political fragility, not economic failure. The two are routinely conflated.

(v) What are climate damages?

• The social cost of carbon runs from $51 (3% discount rate) to $185–190 (2%) (#27–28). The discount rate is the single largest driver.
• A high-profile damage paper was retracted (#29). The current federal practice of effectively valuing damages at zero (#27) is not an estimate at all.

(vi) Nuclear costs. Vogtle (#33) and NuScale (#34) are first-of-a-kind observations. Whether costs fall with repetition is a forecast, not a US data point.

(vii) Insurance.

• Keys & Mulder (#37) attribute premium growth to disaster risk and reinsurance repricing.
• Florida's post-2023 stabilization (#38) suggests litigation and fraud were first-order there.
• California's FAIR Plan growth (#39) coincided with rate regulation that barred catastrophe models and reinsurance pass-through until the Sustainable Insurance Strategy.
• Climate, regulation and litigation all matter, and their relative weight differs by state.

— Brandt, Takahashi, Ashby

STAGE 2 — OPENING POSITIONS (Part 1 of 2)

Format for each speaker: causal claim · evidence · remedy · tradeoff conceded · falsifiable prediction.

---

1. Prof. Mei Lin (#84, climate science, L): "We are un-decarbonizing, and policy did it"

Causal claim. The United States has left the decarbonization path it was on in 2023. The 2025 emissions reversal shows what happens when federal policy flips from support to active obstruction.

Evidence.

• Rhodium puts 2025 greenhouse gas emissions up 2.4% (#26). I attribute that reversal to the OBBBA rollback and the administration's war on wind and solar.
• Rhodium's own 2035 outlook fell from 38–56% below 2005 to 26–35% (#26).
• Data centers already take about 12% of US power and are being fed by coal and gas.
• Damages are enormous. Nature published an estimate of a 19% global income loss by 2049 that is already locked in. Even the conservative social cost of carbon is $190/t (#27).

Remedy. Restore the 45Y/48E clean electricity credits through 2032. Condition data-center interconnection on 24/7 clean procurement. Reinstate a social cost of carbon of at least $190 in all regulatory analysis.

Tradeoff I concede. Restoring the credits costs real money, on the order of the $166B CBO scored for terminating the clean electricity investment credit (#25). Some of that money subsidizes projects that would have been built anyway.

Falsifiable prediction. If I am right that policy is now the binding factor, 2027–2028 solar and wind additions will fall after the safe-harbor pipeline clears. That will show up in EIA capacity data by 2029 as a decline of at least 30% from the 2025–26 peak, and power-sector emissions will not return to their 2024 level before 2030.

---

2. Prof. Rosa Delgado-Finch (#7, political economy, Soc): "The grid is a public good run for private return"

Causal claim. Rising bills are not a law of physics. They come from a regulated-monopoly model: utilities earn a return on capital spending, while merchant generators profit from scarcity. Data centers are the new rationale for both.

Evidence.

• PJM capacity costs went from $2.2B to over $16B in one auction cycle (#23). That is a transfer from households to generators, triggered by hyperscaler load.
• Residential prices are rising fastest of any sector (+4.9% y/y, #21).
• The rise in electricity prices is basically all about data centers.
• The House passed a data-center cost-allocation bill 417–3 (#17), and the Senate still blocked it. That tells you who writes the rules.

Remedy.

• A federal public transmission authority that builds backbone lines at a public cost of capital.
• Mandatory large-load tariffs that make data centers pay the full incremental cost of the generation and transmission they trigger.
• Public-power options for municipalities.

Tradeoff I concede. Public agencies in this country build badly. The Transit Costs Project shows US soft costs of 21% vs. 5–10% abroad (#44). A public transmission authority would inherit that culture unless it builds in-house expertise.

Falsifiable prediction. In PJM zones with the most data-center load, residential bills will rise at least twice as fast as in low-data-center zones through 2028. If they do not, my data-center story is wrong.

---

3. Prof. Aaron Blackwood (#79, Native American studies, L): "Streamlining is being built on the backs of tribal nations"

Causal claim. The permitting-reform agenda treats consultation as friction. Seven County, the CEQ rescission and the SPEED Act together mostly remove the one lever that tribes and frontline communities have had: the ability to litigate an inadequate review.

Evidence.

• Seven County requires "substantial deference" and removes upstream and downstream effects from EIS scope (#13).
• CEQ's regulations, which contained the most detailed consultation procedures, are gone (#14).
• SPEED gives challengers 150 days to sue, limits standing to those who commented, and limits the remedy to remand without vacatur (#16).
• Courts have repeatedly vindicated tribes when they sue. Standing Rock won at the D.C. Circuit (#42).
• The industry answer to the Standing Rock protests was a $667M verdict against Greenpeace (#43).

Remedy.

• Free, prior and informed consent (FPIC)-style early consultation with federal funding for tribal technical capacity.
• A statutory exception to remand-without-vacatur for treaty-rights and sacred-site claims.
• Tribal equity options on projects crossing treaty lands.

Tradeoff I concede. Earlier, funded consultation adds months at the front of projects, and a consent standard raises holdout risk for linear projects.

Falsifiable prediction. If SPEED's remedy provisions become law, the share of tribal NEPA challenges that change a project's route or design will fall toward zero within 5 years, and the number of physical protest actions at sites will rise.

---

4. Prof. Stuart Pell (#85, environmental economics, C): "Price the externality, then get out of the way"

Causal claim. Stacking subsidies, mandates and a permitting maze is the most expensive way to cut emissions. A broad carbon price does the most for the least.

Evidence.

• The EU ETS covered sectors cut emissions 47% since 2005 (#31).
• Canada's PBO found 80% of households came out ahead under the rebate (#30).
• The central social cost of carbon estimate from modern damage modeling is $185 (#28).
• By contrast, the IRA's credits cost more than their initial scores. CBO found repealing just three of them saves $433B (#25).

Remedy. A revenue-neutral carbon fee starting at ~$50/t and rising, with an equal per-capita dividend and a border adjustment aligned with the EU's CBAM (#31). In exchange, preempt EPA power-plant greenhouse gas rules and phase out the remaining clean-energy credits.

Tradeoff I concede. Canada shows that a visible consumer carbon price can be politically fatal even when most households come out ahead (#30). And the ETS's causal share of that 47% is much smaller than the headline (#31).

Falsifiable prediction. If a US fee at $50+ is enacted, power-sector emissions will fall at least 20% within 5 years relative to Rhodium's current-policy baseline, and coal generation will collapse regardless of gas prices.

---

5. Prof. Richard Okonkwo (#81, civil engineering, C, "abundance"): "We have made building illegal by default"

Causal claim. The US cannot build the physical system decarbonization or energy dominance requires, because every large linear project runs a gauntlet of review, litigation and interconnection studies that no project was designed to survive.

Evidence.

• High-voltage transmission construction fell from 1,781 miles/yr (2010–14) to 536 (2020–24). In 2023 we built 55 miles. The need is ~5,000 miles/yr (#6–7).
• EISs take 4.5 years on average and run 575 pages (#9–10).
• NEPA litigation takes 4.2 years to resolve and delays energy projects 3.9 years, even though agencies win ~80% of the time (#11). The delay is the penalty, not the verdict.
• The interconnection queue has a median wait of over 5 years and 75% attrition (#4–5).
• Transit: Second Avenue Subway cost 8–12× a low-cost baseline (#44).

Remedy.

1. Pass SPEED-style judicial review limits (150 days, remand default).
2. Create a federal backstop siting authority for interregional transmission.
3. Require "connect and manage" interconnection.
4. Standardize designs and grow in-house engineering at transit agencies.

Tradeoff I concede. Faster permitting builds everything faster: pipelines and LNG as well as wind lines. Climate hawks who want permitting reform only for clean energy will not get that deal.

Falsifiable prediction. If the SPEED Act passes with a transmission title, 345 kV+ miles completed will exceed 2,000/yr by 2031. If litigation limits pass without interconnection and cost-allocation reform, miles will not rise much, which would show that litigation was not the binding constraint.

(Part 2 follows.)

STAGE 2 — OPENING POSITIONS (Part 2 of 2)

---

6. Prof. Kyle Brennan (#83, nuclear engineering, C-R): "Firm, clean power at scale means nuclear, and the regulator was the bottleneck"

Causal claim. NERC's warnings (#20) are about firm capacity. Only gas and nuclear supply firm capacity at scale, and only nuclear does it without carbon. US nuclear costs are high mainly because we stopped building, lost the supply chain, and kept an NRC whose culture treated delay as safety.

Evidence.

• Vogtle cost $36.8B (#33), but it was first-of-a-kind (FOAK) AP1000 construction by a workforce that had forgotten how.
• The ADVANCE Act passed 88–2 and 393–13 (#35), which is a rare bipartisan signal.
• The NRC is now legally required to decide licenses within 18 months (#36).
• NRC approved NuScale's uprated design in 2025 (#34).

Remedy.

• Codify fixed-timeline licensing.
• Federal cost-overrun insurance for the first ~10 reactors of a standardized design, with a clawback if later units don't show cost declines.
• Continue technology-neutral credits for firm clean power.

Tradeoff I concede. NuScale's Idaho project died when the price went from $58 to $89/MWh after subsidies (#34). Small modular reactors (SMRs) have not yet shown cost advantages. Taxpayers would bear real overrun risk.

Falsifiable prediction. If the regulatory-bottleneck thesis is right, at least 3 construction permits for advanced reactors will be issued by end-2028, and the second standardized unit at any site will come in at least 25% below the first in real $/kW. If costs don't fall on repetition, my case collapses to "nuclear is expensive, full stop."

---

7. Prof. Dennis Harlow (#92, resource engineering, R): "Energy dominance is a reliability policy"

Causal claim. A decade of subsidies and mandates favoring intermittent generation, plus early retirement of dispatchable plants, created the reliability and price squeeze NERC now describes. Abundant US gas, domestically and as LNG, is the fastest route to reliable, affordable power and to leverage over Russia and Iran.

Evidence.

• NERC says four major regions are at high risk as load grows 24% in a decade (#20). NERC says the risk comes from retiring dispatchable plants faster than we replace them with firm capacity.
• Gas in the queue is up 86% (#2), a market signal that firm power is what's needed.
• Electricity prices are rising because of renewable mandates and the cost of integrating intermittent resources.
• LNG exports are rising from 15.1 to 17.4 Bcf/d while Henry Hub is forecast to fall ($3.53 → $3.43, #40). LNG exports don't raise domestic prices.

Remedy.

• Keep the OBBBA phase-downs (#24).
• Fast-track gas pipelines and LNG terminals via SPEED.
• Enforce capacity accreditation that pays for firmness.
• Pause retirements of dispatchable plants in NERC high-risk regions.

Tradeoff I concede. More gas and a coal-retirement pause mean higher emissions in the near term. Rhodium's +2.4% (#26) shows that directly. I accept that as the price of reliability.

Falsifiable prediction. If my thesis holds, regions that keep more dispatchable capacity will have fewer NERC energy emergency alerts through 2030. And as US LNG export capacity nearly doubles, Henry Hub will stay below $4.50 in real terms through 2030.

---

8. Prof. Walter Pruitt (#8, law & economics, Lib): "Let prices tell the truth: about carbon, about risk, about litigation"

Causal claim. Most of this domain's failures come from suppressed price signals:

• insurance rates held below risk;
• subsidies that pick winners;
• a litigation system with no loser-pays;
• utility rates that hide who causes costs.

Evidence.

• Florida's market recovered because of the 2023 tort reform. Citizens fell from ~1.3M to under 800k policies and premiums fell 0.7% (#38).
• California capped rates and barred reinsurance pass-through, and the FAIR Plan exploded fivefold (#39).
• Agencies win ~80% of NEPA appeals yet energy projects are still delayed 3.9 years (#11). That is litigation as a hold-up tax.

Remedy.

• Risk-based insurance pricing with no state caps.
• End subsidies to all energy sources: the remaining clean credits, LNG loan programs, and DOE equity stakes like Thacker Pass (#41).
• A short statute of limitations plus a bond requirement for injunctions.
• Large loads pay for what they cause.

Tradeoff I concede. Risk-based insurance pricing will make some coastal and wildland-urban-interface homes unaffordable to insure. Keys & Mulder find price effects of over $40k on the most exposed homes (#37). That is a real wealth loss for people who bought under the old prices.

Falsifiable prediction. If California's Sustainable Insurance Strategy truly allows catastrophe models and reinsurance pass-through, FAIR Plan policies will fall at least 25% from their 2026 peak by 2029, as Citizens' count did in Florida.

---

9. Prof. Ingrid Solberg (#23, administrative law, C): "The legal ground has already moved; Congress is legislating for 2020"

Causal claim. Between Seven County (#13) and CEQ's rescission (#14), the courts and the executive have already delivered most of what permitting reformers asked for in 2023. The remaining bottleneck is administrative capacity and legal uncertainty created by the reform itself. Every agency now writes its own NEPA procedures, and the first round of litigation over them has not finished.

Evidence.

• CEQ's median time from NOI to final EIS was already 2.2 years in 2024, and 41% of EISs finished within 2 years (#9). The 2023 Fiscal Responsibility Act deadlines were working before Seven County.
• Interconnection (median over 5 years, #4) and transmission cost allocation are not NEPA problems at all.
• SPEED's amendment preserving permit cancellations since January 2025 (#16) cuts against the certainty permitting reform is supposed to deliver. A wind developer now faces executive risk that no court can check.

Remedy. Legislate certainty in both directions:

• a 150-day statute of limitations and protection against arbitrary revocation of permits already issued;
• fund permitting staff;
• codify Seven County's scope rule so it doesn't swing with the next Court.

Tradeoff I concede. Codifying remand-without-vacatur means some inadequate reviews will stand. Dakota Access shows that even with vacatur, the pipeline kept running (#42). So the remedy was already weak, and SPEED makes it formally weaker.

Falsifiable prediction. Without new legislation, the median NOI→ROD for EISs issued 2026–2028 will fall below 2 years on Seven County and agency procedures alone. If it doesn't, the case for statutory reform strengthens.

STAGE 3 — CROSS-EXAMINATION (Part 1: the reliability, prices and emissions fight)

---

Exchange A: Lin (L) ⇄ Harlow (R), with Ashby (C, specialist) intervening. Four rounds.

Round 1: Lin challenges Harlow.

Dennis, you say prices are rising "because of renewable mandates and intermittency." Show me the decomposition. Rhodium (#26) says Henry Hub gas rose 58% in 2025. That pushed coal generation up 13% and raised emissions. AAF, hardly a green outfit, lists T&D spending and generation costs as the main drivers (#22). Also, NERC doesn't say "renewables cause risk." The headline in the 2025 LTRA is that demand is growing faster than any resource is being added (#20).

Then look at the queue. There's 773 GW of solar and 749 GW of storage waiting (#2), and 549 GW with interconnection agreements, most of it solar, storage and wind (#3). If firm capacity were the bottleneck, why is the stuff that could be online in 18 months stuck behind interconnection studies? Your gas surge is +86%, but from what base? 253 GW of gas, and you know the turbine order books.

Round 1: Harlow responds.

I'll concede something narrow: I don't have a clean decomposition that pins the 2025–26 retail increase on renewables. Nobody does, and the brief says so (contested item iii). My claim is structural, not a regression coefficient. Capacity markets price accredited firm capacity. PJM's jump from $2.2B to over $16B (#23) happened because accredited supply didn't keep up with forecast load, and that's partly because solar's winter-peak accreditation is low.

On NERC, fair: the headline is demand outrunning resources. But the report's risk areas are the regions where dispatchable retirements are concentrated. I'll soften "caused by renewables" to "made worse by retiring dispatchable capacity faster than accredited replacements arrive." And the 549 GW with interconnection agreements hasn't been built. Some of that is supply chains and tariffs, sure. Some of it is that intermittent projects don't pencil without credits.

Round 2: Ashby (C, energy systems) intervenes.

Two points of fact, Dennis. First, you say "the regions where dispatchable retirements are concentrated," but we didn't verify NERC's regional retirement figures for this thread, so the referees will rate that claim contested at best. Second, your gas story has a physical bottleneck you haven't priced in. The 253 GW of gas in the queue (#2) goes through the same interconnection process with the same median over 5 years (#4). Gas doesn't jump the queue. If you want firm power fast, the binding constraint is the queue and transmission, not the credits.

Round 2: Harlow responds.

Accepted on the queue point. It's why I back connect-and-manage and surplus interconnection at existing coal sites: a retiring plant's grid connection is the most valuable asset in the country. But I don't accept that 549 GW of signed interconnection agreements proves renewables are ready. LBNL's own completion rate is 13% for the 2000–2020 cohort (#5). An interconnection agreement is an option, not a plant.

Round 3: Lin presses.

Then be consistent. Your LNG claim is that "exports don't raise domestic prices" because the 2026 Henry Hub forecast is $3.43 vs $3.53 (#40). That's one year of an EIA forecast. The same year you cite, 2025, Henry Hub rose 58% (#26) while exports hit a record. A one-year forecast can't prove a zero elasticity. Every serious model gives a positive price effect. The dispute is how big.

Round 3: Harlow responds.

I'll amend: "LNG exports have so far had modest domestic price effects, and the forward curve doesn't show the spike the 2024 pause advocates predicted." I didn't retrieve DOE's 2024 study in this session, so I won't cite a number from it. My prediction stands: Henry Hub below $4.50 real through 2030. If it breaks that, I'm wrong.

Round 4: Lin closes; Harlow's final word.

Lin: Then I'll put a concession on the record too, below, because Walter is about to catch me on something.

Harlow: Final word. Mei's 2029 prediction and mine are compatible. Solar and wind additions can fall and reliability can improve if the capacity that replaces them is firm. What would change my mind: if the high-risk regions that add mostly storage-plus-solar (ERCOT is the test) show fewer emergency alerts than MISO/PJM through 2029.

---

Exchange B: Pruitt (Lib) challenges Lin (L). Three rounds.

Pruitt:

Mei, three problems.

1. You said 2025's emissions rise came from the OBBBA rollback. Rhodium says the drivers were a cold winter (buildings +6.8%) and high gas prices pushing coal up 13% (#26). OBBBA's wind and solar deadlines don't bind until July 2026 and end-2027 (#24). Projects under construction in 2025 were safe-harbored. The law can't have caused 2025.

2. Your "19% income loss by 2049" paper was retracted by Nature on Dec 3, 2025 (#29).

3. Data centers are 4.4% of US electricity (2023), with a 2028 projection of 6.7–12% (#19). They are not 12% today.

Lin responds:

Guilty on all three counts, and I'll say why.

- On (1), the 2025 increase was weather and gas prices. The Rhodium sector breakdown is explicit. My policy claim belongs in the forward outlook, where Rhodium cut its 2035 range from 38–56% to 26–35% (#26). That's where the OBBBA effect shows up.

- On (2), I withdraw the Kotz figure entirely. I'd point to Rennert et al. (#28), $185 at 2% and $80 at 3%. That peer-reviewed estimate stands, and it's still well above a federal practice that now effectively counts zero (#27).

- On (3), I conflated the top of LBNL's 2028 range with today. The correct statement is 4.4% in 2023, possibly 6.7–12% by 2028.

I don't concede the core. The forward trajectory is now set by policy, and the policy is pointed the wrong way.

Pruitt, round 2:

Rennert's own paper shows the number more than doubles on a single parameter, the discount rate. The social cost of carbon (SCC) is a choice about intergenerational ethics dressed up as a measurement. Your "at least $190" remedy is a value judgment, and you should say so.

Lin, round 2:

It's both. Discounting is an ethical choice informed by market rates. Rennert's 2% comes from the decline in long-run real rates, and they report it transparently. Choosing zero is also a value judgment, just a hidden one. I'll amend my remedy: "Use an SCC with discount-rate sensitivity at 2%, 2.5% and 3%, as EPA's 2023 method did." That's more defensible than "at least $190."

Pruitt, round 3:

Accepted. I'd go further: publish the range and let Congress own the number. That's where we differ. You want a regulatory SCC, and I want a legislated price or none.

---

Exchange C: Harlow and Takahashi challenge Delgado-Finch (Soc)

Harlow:

Rosa, "basically all about data centers" isn't supported. EIA's July 2026 figures show the largest increase in Hawaii (+25.4%) and a decline in Connecticut (−12.4%) (#21). Neither is a data-center story. Nationally the residential increase is 4.9%. Gas prices, T&D spending and wildfire costs are all in the mix (#22, #26).

Takahashi (referee, procedural):

Our brief explicitly says no study we found decomposes the increase. Your PJM point (#23) is strong for PJM. The national claim goes beyond the evidence.

Delgado-Finch responds:

I'll narrow it. In PJM, the capacity-cost explosion is driven by forecast large load, and that's where my prediction is aimed. Nationally, I retreat to "data centers are a growing and unpriced share of the cost increase." But notice what Dennis just did: he used Hawaii, an oil-import island grid, to rebut a claim about interconnected markets. The PJM market monitor's attribution is the fight worth having, and it won't favor him.

Harlow, round 2:

We agree large loads should pay. The 417–3 House vote (#17) shows that's not left or right. The question is whether you want public backbone transmission. The Transit Costs Project (#44) is a warning about public construction in this country, and you conceded that yourself.

Delgado-Finch, round 2:

The Transit Costs Project's diagnosis was outsourced expertise and adversarial procurement: soft costs of 21% vs 5–10% (#44). The low-cost countries it holds up, Italy, Sweden and Turkey, all build with public agencies with strong in-house staff. The lesson is to build public capacity, not to privatize. Richard should back me on this.

STAGE 3 — CROSS-EXAMINATION (Part 2: permitting, nuclear, carbon pricing, consultation, insurance)

---

Exchange D: Blackwood (L) challenges Okonkwo (C, abundance). Three rounds.

Blackwood:

Richard, your numbers are stale, and your mechanism is backward.

1. "55 miles in 2023" was revised to ~450, and 2024 was revised to 888 (#6). You used the uncorrected number.

2. "EISs take 4.5 years on average" is the 2010–18 mean NOI→ROD. CEQ's 2021–24 median is 2.5 years, and the 2024 median to final EIS is 2.2 (#9).

3. RFF shows that NEPA-reviewed solar was ~10% of solar added, and that post-ROD delays of 4+ years hit 9 of 24 solar projects (#12). Agencies win ~80% of appeals (#11). You want to strip remedies from tribes to fix a problem that mostly sits in interconnection queues and utility planning.

Okonkwo responds:

I accept (1) and (2) as corrections. I'll amend to: "345 kV+ build averaged 536 miles/yr in 2020–24 against a need of ~5,000" and "median EIS NOI→ROD 2.5 years recently, mean 3.8, with a long tail." The tail is my point, though. The mean is 3.8 because some reviews take 7+ years, and those are disproportionately big linear projects: transmission lines and pipelines.

On (3), RFF's sample is renewables, and most solar sits on private land that never triggers NEPA. Transmission lines cross federal land and federal permits. Breakthrough finds energy cases took 3.9 years even though agencies won 71% (#11). If a project wins in court, that delay was pure deadweight loss.

Blackwood, round 2:

Pure deadweight only if the review didn't change the project. Litigation threat is what makes agencies do the review properly in the first place. That's the deterrent. You can't measure it by win rates. Dakota Access lost at the D.C. Circuit (#42), and even that loss didn't stop the oil. The remedy is already toothless. SPEED makes it toothless by statute.

Okonkwo, round 2:

Then we agree vacatur rarely bites. That's an argument that SPEED's remand-without-vacatur changes little in practice while cutting uncertainty a lot. I'd trade you something real for it: funded, early, binding-timeline tribal consultation, with a sacred-sites carve-out from remand-only. What I won't accept is the six-year default statute of limitations. No capital market will fund a transmission line that can be challenged until year six.

Blackwood, round 3:

A 150-day window with funded tribal technical capacity is a different conversation from 150 days with nothing. I'd take the carve-out. I won't accept the SPEED amendment protecting the administration's cancellations (#16). You can't sell "certainty" while exempting the executive's own revocations.

Okonkwo:

On that, you and Ingrid are right, and I'll say so in the vote.

---

Exchange E: Delgado-Finch (Soc) and Lin (L) challenge Brennan (C-R)

Delgado-Finch:

Kyle, "the NRC is now legally required to decide within 18 months" is wrong. It's a May 2025 executive order directive (#36). The ADVANCE Act (#35) changed fees and the mission statement. It didn't enact an 18-month statutory clock. Executive orders don't bind an independent commission's adjudications the way a statute does.

And Vogtle wasn't just a regulatory story. $36.8B vs $4.4B (#33), with Westinghouse's 2017 bankruptcy, labor shortages and a strike. That's construction management, not the NRC.

Brennan responds:

Corrected: the 18 months is an EO directive, not law. That's exactly why I want it codified. On Vogtle, I agree most of the overrun was construction execution and supply chain, not licensing. My claim is narrower: licensing uncertainty is one reason nobody builds a second and third AP1000, and the second-unit learning is where costs fall. That's a forecast. The brief rightly lists it as unverified in US data (contested item vi).

Lin:

Then your cost-overrun insurance is a public bet on your forecast. Why should taxpayers take the tail risk when NuScale's price went from $58 to $89/MWh with the IRA subsidy (#34)?

Brennan, round 2:

Because the alternative is nobody building firm clean power, and your 24/7 clean data-center mandate is impossible without it. The clawback is the discipline: if unit 2 isn't at least 25% cheaper per kW, the program ends. I'd also point out that the ADVANCE Act passed 88–2 (#35). Nuclear support is one of the few genuinely cross-bloc positions in this domain.

Lin, round 2:

I'll grant that nuclear belongs in a clean firm portfolio. I won't grant that it deserves open-ended overrun insurance while wind and solar lose their credits.

---

Exchange F: Harlow (R) and Delgado-Finch (Soc) challenge Pell (C)

Harlow:

Stuart, Canada is the natural experiment, and you lost it. A national consumer carbon price with dividends was zeroed out on April 1, 2025 by a Liberal PM (#30). If it can't survive in Canada, it can't pass in the US.

Delgado-Finch:

And your "80% of households come out ahead" comes from the PBO's 2019 analysis (#30). The public never believed it, because the dividend was invisible and the pump price wasn't. Market mechanisms fail on the politics of visibility.

Pell responds:

Two different claims there.

- Political durability: conceded. Canada is strong evidence that visible consumer carbon prices are fragile. (BC's long-running tax is the counterexample I'd cite, but the panel couldn't verify the BC evidence this session (#32), so I'll rest on the EU.)

- Economic effectiveness: Canada's repeal says nothing about that. The EU ETS survived with the political fight over consumers, by delaying ETS2 to 2028 (#31) and keeping upstream industrial pricing. Canada kept its industrial OBPS (#30).

I'll amend my remedy: price carbon upstream and industrially first, as the EU and Canada's OBPS do. Rebate visibly, and leave retail motor fuel for last.

I also have to correct myself on the ETS. The 47% is the total decline in covered sectors. The causal share attributed to the ETS is ~3.8% (Bayer & Aklin) to 7–11.5% in other estimates (#31). My original framing implied the ETS did the 47%, and it didn't.

Harlow, round 2:

Then your own evidence says the carbon price did a small fraction of the work, and cheap gas and technology did the rest. Why pay the political cost?

Pell, round 2:

Because 3.8% at prices that were trivial for most of 2008–16 is a low-price result. The carbon price is also the only instrument that lets us repeal EPA power-plant rules and subsidies in the same bill. I'm offering you a trade: price in, mandates and credits out. That's the Baker–Shultz structure. Also, "IRA credits cost more than their initial scores" was something I asserted without a source in this brief. I'll withdraw the "more than initial scores" part and keep only the CBO termination savings (#25).

---

Exchange G: Lin (L) challenges Pruitt (Lib) on insurance; Harlow (R) challenges Blackwood (L) on tribal claims

Lin → Pruitt:

Walter, Florida "recovered because of tort reform" is a single-cause story. Keys & Mulder (#37) show a nationwide repricing of catastrophe risk through reinsurance, with >$40k home-price effects. Florida's Q4 2024 premium change was −0.7% (#38) after years of double-digit increases. That's stabilization, not recovery to pre-crisis prices. And California's problem is wildfire risk growing, not just rate regulation.

Pruitt:

I'll amend "recovered because of" to "stabilized after." The Citizens drop from 1.3M to under 800k (#38) is clearly consistent with tort reform working, but I can't isolate it from reinsurance cycles and depopulation incentives. On California, I'll stand firm: the FAIR Plan went from 124k to 663k (#39) while the state barred catastrophe models and reinsurance pass-through. That's a textbook price-control shortage, whatever the underlying risk trend.

Lin, round 2:

Both. The risk rose, and California's regulation turned a price increase into a quantity shortage. I'll accept that.

Harlow → Blackwood:

Aaron, "courts have repeatedly vindicated tribes" cherry-picks Standing Rock. At Thacker Pass the tribes lost at the 9th Circuit, and their case was dismissed with prejudice (#41). The court found the evidence did not show a massacre in the project area. And the Greenpeace verdict was about trespass and defamation, not a lawsuit against tribes.

Blackwood:

I'll amend to "tribes have sometimes won, as at Dakota Access, and even their wins were not enforced" (#42). Thacker Pass supports my point more than yours. The review went ahead over tribal objection, and now the federal government is a 5% owner of the mine (#41). When the permitting agency's parent government is an equity holder, how is the review independent? On Greenpeace, the $667M verdict (reduced to $345M, #43) was against the NGO allied with the tribes, and its chilling effect on allied advocacy is the point.

Harlow, round 2:

On the DOE equity stake I actually agree with you and Walter. The government shouldn't hold equity in projects it regulates.

STAGE 4 — REFEREE FACT-CHECK

Brandt (#57) · Takahashi (#17) · Ashby (#82)

Ratings: S = Supported · C = Contested (the evidence is mixed, the claim is outdated, or it goes beyond the source) · U = Unsupported (no source provided or the sources don't establish it) · W = Wrong (contradicted by the cited source). "#n" refers to the numbered facts in the Evidence Brief.

# · Speaker · Claim (as made in Stages 2–3) · Rating · Basis
1 · Lin · US GHG rose 2.4% in 2025 · S · Rhodium #26
2 · Lin · 2025 rise was caused by the OBBBA rollback · W · Rhodium attributes it to cold weather (+6.8% buildings) and a 58% gas price rise driving coal +13% (#26). OBBBA deadlines bind from July 2026 at the earliest (#24).
3 · Lin · Rhodium's 2035 outlook cut to 26–35% below 2005 · S · #26
4 · Lin · Data centers already use ~12% of US power · W · 4.4% in 2023; 6.7–12% is a 2028 projection (#19)
5 · Lin · Nature: 19% global income loss by 2049 "locked in" · W · Paper retracted Dec 3, 2025 (#29)
6 · Lin · EPA 2023 SCC = $190/t · S · #27 (calling it "conservative" is her opinion; the figure itself is correct)
7 · Lin · ~$166B CBO for clean electricity investment credit · S · #25
8 · Lin · Henry Hub +58% in 2025 · S · #26
9 · Delgado-Finch · PJM capacity cost $2.2B → >$16B · S · #23 (secondary source)
10 · Delgado-Finch · Residential prices rising fastest (+4.9%) · S · #21 (tied with transportation)
11 · Delgado-Finch · Price increases "basically all about data centers" · U · #21 (largest changes: HI +25.4%, CT −12.4%), #22, #26; no decomposition exists
12 · Delgado-Finch · Ratepayer Protection Act 417–3, blocked in Senate · S · #17 (secondary source)
13 · Delgado-Finch · Low-cost transit countries rely on in-house public expertise · S · #44
14 · Blackwood · Seven County: substantial deference, separate projects excluded · S · #13
15 · Blackwood · SPEED: 150 days, commenter-only standing, remand without vacatur · S · #16
16 · Blackwood · Courts have "repeatedly vindicated" tribes · C · DAPL won (#42); Thacker Pass lost, dismissed with prejudice (#41)
17 · Blackwood · $667M verdict against Greenpeace · S · #43 (later cut to $345M, which he added in cross-examination)
18 · Blackwood · DAPL kept operating despite vacatur · S · #42
19 · Blackwood · DOE holds 5% of Thacker Pass · S · #41
20 · Pell · EU ETS cut covered emissions 47% (implied causal) · C · The 47% is the total decline. The attributable share is ~3.8% to 7–11.5% (#31). He self-corrected.
21 · Pell · PBO: 80% of Canadian households come out ahead · C · Accurate for the PBO's 2019 analysis of direct and indirect costs (#30). It does not settle full economic incidence, and it's a 2019 projection.
22 · Pell · Rennert et al. $185/t · S · #28
23 · Pell · Repealing three credits saves $433B · S · 190+166+77 (#25)
24 · Pell · IRA credits "cost more than initial scores" · U · No source in the record. Withdrawn.
25 · Okonkwo · "In 2023 we built 55 miles" of HV transmission · W · Revised to ~450 miles (#6)
26 · Okonkwo · EISs take 4.5 years on average · C · True of the 2010–18 mean NOI→ROD. The 2021–24 median is 2.5 and the mean 3.8 (#9). Outdated as a current description.
27 · Okonkwo · EISs average 575 pages · S · #10 (2013–18 sample)
28 · Okonkwo · Litigation 4.2 yrs; energy cases delayed 3.9 yrs; agencies win ~80% · S · #11
29 · Okonkwo · Queue median over 5 yrs; 75% attrition · S · #4–5
30 · Okonkwo · 345 kV+ avg 536 mi/yr (2020–24) vs ~5,000 needed · S · #6–7 (the need figure is an advocacy group's translation of DOE modeling)
31 · Okonkwo · Second Ave Subway 8–12× baseline · S · #44
32 · Brennan · Vogtle $36.8B · S · #33
33 · Brennan · NRC is "legally required" to decide within 18 months · W · This comes from an executive order (#36). The ADVANCE Act (#35) contains no such statutory deadline.
34 · Brennan · ADVANCE Act 88–2 / 393–13 · S · #35
35 · Brennan · NuScale $58 → $89/MWh · S · #34
36 · Harlow · NERC attributes risk to dispatchable retirements · C · We verified NERC's demand-growth headline (#20), not its retirement attribution. NERC frames the risk as demand outpacing resources.
37 · Harlow · Gas in queue +86% · S · #2
38 · Harlow · Prices rising "because of renewable mandates/integration" · U · No decomposition supports this. The brief's evidence points to gas, T&D and capacity costs (#22, #23, #26).
39 · Harlow · Exports 15.1 → 17.4 Bcf/d; Henry Hub $3.53 → $3.43 (f) · S · #40
40 · Harlow · "LNG exports don't raise domestic prices" · U · A one-year forecast can't establish a zero effect. Amended to "modest so far."
41 · Harlow · Thacker Pass tribes lost; case dismissed with prejudice · S · #41
42 · Pruitt · Florida "recovered because of" 2023 tort reform · C · Stabilization is documented (#38), but causation is not isolated from reinsurance cycles (#37). Amended.
43 · Pruitt · FAIR Plan grew ~fivefold · S · 124k → 663k (#39)
44 · Pruitt · CA barred catastrophe models/reinsurance pass-through before the Sustainable Insurance Strategy · S · #39 (secondary source)
45 · Solberg · 2024 median NOI→FEIS 2.2 yrs; 41% within 2 yrs · S · #9
46 · Solberg · SPEED preserves post-Jan 20, 2025 permit cancellations · S · #16
47 · Solberg · FRA 2023 deadlines "were working" before Seven County · C · Timelines were already falling in 2021–24 (#9), but no causal attribution to the FRA exists in the record

Tally: 47 claims rated

• Supported: 31
• Contested: 8 (#16, 20, 21, 26, 36, 42, 47, and #11-adjacent items are counted as U, not C; see the note below)
• Unsupported: 4 (#11, 24, 38, 40)
• Wrong: 5 (#2, 4, 5, 25, 33)

Correction by Takahashi on recount: the Contested count is 7 (#16, 20, 21, 26, 36, 42, 47). Final tally: 31 S / 7 C / 4 U / 5 W = 47.

---

Retractions and amendments on the record

• Lin: "I retract #2. The 2025 rise was weather- and gas-price-driven. My policy claim applies to the forward outlook. I retract #4; the correct figure is 4.4% in 2023. I retract #5 entirely and will not cite Kotz et al. My damages case rests on Rennert et al. and EPA 2023, with discount-rate sensitivity."
• Okonkwo: "I retract '55 miles' and substitute the 536-mile 2020–24 average. I amend '4.5 years on average' to 'median 2.5, mean 3.8 years (2021–24), with a long tail concentrated in linear infrastructure.'"
• Brennan: "I retract 'legally required.' The 18-month deadline is an executive order directive. That's why I propose codifying it."
• Harlow: "I withdraw 'because of renewable mandates' as a price explanation and keep the narrower claim about accreditation and firm capacity. I amend the LNG claim to 'domestic price effects modest so far,' tested by my $4.50 prediction."
• Delgado-Finch: "I amend to: 'In PJM, capacity costs driven by forecast large load are a major driver. Nationally, data centers are one of several drivers.'"
• Pell: "I withdraw the 'more than initial scores' assertion. The 47% was the total decline, not the ETS's causal effect."
• Pruitt: "'Stabilized after,' not 'recovered because of.'"
• Blackwood: "'Sometimes,' not 'repeatedly.'"

Referee note (Brandt): Nearly every error ran in the direction of the speaker's priors. Left speakers overstated current damages and data-center load, and right speakers overstated the role of renewables in prices. The abundance center used an uncorrected, dramatic transmission number. None of the errors came from nowhere: each misread a real source (a projection read as a current figure, a retracted paper, a superseded estimate).

STAGE 5 — STEELMAN

Each bloc writes the strongest version of an opposing bloc's case, and the opposing bloc responds.

---

5.1 The Right steelmans the Left

Written by Harlow (#92, R) and Pruitt (#8, Lib).

The Left's best case is not about 2025's emissions blip. It's about irreversibility and asymmetric risk.

- CO₂ accumulates, so every year of delay commits warming that cannot be recalled.

- Even the low end of peer-reviewed damage estimates ($80/t at 3%, Rennert, #28) is well above zero, which is what federal analysis now effectively uses (#27). Setting a known-positive externality to zero is an unforced analytic error, whatever your view of discount rates.

- Rhodium's forward range fell from 38–56% to 26–35% below 2005 by 2035 (#26). That is a policy choice with a measurable emissions consequence.

- The Left is also right that people living near projects bear concentrated costs while benefits are diffuse. If you strip litigation remedies (#16) without replacing them with something, you are asking the Standing Rock Sioux to trust an executive that just claimed the power to cancel permits it dislikes.

- Finally, the Left's point about cost causation is correct in market terms. If data centers trigger capacity costs (#23), charging residential customers for them is a cross-subsidy that a free-marketer should hate.

Left response (Lin, Blackwood, Delgado-Finch): "Accepted as fair, with one correction." Our case is not only about damages. It's also that clean-energy capital is already cheaper per MWh in many markets, and that the OBBBA phase-down raises costs for ratepayers in the near term. We have not proven that point in this thread's evidence record, so we accept the steelman as written and flag it as our research burden.

---

5.2 The Left steelmans the Right

Written by Lin (#84, L) and Blackwood (#79, L).

The Right's best case is reliability and physics before preferences.

- Peak demand is forecast to grow 224 GW in a decade, a forecast raised 69% in one year (#20). A grid that fails in a winter storm kills people. Unlike emissions, a blackout's harm is immediate and local.

- Firm capacity has a value that energy-only metrics miss, and gas remains the only firm resource that can be built at scale this decade.

- US LNG exports (#40) displace coal abroad and give allies an alternative to Russian gas. On that view, it is a climate and security asset.

- Second, the Right is right that subsidies are a poor substitute for markets: CBO scored $496B in savings from ending energy credits (#25), and those dollars subsidized projects that were often inframarginal.

- Third, and most uncomfortable for us: the Right is right that litigation delay is costly even when challengers lose. Energy cases took 3.9 years while agencies won 71% (#11). A process where losing still wins is a hold-up tool, and it has been used against transmission lines climate policy needs.

Right response (Harlow, Brennan, Pruitt): "Accepted as fair, with one correction." Pruitt notes that the libertarian case is not pro-gas. It is against all subsidies, including LNG loan support, DOE equity stakes (#41) and the capacity payments that prop up legacy plants. Harlow accepts the "coal displacement abroad" framing as ours but says the record here doesn't verify it.

---

5.3 The Center steelmans both flanks against itself

Written by Okonkwo (#81, C) and Solberg (#23, C).

The abundance center's weakness, stated fairly from both flanks:

- From the Left: "Permitting reform" in practice delivered Seven County and SPEED's remedy cuts (#13, #16), which apply equally to pipelines and LNG. The clean-energy titles (transmission cost allocation, interconnection) are the ones that keep getting dropped. The Left has seen EPRA die (#15) and SPEED pass without a transmission title (#16), and it reasonably concludes that the "grand bargain" is a one-sided trade.

- From the Right: The center overestimates what process reform buys. Vogtle's overrun (#33) and the Second Avenue Subway (#44) were execution failures, not permitting failures. The center keeps prescribing "state capacity" for a state that builds at 8–12× international costs.

Left response (Blackwood): "Accepted as fair."
Right response (Harlow): "Accepted. Add that the center's transmission needs number (#7) is an advocacy translation of a modeled scenario, and we shouldn't legislate against it as if it were measured demand."
Center rejoinder (Okonkwo): Accepted. The 5,000 miles/yr figure is a scenario, and I'll use it as an upper bound, not a target.

---

5.4 The Right steelmans the Center's carbon-price case

Written by Brennan (#83, C-R).

Pell's trade, a price in exchange for mandates and subsidies out, is the most market-conservative climate policy on offer. Canada kept industrial output-based pricing (#30), and the EU ETS survived for 20 years (#31). A conservative who takes climate risk as non-zero should prefer one transparent price to the IRA/OBBBA whiplash of credits created and then repealed.

Center response (Pell): "Accepted as fair."
Right majority (Harlow) dissent: A price will not stay revenue-neutral, and the Canadian and EU evidence (ETS2 delayed, consumer price repealed) shows it will not stay stable either.

STAGE 6 — REVISED PROPOSALS

Scoring note: none of these proposals has a CBO/JCT score in this exact form. Where a related official score exists, we cite it. Otherwise we say "no official score."

---

P1. Two-way permit certainty: a 150-day statute of limitations plus protection against arbitrary revocation

Sponsors: Solberg (C), Okonkwo (C), with Blackwood's (L) carve-out.

Mechanism.

• A 150-day statute of limitations for NEPA challenges, as in EPRA and SPEED (#15, #16).
• Remand without vacatur as the default remedy. Courts may still vacate on a showing of imminent irreparable harm to treaty rights, sacred sites or public health.
• Codify Seven County's scope rule (#13).
• Strike SPEED's carve-out for cancellations since January 2025 (#16). Instead, bar executive revocation of issued permits except for documented violations or new safety information, with judicial review.
• Fund federal and tribal staff for early, time-limited consultation (see P6).

Cost. No official score. The staffing component likely costs low hundreds of millions per year; we could not verify a figure.

Precedent. EPRA (15–4 in committee, #15) and SPEED (221–196, #16) both contain the 150-day core. No enacted precedent exists for the anti-revocation provision.

Key risk. Remand-without-vacatur leaves flawed reviews standing. DAPL shows even vacatur rarely stopped operations (#42), so the marginal harm may be small, but it is not zero.

---

P2. Interregional transmission: federal backstop siting, a minimum transfer capability, and broad cost allocation

Sponsors: Okonkwo (C), Ashby (C, as specialist, non-voting on authorship), Delgado-Finch (Soc).

Mechanism.

• FERC gets backstop siting authority for lines that have been stalled at the state level for more than 2 years.
• A statutory minimum interregional transfer capability, per EPRA's approach (#15).
• Beneficiary-pays cost allocation across regions.
• The package is paired with pipeline permitting on the same statute-of-limitations terms (the political price of right-bloc votes).

Cost. No official score. The capital cost is borne by ratepayers, not the federal budget.

Precedent. The 2010–14 build rate of 1,781 miles/yr (#6) shows the US has built at three times today's pace. The 2024 revision to 888 miles (#6) shows the pace can recover.

Key risk. Cost allocation fights shift to FERC litigation. Backstop siting was weakened by courts after 2005 (not re-verified this session).

---

P3. Interconnection and large-load reform: "connect and manage," surplus interconnection, and large loads pay their own way

Sponsors: Ashby/Harlow/Delgado-Finch cross-bloc.

Mechanism.

• Allow generators to connect on an energy-only, curtailable basis while network upgrades are built.
• Fast-track surplus and replacement interconnection at existing plant sites (retiring coal plants).
• Hard deadlines for grid operator studies.
• A mandatory large-load tariff: loads above 50 MW pay incremental generation and transmission costs, post collateral, and accept curtailment during emergencies. This is modeled on the House-passed Ratepayer Protection Act (417–3, #17).

Cost. No official score. Largely regulatory.

Precedent. Texas's grid (ERCOT) has used a connect-and-manage style approach (not re-verified this session). Texas's August 2026 data-center pause cut EIA's 2027 Texas growth forecast from 14% to 6% (#18), which shows large-load policy materially changes demand.

Key risk. Curtailment risk raises financing costs for renewables. Data centers may relocate to jurisdictions without the tariff.

---

P4. Firm clean power: codify NRC licensing timelines, plus a capped, clawback-protected cost-overrun backstop for standardized reactors

Sponsor: Brennan (C-R), amended in response to Lin.

Mechanism.

• Put the 18-month licensing decision deadline into statute (currently an executive order directive, #36), with NRC staffing funded.
• A federal overrun backstop covering the first 10 units of standardized designs: 50% of overruns above a 30% contingency, capped at $2B per unit.
• The program ends if unit 2 of any design does not beat unit 1's real $/kW by 20% or more.
• Technology-neutral credits for firm clean power (nuclear, geothermal, long-duration storage).

Cost. No official score. The backstop's maximum exposure is ~$20B over 15 years by construction.

Precedent. Vogtle ($36.8B vs $4.4B, #33) is the cautionary FOAK case. The ADVANCE Act (88–2, #35) is the political precedent.

Key risk. Moral hazard, and the history of NuScale's cost escalation (#34).

---

P5. Upstream carbon fee-and-dividend with a border adjustment, in exchange for preempting EPA power-plant greenhouse gas rules and ending the remaining clean credits

Sponsor: Pell (C), amended after Stage 3.

Mechanism.

• $50/t at the upstream level (mine, well, port) rising 5% per year above inflation.
• 100% of revenue returned as an equal per-capita dividend paid quarterly and shown on utility bills.
• A border adjustment compatible with CBAM (#31).

Cost. Revenue-neutral by design. No official score for this version.

Precedent. The EU ETS: covered emissions −47% overall, with the attributable effect ~3.8–11.5% (#31). Canada's consumer price was repealed in 2025, while the industrial OBPS survived (#30). BC evidence is unverified this session (#32).

Key risk. Political durability (Canada). Pressure to spend the revenue rather than return it.

---

P6. Funded early tribal consultation with a binding timeline

Sponsor: Blackwood (L), co-signed by Okonkwo (C).

Mechanism.

• A mandatory consultation window before NOI for projects crossing treaty lands or known sacred sites.
• Federal grants for tribal technical review capacity.
• A 12-month consultation clock. Its findings create a record that courts must weigh in any post-ROD challenge.
• An optional tribal equity or benefit-sharing framework.

Cost. No official score.

Precedent. DAPL (#42) and Thacker Pass (#41) both show that late consultation produced years of conflict without changing outcomes.

Key risk. Consultation turns into a de facto veto, or into box-checking.

---

P7. Insurance: risk-based pricing, catastrophe-model and reinsurance pass-through, and means-tested mitigation vouchers

Sponsors: Pruitt (Lib), Lin (L), cross-bloc.

Mechanism.

• States permit catastrophe models and reinsurance cost pass-through.
• States wind down rate caps that force residual-market growth.
• Federal block grants for means-tested home hardening vouchers, with premium credits mandated for verified mitigation.
• Residual markets (Citizens, FAIR Plan) priced at actuarial rates for new policies.

Cost. No official score for vouchers.

Precedent.

• Florida: Citizens fell from 1.3M to under 800k policies, and premiums fell 0.7% in Q4 2024 after reform (#38).
• California: the FAIR Plan hit 663k policies and $768B exposure (#39), then the state adopted the Sustainable Insurance Strategy.

Key risk. An affordability shock and >$40k home-value losses in exposed ZIP codes (#37). The political backlash could reimpose caps.

---

P8. Restore the IRA clean electricity credits (45Y/48E) for wind and solar through 2032

Sponsor: Lin (L).

Mechanism. Repeal OBBBA's July 4, 2026 / end-2027 deadlines (#24). Keep the FEOC restrictions.

Cost. Roughly the reverse of CBO's $166B clean electricity investment credit savings (#25), so on the order of $150–170B over 10 years.

Precedent. The IRA 2022–2025 period.

Key risk. Fiscal cost, inframarginal subsidy, and policy whiplash if repealed again.

STAGE 7 — RECORDED VOTE

Blocs: Left 34 · Center 33 · Right 32. The moderator (#100) abstains.
Bridge rule: a proposal is CONSENSUS only if it gets ≥60% in every bloc (Left ≥21, Center ≥20, Right ≥20).

Proposal · Left yes · Center yes · Right yes · Consensus?
P1 Two-way permit certainty (150-day SOL + anti-revocation + carve-out) · 20 / 34 (58.8%) · 29 / 33 (87.9%) · 29 / 32 (90.6%) · NO (Left short by 1)
P2 Interregional transmission backstop siting + cost allocation, paired with pipeline permitting · 31 / 34 (91.2%) · 30 / 33 (90.9%) · 20 / 32 (62.5%) · YES (narrow on Right)
P3 Connect-and-manage + surplus interconnection + large-load tariff · 29 / 34 (85.3%) · 31 / 33 (93.9%) · 26 / 32 (81.3%) · YES
P4 Codified NRC timeline + capped overrun backstop w/ clawback · 13 / 34 (38.2%) · 24 / 33 (72.7%) · 27 / 32 (84.4%) · NO
P5 Upstream carbon fee-and-dividend + border adjustment, in exchange for preempting rules and ending credits · 24 / 34 (70.6%) · 22 / 33 (66.7%) · 7 / 32 (21.9%) · NO
P6 Funded early tribal consultation with a 12-month clock · 32 / 34 (94.1%) · 24 / 33 (72.7%) · 12 / 32 (37.5%) · NO
P7 Risk-based insurance pricing + means-tested mitigation vouchers · 21 / 34 (61.8%) · 28 / 33 (84.8%) · 23 / 32 (71.9%) · YES (at the Left threshold)
P8 Restore 45Y/48E wind and solar credits through 2032 · 32 / 34 (94.1%) · 14 / 33 (42.4%) · 2 / 32 (6.3%) · NO

Consensus reforms: P2, P3, P7. P1 missed by one Left vote.

---

Vote explanations

Prof. Aaron Blackwood (#79, L). YES on P1, P6; NO on P4.

I voted for P1, against much of my bloc, because it strikes SPEED's revocation carve-out and keeps vacatur for treaty and sacred-site harms. That is better than the status quo, where the remedy is formally strong and practically empty (DAPL, #42). Thirteen Left colleagues voted no because they don't trust "imminent irreparable harm" to be applied generously. What would change my mind: if courts under Seven County (#13) start denying vacatur even for documented sacred-site destruction, I'll move to opposing any SOL cap.

Prof. Rosa Delgado-Finch (#7, Soc). NO on P1, P5, P7; YES on P2, P3.

P5 and P7 are market fixes to problems created by markets. Risk-based insurance pricing without a public reinsurance option leaves working-class homeowners with the >$40k capital loss (#37). I voted for P2 and P3 because they put cost causation where it belongs, on large loads (#23). What would change my mind on P7: evidence that Florida's post-reform stabilization (#38) held for low-income policyholders specifically, not just the average premium.

Prof. Dennis Harlow (#92, R). YES on P1, P2, P3, P4; NO on P5, P6, P8.

I voted for P2 only because it's paired with pipeline permitting. Without that pairing, a majority of my bloc votes no, and the 20 Right votes would drop to about 12. P3 is the real reliability fix. Gas doesn't skip the queue either (Ashby's point). What would change my mind on P5: a carbon price that is constitutionally or statutorily locked to revenue neutrality and actually repeals the power-plant greenhouse gas rules. I don't believe either is enforceable.

Prof. Stuart Pell (#85, C). YES on P1, P2, P3, P5, P7; NO on P8; YES on P4.

I voted against P8 because subsidy whiplash is the worst of both worlds. CBO's $166B (#25) buys emissions cuts at an unknown and probably high $/t. P5 is the better instrument even though it lost. What would change my mind on P8: a credible study showing that the post-OBBBA drop in wind and solar additions raises consumer electricity costs by more than the credit's fiscal cost. That's Mei's burden, and she acknowledged it.

Prof. Mei Lin (#84, L). YES on P2, P3, P5, P6, P7, P8; NO on P1, P4.

I voted against P4 not because nuclear is bad but because a $20B overrun backstop for one technology, while wind and solar lose credits, isn't neutral. What would change my mind: if P4 were paired with P8, or if any standardized second unit meets Kyle's 20% cost-decline test, I'd vote yes. I voted for P5 despite my bloc's skepticism because it's the one instrument that survives changes of administration better than regulation does (the EU, #31).

Prof. Walter Pruitt (#8, Lib). YES on P1, P3, P7; NO on P2, P4, P6, P8; NO on P5.

P4's backstop and P8's credits are both corporate welfare. P2's backstop siting is eminent-domain power I can't support. P5 lost me when the revenue went to dividends and a border adjustment. That's trade policy wearing a green hat. What would change my mind on P2: a compensation standard for affected landowners above fair market value, which would make siting a transaction rather than a taking.

STAGE 8 — VERDICT

Moderator: Prof. Adelaide Wainwright (#100, political science & institutional design, C; abstaining)

Established (the evidence is strong enough that no bloc disputed it after cross-examination)

1. The grid-connection queue is the largest single measured bottleneck.

• Median wait from request to operation is over 5 years (#4).
• 75% of 2000–2020 capacity withdrew (#5).
• 549 GW already has interconnection agreements but isn't built (#3).
• Gas faces the same queue (#2). This finding is ideology-neutral.

1. High-voltage transmission construction collapsed after 2013. 345 kV+ build fell from 1,781 miles/yr (2010–14) to 536 (2020–24) (#6). The dramatic "55 miles" figure was revised to ~450, and anyone still quoting it is wrong.
2. NEPA timelines have improved, but the tail is real.

• Median NOI→ROD was 2.5 years in 2021–24, with a mean of 3.8 (#9).
• Energy litigation added 3.9 years even though agencies won 71% (#11).
• The "4.5 years average" is a 2010–18 figure.

1. Seven County (8–0 judgment, 5–3 on reasoning) and the CEQ rescission already narrowed NEPA substantially (#13–14), before any statute.
2. Demand is growing after two flat decades. Consumption was 4,195 BkWh in 2025, forecast at 4,391 BkWh for 2027 (#18). Data centers were 4.4% of load in 2023 (#19). Forecasts are volatile: NERC raised its forecast 69% in one year, and EIA halved Texas's 2027 growth forecast (#18, #20).
3. US emissions rose 2.4% in 2025 because of weather and gas prices, not (yet) OBBBA (#26). Rhodium's forward outlook did worsen materially (#26).
4. Nuclear FOAK costs were catastrophic: Vogtle cost $36.8B against a $4.4B estimate, and NuScale's CFPP rose to $89/MWh after subsidies (#33–34).
5. California's residential insurance market shows a price-control shortage layered on rising risk (#37, #39).

Contested (a reasonable reading of the evidence divides the blocs)

• Why retail prices are rising. Gas prices, T&D, capacity costs and data centers each have support. No decomposition exists.
• Whether NEPA or interconnection is the binding constraint for transmission specifically (RFF vs. Breakthrough; #11–12).
• The causal effect of carbon pricing (3.8% to 11.5% for the ETS; #31), and whether it survives politically (Canada, #30).
• The value of the social cost of carbon, mostly a discount-rate dispute ($80 vs $185; #28). But a value of effectively zero (#27) is not a defensible estimate, and even the Right's steelman conceded that.
• Whether LNG exports materially raise domestic gas prices over a 5-year horizon (#40).
• Whether Florida's stabilization came from tort reform or the reinsurance cycle (#37–38).

Unknown

• Whether data-center load materializes at the top or bottom of LBNL's 6.7–12% range (#19).
• Whether nuclear costs fall on the second US unit. No US data point exists.
• How courts will apply Seven County's "substantial deference" to agency-specific NEPA procedures written after CEQ's rescission.

Research agenda

1. A decomposition of 2024–27 retail price changes by driver, by region: gas, T&D, capacity, wildfire liability, large loads, and renewables integration.
2. Project-level tracking of how SPEED/*Seven County*-era reviews change outcomes (routes, mitigation), not just timelines. This tests Blackwood's deterrence claim.
3. An ex-post evaluation of connect-and-manage in ERCOT versus PJM's study-first approach.
4. A replication of damage-function estimates after the Kotz retraction (#29).
5. Low-income distributional outcomes of Florida's insurance reform.

Consensus reforms (≥60% in every bloc)

• P2, interregional transmission with backstop siting, only because it was paired with pipeline permitting. That pairing is the actual bridge.
• P3, connect-and-manage plus large-load cost allocation. This is the strongest cross-bloc result, echoing the House's 417–3 vote (#17).
• P7, risk-based insurance pricing with means-tested mitigation vouchers. It barely cleared the Left.
• Near miss: P1 (two-way permit certainty), one Left vote short. The anti-revocation clause pulled the Center and Right into supporting the Left's concern about executive cancellations. That's a live deal the Senate is currently failing to make (#17).

Where Symposium 1 was wrong or too glib on this domain

• Too glib that "permitting reform" means NEPA reform. In this record the queue, cost allocation and execution costs (#4, #44) matter at least as much. NEPA has already been narrowed by the Court and CEQ.
• Wrong to repeat the "55 miles" transmission figure and "EISs take 4.5 years." Both are superseded.
• Too confident about damages. Symposium 1 leaned on a now-retracted paper for headline damages. The defensible range is the discount-rate-sensitive $80–$190.
• Too glib that data centers "are" the price story. They are a PJM story with national spillover, not the whole national story.
• Too dismissive of reliability concerns from the Right. NERC's 224 GW forecast (#20) is real, even though the Right's attribution to renewables was unsupported.

Three-sentence verdict. The United States' binding energy constraints are physical and institutional: a 5-year interconnection queue, transmission build at roughly a tenth of modeled need, and FOAK construction costs. NEPA, already narrowed by Seven County and CEQ's rescission, is a real but secondary tail risk. The strongest cross-bloc agreement is on making large loads pay their own way, fixing interconnection, and building interregional transmission, but only when paired with pipeline permitting. Carbon pricing, tax credits, nuclear backstops and tribal-consultation standards remain partisan because they rest on contested values (discounting, subsidy, consent) that more data alone won't settle.

— A. Wainwright, #100

Independent Audit

Audit Panel: Prof. Laura Pennington (#97, operations research, C), lead, and Prof. Olivia Brandt (#57, statistics, C). Brandt sat on this thread's referee panel, so she recuses from judging any Stage 1 or Stage 4 ruling that was hers. Pennington made every finding below on those items. Corrections are appended. Nothing above has been deleted or edited.

1. Source check (16 claims, taken from the Evidence Brief and the Verdict)

# · Claim · Figure in thread · Verified figure · Status · URL
1 · Interconnection queue, end-2025 · 1,312 GW gen + 749 GW storage; −10% y/y; gas 253 GW (+86%) · Same · Confirmed · emp.lbl.gov/publications/queued-2026-edition-characteristics
2 · Signed but unbuilt; wait; attrition · 549 GW; median >5 yrs; 13% built / 75% withdrawn · Same · Confirmed · same
3 · 345 kV+ miles · 1,781 / 942 / 536; 2023: 55 revised to ~450; 2024: 322 revised to 888 · Same (2023 = 400 mi at 345 kV + 50 mi at 500 kV) · Confirmed · gridstrategiesllc.com …Fewer-New-Miles-2025_vF.pdf
4 · Need of ~5,000 mi/yr; NTPS growth 2.1–3.3× · as stated · Base case 2.1–2.6×, high-demand case 2.5–3.3× · Confirmed · same
5 · CEQ EIS timelines · median 2.8 yrs "overall"; 2.2 in 2024; NOI→ROD 2.5/3.8 vs 3.5/4.5; 41% vs 24% · The 2.8-yr median covers 2019–2024, not the whole 2010–24 sample. Everything else matches. · Minor discrepancy · nepa.gov …CEQ_EIS_Timeline_Report_2025-1-13.pdf
6 · Breakthrough litigation study · 387 cases; ~80% agency wins; 4.2 yrs; energy 29%, 3.9 yrs, 71% wins; NGOs 72% · Same · Confirmed · thebreakthrough.org/issues/energy/understanding-nepa-litigation
7 · RFF clean-energy NEPA · solar 27 mo, wind 45 mo; ~10% / 3.7% of capacity; 9 of 24 and 7 of 14 took 4+ yrs · Same · Confirmed · rff.org …how-long-does-it-take…
8 · Seven County · Kavanaugh +4; Sotomayor, Kagan, Jackson concur in judgment; Gorsuch recused; "substantial deference" · Same · Confirmed · supreme.justia.com/cases/federal/us/605/23-975/
9 · SPEED Act · 221–196, Dec 18 2025; 150 days; remand only; commenter standing; cancellation carve-out · Same · Confirmed · bipartisanpolicy.org/issue-brief/whats-in-the-speed-act/
10 · Data-center load · 176 TWh = 4.4% (2023); 325–580 TWh = 6.7–12% (2028) · Same · Confirmed · newscenter.lbl.gov/2025/01/15/…
11 · NERC 2025 LTRA · +224 GW (24%); 69% above prior forecast; winter +246 GW; MISO, PJM, ERCOT, PNW · Same. The source also cites >105 GW of planned retirements and a 21 GW drop in fossil capacity as contributing factors (see §2c). · Confirmed · utilitydive.com/news/nerc-10-year-peak-demand-forecast…
12 · Retail prices, July 2026 · 14.99¢, +4.4%; residential +4.9% · Same. Residential and transportation are tied at +4.9%. · Confirmed · eia.gov/electricity/monthly/update/end-use.php
13 · Rhodium 2025 · +2.4%; 18% below 2005; power +3.8%; coal +13%; Henry Hub +58%; buildings +6.8% (cold winter); 2035 outlook 26–35% (was 38–56%) · Same · Confirmed · rhg.com/research/us-greenhouse-gas-emissions-2025/
14 · PJM capacity costs · $2.2B (2024 auction) → >$16B (2025), "~7×"; Dec 2025 shortfall · The Wikipedia text supports $2.2B → >$16B and the Dec 2025 shortfall, but says "eight times". Auditor's note, not verified against PJM primary reports this session: the $2.2B auction was for delivery year 2024/25 and the ~$16B auction for 2026/27, so the jump spans two auction cycles, with ~$14.7B in between. Delgado-Finch's "in one auction cycle" is therefore probably overstated. · Minor discrepancy (secondary source) · en.wikipedia.org/wiki/PJM_Interconnection
15 · CBO/OBBBA savings · $190B / $166B / $77B; $496B energy total · Same · Confirmed · pgpf.org/article/energy-tax-policy-under-the-obbba/
16 · Vogtle 3 & 4; nuclear EOs · $36.8B vs $4.4B, 15 yrs; 400 GW by 2050, 10 reactors by 2030, 18-month deadline set by EO (May 23 2025) · Same · Confirmed · georgiarecorder.com/2024/05/31/…; energy.gov/ne/articles/9-key-takeaways…

Counts: 14 Confirmed · 2 Minor discrepancy · 0 Not supported · 0 Could not access.

2. Internal consistency

a. Fact-check tally. The first tally line reads "Contested: 8 (#16, 20, 21, 26, 36, 42, 47, and #11-adjacent items…)". It lists only seven items, and #11 is rated U in the table. Takahashi's recount is correct. We re-counted every row: S = 31, C = 7 (#16, 20, 21, 26, 36, 42, 47), U = 4 (#11, 24, 38, 40), W = 5 (#2, 4, 5, 25, 33), total 47. Correction: treat the "Contested: 8" line as superseded by the recount note. The final tally is 31 S / 7 C / 4 U / 5 W.

b. Vote math. We checked all 24 cells (yes-count ÷ bloc size) and every percentage is correct to one decimal place. Consensus labels are correct. P2 passes (Right 20/32 = 62.5%). P3 passes. P7 passes (Left 21/34 = 61.8%, exactly at the threshold). P1 misses by one Left vote (20/34). The rest fail. One count error: Blackwood's explanation says "Thirteen Left colleagues voted no" on P1. With 20 of 34 voting yes, 14 Left members voted no; Lin and Delgado-Finch both did. Correction: "Fourteen".

c. Verdict against the evidence.

• (i) The three-sentence verdict says transmission is being built at "roughly a tenth of modeled need". That uses 536 mi/yr against 5,000. In Stage 5, Okonkwo accepted that 5,000 is an upper bound from an advocacy group's reading of a scenario, and the revised 2024 figure (888) is ~18% of it. Suggested wording: "at roughly one-tenth to one-fifth of an advocacy-derived upper-bound scenario."
• (ii) Fact-check #36 rated C Harlow's claim that "NERC attributes risk to dispatchable retirements". The referees said they hadn't verified it. The Utility Dive piece they cited does name >105 GW of expected retirements and a 21 GW fall in fossil capacity as contributing factors, while keeping demand growth as the headline. C is still the right rating, since demand is the main framing. But the basis should say the cited source partly supports the claim, not that the claim is unverified.
• (iii) No claim rated Wrong comes back as established. The Kotz retraction, the "55 miles" figure, "legally required" and the 12% data-center share are all handled correctly in the Verdict.
• (iv) Names and fields match across all stages. Pruitt is #8 (Lib) and Wainwright is #100 (abstaining), consistent with the other threads.

3. Balance

The five Wrong ratings are spread across the spectrum: Lin ×3 (L), Okonkwo (C), Brennan (C-R). All five are backed by the cited sources: a retracted paper, a superseded number, a projection read as a current figure, and an EO described as a statute. The Left's three Ws are the most in any bloc, but each one is clean. The Right drew two U ratings (#38, #40). Both were fair, since neither had a decomposition or elasticity behind it. #36 is the one place the referees were stricter with the Right than the record required (see 2c-ii). The Verdict's "Too dismissive of reliability concerns from the Right" section offsets this in part. The Verdict does not tilt beyond the evidence: its contested and unknown sections line up with the brief's contested items (i)–(vii).

4. Grade: A−

Sources were accurate in 14 of 16 checks, and the vote arithmetic has no errors. We deducted for a messy tally line that needed a recount, one miscount in a vote explanation, and a PJM "one cycle" framing that probably overstates the speed of the jump.

— L. Pennington (#97), lead; O. Brandt (#57), recused on Stage 1 and Stage 4 rulings

Verdict v2 (audit-corrected)

Prof. Adelaide Wainwright (#100), moderator. This post adds to the Stage 8 verdict (post 244) and the Independent Audit (post 249). Neither of those posts is edited. Every open item was re-checked on 2026-09-24. Ratings use the standard scale: Supported / Contested / Unsupported (the last includes "could not verify") / Wrong.

1. Unverified and audit-flagged items: resolutions

Item · Origin · Result · Figure and source
British Columbia carbon tax (Brief #32) · "[unverified this session]" · Now verified · Murray & Rivers (Energy Policy 86, 2015): "Empirical and simulation models suggest that the tax has reduced emissions in the province between 5% and 15%" and "negligible effects on the aggregate economy." IDEAS/RePEc
Harlow: NERC attributes risk to dispatchable retirements (#36) · Rated C as "not verified" · Now verified (as a contributing cause) · NERC's 2025 LTRA, as reported, cites >105 GW of confirmed and announced retirements over 10 years. Fossil capacity fell 21 GW (2024–25) against 23 GW of renewable additions. Capacity fell short of projections because of "delays in connecting new resources and unanticipated generator retirements." Solar and storage, two-thirds of planned additions, have weak winter capability. Demand growth remains the headline. Utility Dive
PJM capacity costs (#23; Delgado-Finch #9) · Secondary source; audit flagged "two cycles" · Corrected with primary data · $2.2B (2024/25 BRA) → $14.7B (2025/26 BRA, July 2024): a ~6.7× jump in one auction (PJM 2025/26 BRA report). Then $16.1B (2026/27, at the price cap) (PJM). Then $16.4B (2027/28, Dec 17 2025), which fell 6,623 MW short of the reliability requirement (PJM Inside Lines). "$2.2B → >$16B" therefore spans two auctions. Delgado-Finch's "one auction cycle" holds for the ~6.7× step to $14.7B, so the claim stays S with the corrected figure.
Federal backstop siting weakened by courts after 2005 (P2) · "not re-verified" · Now verified · Piedmont Env. Council v. FERC (4th Cir. 2009) held that FERC cannot override an outright state denial. Cal. Wilderness Coalition v. DOE (9th Cir. 2011) vacated the 2007 corridor designations. The 2021 IIJA sought to revive §216 authority. Wikipedia summary of cases
ERCOT connect-and-manage (P3) · "not re-verified" · Now verified · ERCOT studies only local upgrades and manages congestion by curtailment. It takes about 3.5 yrs to operation, against 6+ yrs in PJM and SPP. ERCOT connected 14.2 GW in 2021–22 versus 5.6 GW in PJM. The trade-off was 2022 curtailment of about 9% of solar and 5% of wind. Utility Dive, citing LBNL
CEQ "2.8-yr median overall" (#9) · Audit · Corrected · The 2.8-yr median NOI→FEIS covers 2019–2024, not 2010–24 (CEQ, Jan 2025).
Blackwood: "Thirteen Left colleagues voted no" on P1 · Audit · Corrected · Fourteen.
"Transmission at roughly a tenth of modeled need" (Verdict) · Audit · Corrected · "Roughly one-tenth to one-fifth of an advocacy-derived upper-bound scenario" (536 or 888 mi/yr vs ~5,000).
P1 staffing cost ("low hundreds of millions") · Stage 6 · Still unverifiable · No official score exists.
DOE 2024 LNG study; "LNG displaces coal abroad" · Stage 3 / 5.2 · Still unverifiable · Not retrieved. No number is entered into evidence.
Solberg: FRA 2023 deadlines "were working" (#47) · Rated C; no attribution in the record · Still unverifiable · Timelines fell in 2021–24, but nothing in the record attributes the decline to the FRA.

2. Rating normalization and revised tally

• #36 (Harlow): C → S. NERC names retirements as a contributing cause. The qualifier is that demand growth, not retirements, is NERC's headline, and Harlow's "comes from" overstates retirements as the sole source.
• #47 (Solberg): C → U. The claim was merely unverified, not disputed by any source.
• #9 (Delgado-Finch): remains S, with the primary-source correction above.
• All other ratings are unchanged. #16, #20, #21, #26 and #42 are genuine disputes, overstatements or superseded figures, so they stay C.

Tally: 31 S / 7 C / 4 U / 5 W → 32 S / 5 C / 5 U / 5 W (47 claims). The one strictness the audit found against the Right (#36) is now removed.

3. Corrected verdict

Established

1. The interconnection queue is the largest single measured bottleneck. The median wait from request to operation is over 5 years. 75% of 2000–2020 capacity withdrew, and 549 GW holds signed agreements but is unbuilt. Gas faces the same queue. New: ERCOT's connect-and-manage approach reaches operation in about 3.5 years and connected 14.2 GW in 2021–22, against PJM's 5.6 GW, at the cost of higher curtailment.
2. High-voltage transmission construction collapsed after 2013. It ran at 1,781 mi/yr (2010–14) and 536 (2020–24), with 2024 revised up to 888. (Corrected.) That is roughly one-tenth to one-fifth of an advocacy-derived upper-bound scenario of ~5,000 mi/yr. It is not a measured shortfall.
3. NEPA timelines have improved, but the tail is real. The median NOI→ROD for 2021–24 is 2.5 years and the mean 3.8. The median NOI→FEIS is 2.8 years for 2019–24 and 2.2 in 2024. Energy litigation added 3.9 years even though agencies won 71% of cases.
4. Seven County (8–0 judgment, 5–3 on reasoning) and the CEQ rescission had already narrowed NEPA before any statute.
5. Demand is growing after two flat decades. Consumption was 4,195 BkWh in 2025 and is forecast at 4,391 in 2027. Data centers used 4.4% of electricity in 2023. Forecasts are volatile.
6. Reliability risk is real, and it comes from both demand and supply. (Revised.) NERC's 10-year summer peak is +224 GW. NERC also cites >105 GW of expected retirements, a 21 GW fossil decline against 23 GW of additions, and weak winter capability in the solar and storage additions. The Right's reliability attribution to retirements is supported as a contributing cause. Its price attribution to renewables is still unsupported.
7. PJM capacity costs jumped about 6.7× in one auction (corrected): $2.2B to $14.7B, then $16.1B and $16.4B. The Dec 2025 auction fell 6,623 MW short of the reliability requirement.
8. U.S. emissions rose 2.4% in 2025 because of weather and gas prices, not (yet) OBBBA. The forward outlook worsened.
9. First-of-a-kind nuclear costs were catastrophic: Vogtle at $36.8B vs $4.4B, and NuScale's CFPP at $89/MWh.
10. California's insurance market shows a price-control shortage layered on rising risk.

Contested

• Why retail prices are rising. No decomposition exists.
• Whether NEPA or interconnection binds for transmission specifically.
• The causal effect of carbon pricing. The EU ETS share is ~3.8–11.5%. Now verified: British Columbia's tax cut emissions an estimated 5–15% with negligible aggregate economic effect. Political durability is a separate question (Canada).
• The value of the social cost of carbon ($80 vs $185; zero is not an estimate).
• LNG exports' effect on domestic gas prices.
• Florida's tort reform versus the reinsurance cycle as the cause of stabilization.

Unknown

• Where data-center load lands within the 6.7–12% range.
• Whether nuclear costs fall on the second unit.
• How courts will apply Seven County to agency-specific procedures.
• Whether the FRA 2023 deadlines caused the 2021–24 timeline improvement.

Conclusions that changed

• (a) Carbon pricing. "BC evidence is modest but positive" was an unverified placeholder. It is now a peer-reviewed 5–15% range. That strengthens the evidence for P5's effectiveness, though not its political durability.
• (b) "Too dismissive of reliability concerns from the Right" gets stronger. NERC's retirement framing partly vindicates Harlow on reliability, though not on prices.
• (c) Transmission is not "a tenth of need." It runs at one-tenth to one-fifth of a scenario-based upper bound.
• (d) P2's key risk is confirmed. Courts did gut the 2005 backstop, which is why the statute must be explicit.
• (e) P3's ERCOT precedent is now verified, including its curtailment cost.
• (f) The consensus results recorded in Stage 7 are unchanged.

— A. Wainwright (#100)