11 — Defense, Foreign Policy & National Security

11 — Defense, Foreign Policy & National Security

Stage 1: EVIDENCE BRIEF

Referee panel: Prof. Olivia Brandt (#57, statistics, C) · Prof. June Takahashi (#17, econometrics, C) · domain specialist Prof. Dmitri Volkov (#33, national security, C-R)

Ground rules from the referees. Today is 2026-09-24. Figures are taken from the source as stated, with the year of the data. Where we could not verify a figure in this session we say so. One warning up front: in this domain, "the defense budget" means at least four different numbers. There is the NDAA authorization, the appropriation, the mandatory reconciliation money, and SIPRI's broader "military expenditure." Anyone who mixes them without saying so will be flagged. — OB

---

A. Budget level and trend

1. FY2026 NDAA topline: $890.6 billion for national defense. That is $8.0 billion above the President's $882.6 billion request. It was signed December 18, 2025 as P.L. 119-60. The CRS notes that it sits alongside the mandatory defense money in Title II of the 2025 reconciliation act (P.L. 119-21). — CRS IN12641 (2025)
2. FY2026 defense appropriations were enacted January 30, 2026, four months into the fiscal year. The act gives $866.6 billion, which AEI describes as "$18 billion over the President's request." — AEI, McCusker & Ferrari (2026)
3. 2025 reconciliation act (P.L. 119-21): $150 billion for defense. The Senate Armed Services Committee's allocation in the enacted version was $29B shipbuilding/maritime industrial base, $25B Golden Dome, $25B munitions (plus $5B critical-minerals supply chains), $16B drones/AI/low-cost weapons, $15B nuclear modernization, $12B Indo-Pacific, and $9B air superiority. — Breaking Defense (Jul 2025)
4. DoD's 2026 request totaled $961 billion, including $113 billion from reconciliation. Most of the reconciliation share went to acquisition. Smaller amounts went to operation and support and to infrastructure "without corresponding decreases in base-budget funding." — CBO (Apr 30, 2026)
5. FY2027 request: $1.5 trillion. That is about $1.15T discretionary national defense plus $350B DoD reconciliation. CSIS calls it a 24% real increase in discretionary funding and a 38% real increase overall versus FY2026. Outlays would go from 3.0% of GDP (FY26) to 3.6% (FY27), and budget authority would be 4.6% of GDP. — CSIS, Daniels (Apr 2026)
6. CBO on the FY27 budget: total defense funding would rise "by $599 billion (or 66 percent) for 2027." Defense outlays would reach 3.4% of GDP in 2030 and then fall to 2.8% by 2036. CBO's current-law projection for 2036 is 2.4%. — CBO (Jun 2026)
7. SIPRI 2025: world military expenditure was $2,887 billion, up 2.9% in real terms. The US spent $954 billion, 7.5% lower than in 2024, which was 3.1% of GDP (down from 3.4%) and about a third of the world total. SIPRI attributes the drop to the absence of supplemental money for military aid to Ukraine or Israel in 2025. — SIPRI press release; SIPRI fact sheet (Apr 2026)
8. Other big spenders, 2025 (SIPRI): China was about $336B, up 7.4%, its 31st straight annual increase. Russia spent $190B, which is 7.5% of GDP. Ukraine spent $84.1B, about 40% of GDP. European spending rose 14% to $864B. Germany reached $114B (+24%, 2.3% of GDP). The 32 NATO members together spent $1,581B, 55% of the world total. — same sources
9. CBO long-run cost growth: if DoD's costs grow at historical rates, the FYDP understates them by about 4% for 2025–29 ($154B) and 5% for 2025–39 ($677B). Under the FYDP, costs rise from $850B (2025) to $965B (2039) in 2025 dollars. — CBO (Nov 2024)

B. Procurement, sustainment, audit

1. GAO 2026 Weapon Systems Annual Assessment: it covers 104 of DoD's costliest programs, worth over $2.4 trillion. "The overall average time frame to deliver a capability increased this year to over 12 years." Of 40 programs on the rapid "middle tier" pathway, 18 entered with immature technologies (2018–25). — GAO-26-108457 (Jul 2, 2026)
2. F-35 sustainment: US-only lifecycle sustainment is estimated at $1.6 trillion (as of 2024). In FY2025 the fleet was 44% mission capable and 25% fully mission capable, against 2030 goals of 80% and 65%. Steady-state annual cost is $7.4M per F-35A against an affordability target of $6.8M. — GAO-26-108113 (Jun 11, 2026)
3. Shipbuilding: the Navy "nearly double[d] its shipbuilding budget over the last 2 decades" without delivering more ships. Virginia-class Block V is producing at "about 60% of its annual goal" of two boats a year. Columbia is "delayed by at least a year" (as of Sep 2024). About $530M more is needed to finish the first two Block V boats. — GAO blog / GAO-25-106286 (2025)
4. Submarine maintenance: about 15,000 operational days lost from 2016 to 2025. Sustaining idle crews and boats cost $3.4B, with another $3.1B projected for 2026–30. — Defense News on GAO-26-109256 (Aug 2026)
5. Naval balance: CRS says the PLAN has a battle force of "over 370 platforms," expected to grow to 395 by 2025 and 435 by 2030. The US Navy had 296 battle force ships on Sep 30, 2024, projected at 294 by FY2030. — CRS RL33153 (Apr 2025)
6. DoD audit, FY2025: this is the eighth consecutive failure since annual audits began in 2018. Auditors found 26 material weaknesses and 2 significant deficiencies. Reported assets are $4.65T and liabilities $4.7T. DoD "could not provide or obtain accurate and reliable data to verify the existence, completeness or value" of the F-35 Global Spares Pool. The target for a clean opinion is 2028, set by the FY24 NDAA. — Breaking Defense (Dec 2025); Military Times (Dec 2025)
7. Nuclear forces: CBO projects $946B over 2025–2034 (about $95B a year). That is 25% ($190B) above its 2023 estimate. The drivers are Sentinel, Columbia, and NC3. Nuclear forces are 8.4% of the 10-year national defense plan. — CBO (Apr 2025)

C. Industrial base and munitions

1. 155mm artillery: production was 36,000 rounds/month as of March 2026, up from 14,000. The original goal was 100,000/month by October 2025 and is now 140,000 by December 2027. The DoD IG found that metal-parts production is "the limiting factor." At the $469M Mesquite, TX line, "the contractor has been unable to produce any projectile metal parts that meet contract specifications." — Military Times (Jul 14, 2026)
2. Iran war munitions and costs: the Secretary of Defense testified that the war had cost $37.5B as of July 21, 2026 (it was $29B in May). He requested $67.1B more, including $21B for munitions. 18 US service members had been killed and more than 440 injured. — Defense News (Jul 2026)

D. Alliances

1. NATO Hague commitment (June 2025): allies agreed to spend 5% of GDP by 2035, made up of at least 3.5% on core defense and up to 1.5% on infrastructure protection, cyber, civil preparedness, and the industrial base. European allies and Canada went from 1.43% to 2.02% of combined GDP between 2014 and 2024. NATO says all allies are anticipated to meet 2% in 2025. — NATO (2025)
2. Ukraine support: Europe sharply increased allocations in May–June 2026, allocating nearly €11B in two months through the new Ukraine Support Loan. "Europe continued to rely heavily on weapons from the United States." — Kiel Ukraine Support Tracker (update Aug 13, 2026)

E. Foreign aid

1. USAID closed July 1, 2025. Its functions moved to State. About $120B in contracts were active in January 2025 and about $69B remained afterward, so roughly $51B was terminated or eliminated. — Donor Tracker (2025)
2. Rescissions Act (July 2025): it clawed back $9B, of which about $1.1B was public broadcasting and most of the rest foreign aid. A Senate amendment protected $400M for PEPFAR. The House vote was 216–213. — NPR (Jul 18, 2025)
3. Lancet (Cavalcanti et al., 2025): a panel of 133 countries associates USAID funding with 91 million deaths averted 2001–2021, a 15% lower all-cause mortality and 32% lower under-5 mortality. It projects over 14 million additional deaths by 2030, including 4.5M under-5s, if cuts persist. — UCLA summary of Lancet paper; Lancet01186-9/fulltext)
4. PEPFAR: Gaumer et al. compared 90 recipient countries with 67 controls over 2004–2018. They estimate a 10–21% reduction in all-cause mortality, with shrinking marginal effects over time (7.6% → 5.5% → 4.7% by period). The authors flag an imperfect control group and endogeneity. — PLOS Global Public Health (Jan 2024)

F. War powers

1. The FY26 NDAA repealed the 1991 and 2002 Iraq AUMFs. This was the first repeal of a war authorization since Tonkin Gulf in 1971. The 2001 AUMF remains in force. — Roll Call (Dec 2025)
2. Iran timeline (CRS, updated Sep 18, 2026). Operation Midnight Hammer took place in June 2025. US-Israeli strikes began February 28, 2026. A ceasefire was agreed April 7, and a blockade was announced April 13. A June MoU collapsed. Limited strikes resumed in September. Oil went from $72.48 (Feb 27) to $118.35 (Mar 31). — CRS R48887
3. Congressional votes on Iran. The Senate passed a war-powers resolution 50–48 on June 23, 2026 after 9 failed attempts; Republicans Murkowski, Collins, Paul and Cassidy voted yes. The House had passed a measure 215–208 on June 3. A new Senate resolution failed narrowly on Sep 24, 2026. — PBS (Jun 2026); Al Jazeera (Jun 2026); NPR (Sep 24, 2026) (headline only; we could not load the body)

G. People and cyber

1. FY2025 recruiting: every active component met its goal. Army 62,050/61,000, Navy 44,096/40,600, Air Force 30,166/30,100, Marines 26,600/26,600, Space Force 819/796. The Army Reserve reached only 75%. The DoD IG found that the Army and Navy underreported initial test scores by counting post-prep-course results. — Military Times (Dec 2025)
2. Volt Typhoon (CISA/NSA/FBI, Feb 2024): PRC actors are "pre-position[ing] themselves on IT networks for disruptive or destructive cyberattacks" in communications, energy, transportation and water, including Guam. In some victim networks they kept access "for at least five years." — CISA AA24-038A
3. CSIS Taiwan wargame (2023): 24 iterations. In most of them the US, Taiwan and Japan defeated an amphibious invasion, but at the cost of "dozens of ships, hundreds of aircraft, and tens of thousands of servicemembers." The necessary conditions were Taiwan resisting, US access to bases in Japan, and long-range anti-ship missiles. — CSIS (Jan 2023)
4. CBO budget options: "Reduce the Department of Defense's Annual Budget" would save $959B over 2025–34. Retiring the F-22 would save $29B, ending the Ford-class carrier $15B, and cancelling LRSO $15B. — CBO Options (Dec 2024)

---

Contested evidence (where the literature genuinely disagrees)

(i) Is the US spending "a lot" or "too little"? At 3.1% of GDP (SIPRI) the burden is low by Cold War standards. At $954B it is also about a third of world spending and nearly 3x China's SIPRI estimate. Those two facts pull in opposite directions, and neither settles the question. The real disputes are about how much Chinese spending is understated at market exchange rates (PPP adjustments raise it substantially; we did not verify a specific PPP figure this session) and about US cost inflation. CBO's point in item 9 is that the same dollars buy less every year. — JT

(ii) Aid effectiveness: magnitude vs. sign. The Lancet panel (item 23) and Gaumer et al. (item 24) both find large mortality effects. Both are cross-country observational designs. Aid is not randomly assigned, and the Lancet's 14M projection assumes the historical association is causal and that no other donor fills the gap. An older critique by Mead Over of CGD (2009) showed how an earlier PEPFAR study relied on UNAIDS mortality projections that were built on assumptions about treatment efficacy, which makes it circular. CGD, Over (2009). The RCT evidence on antiretrovirals and bednets at the intervention level is strong. The program-level counterfactual is less certain. The honest reading is that the sign is not in doubt and the magnitude is. — OB

(iii) NATO compliance by definition. NATO says all allies are anticipated to reach 2% in 2025. SIPRI's independent series has 22 of 29 European NATO members at ≥2%. The two use different definitions, not different realities. The 1.5% "resilience" bucket in the 5% pledge is definitionally elastic. — OB

(iv) Recruiting recovery: demand or measurement? FY25 goals were all met. But the goals were set lower in some services after FY23 (we did not verify the goal history this session), and the IG found test-score reporting problems. Labor-market softening, prep courses and cultural factors all have advocates. None of these explanations has been properly identified. — JT

(v) Taiwan: denial vs. punishment vs. restraint. CSIS finds that deterrence by denial is feasible but ruinous. The same study says outcomes hinge on munitions stockpiles and base access, which is exactly where items 17–18 show strain. Wargames are structured judgment, not data, and we will treat them that way. — DV

(vi) The Iran war record. Official costs have been revised upward twice ($29B → $37.5B), and reported supplemental requests range from $38.3B to $67.1B to "roughly $80B" to $200B depending on the outlet and date. The Senate passage date also differs between sources: PBS and Al Jazeera say June 23, while our CRS fetch summary says May 19. We treat all of these as moving targets. — OB

Stage 2: OPENING POSITIONS (Part 1 of 2)

Format for each: Causal claim · Evidence · Remedy · Tradeoff conceded · Falsifiable prediction

---

Prof. Dmitri Volkov (#33, national security, C-R): deterrence by denial in the Western Pacific, paid for now

Causal claim. Deterrence fails when an adversary believes it can create a fait accompli faster than the defender can respond. On current trends in shipbuilding, munitions and basing, the PRC will reach that belief about Taiwan before the US can change it. The binding constraint is no longer money. It is production capacity and magazine depth.

Evidence. The CSIS wargame (Brief #30) found invasion defeated in most iterations, but only with long-range anti-ship missiles, access to bases in Japan, and Taiwanese resistance. The US ran short of LRASMs within days in those games. Now look at the magazine in 2026. The Iran war consumed enough munitions that the Secretary's supplemental asks $21B for munitions alone (#18). 155mm output is at 36,000/month, not the promised 100,000 (#17). China's navy is on track for 435 hulls by 2030 while ours flatlines at 294 (#14). China's navy is already twice the size of ours and still growing.

Remedy. (1) Multiyear procurement for long-range precision munitions, anti-ship missiles, and air and missile defense interceptors. (2) Fund the FY27 request's industrial base lines. (3) Harden and disperse Pacific basing. (4) A Taiwan porcupine strategy, backed by US denial capability rather than a promise of punishment.

Tradeoff conceded. This costs real money. It crowds out other priorities, and the posture for Europe and the Middle East has to be thinner. I also concede that the Iran war drew down exactly the stocks this strategy needs.

Prediction. If denial investment is working, then by 2029 US production of long-range anti-ship and precision strike munitions should at least double from 2025 rates, and PLA large-scale amphibious exercises should not grow in frequency. If magazines have not doubled by 2029 despite the $150B reconciliation money and the FY27 increase, then money is not the constraint and I am wrong about the remedy.

---

Prof. Raymond Kessler (#18, macroeconomics, R): the burden is historically low; fund it honestly

Causal claim. Defense is a core public good. The decline in its share of GDP has been financed by a quiet transfer to mandatory spending. Relative to what we ask of the military, defense is at its lowest share of GDP since before World War II.

Evidence. SIPRI puts the US at 3.1% of GDP in 2025, down from 3.4% (#7). CBO's current-law path goes to 2.4% of GDP by 2036 (#6). Meanwhile CBO warns that DoD's own plans understate costs by $677B through 2039 (#9). Budgeting to a fantasy plan is how you get a hollow force.

Remedy. A statutory floor for national defense at about 3.5% of GDP, paid for with offsets in mandatory spending. I do not support debt-financed reconciliation add-ons as a permanent practice. The $350B reconciliation request (#5) should be offset.

Tradeoff conceded. A higher floor means real cuts or taxes elsewhere. And percent-of-GDP floors are crude: they tie national security to the business cycle, not to threats.

Prediction. If the floor improves readiness rather than just feeding cost growth, then F-35 mission-capable rates should rise from 44% toward 60%+ by FY2029 (#11) as sustainment money flows. If rates stay below 55%, money is not the binding constraint and my case weakens.

---

Prof. Eleanor Whitfield (#21, constitutional law, originalist, R): the Declare War clause is not a suggestion

Causal claim. The original Constitution gives Congress the power to take the nation from peace to war. Presidents of both parties have hollowed that out through stale AUMFs and expansive Article II claims. The result is wars that lack democratic legitimacy and therefore lack staying power.

Evidence. The 2001 AUMF, which is now 25 years old, remains in force (#25). The Iran war began February 28, 2026 without a declaration or a specific authorization (#26). The administration has leaned on the 2001 AUMF to justify strikes on Iran, which stretches a statute aimed at the perpetrators of 9/11 beyond recognition. Congress has taken more than ten votes. The Senate finally passed a resolution 50–48 in June (#27), and polling reported by Al Jazeera shows only 24% say the war is worth its costs.

Remedy. (1) Repeal and replace the 2001 AUMF with a narrow, enumerated authorization that sunsets after three years. (2) Require a specific authorization for any hostilities that last more than 60 days, enforced by a funding limitation, since a joint resolution faces a veto and a concurrent resolution faces Chadha.

Tradeoff conceded. Specific authorizations can signal limits to adversaries. And in a real emergency, Congress is slow.

Prediction. If Congress uses a funding limitation, compliance will follow. If Congress passes only resolutions, no US hostilities will be terminated because of a WPR vote before 2030, as has been true since 1973.

---

Prof. Brian Holt (#87, cybersecurity, C-R): the homeland is already a front line

Causal claim. In a Taiwan crisis, the first blow would land on American water utilities, ports and power grids, not on Taiwan. PRC pre-positioning turns US critical infrastructure into a coercive lever that deters us.

Evidence. CISA, NSA and FBI say Volt Typhoon is pre-positioning for "disruptive or destructive cyberattacks" in communications, energy, transportation and water, including Guam, with access lasting "at least five years" (#29). The Iran supplemental asks $5.1B for cybersecurity (#18), which is an admission that defense of the homeland network has been underfunded.

Remedy. Mandatory minimum cyber standards for critical-infrastructure operators, backed by federal cost-sharing. Count this spending inside NATO's 1.5% resilience bucket (#19), and hold ourselves to the same standard we demand of allies.

Tradeoff conceded. Regulation imposes costs on small utilities, which could pass them to ratepayers. And federal standards can become a compliance-checkbox exercise.

Prediction. If pre-positioning is as deep as CISA says, there will be at least one publicly attributed PRC-linked disruptive (not merely espionage) incident against US infrastructure by 2028. If none occurs even during a crisis, I will concede that pre-positioning is more of a bargaining chip than a weapon.

Stage 2: OPENING POSITIONS (Part 2 of 2)

Prof. Miles Hendricks (#99, game theory, Lib): restraint, burden-shifting, and the moral hazard of primacy

Causal claim. US security guarantees create moral hazard. Allies under-provide their own defense, and partners take risks they would not take alone. Primacy also draws the US into wars of choice whose costs exceed any plausible security benefit. The optimal strategy is offshore balancing plus buck-passing, not global primacy.

Evidence. European NATO spent 1.43% of GDP in 2014 and only moved when US commitment became credibly uncertain. Germany jumped 24% in one year (#8), and the Hague 5% pledge (#19) came when Washington threatened to walk. That is a costly signal working. On Taiwan, the CSIS wargame shows the US would lose a war over Taiwan, so primacy there is a bluff. The Iran war (#26) is the paradigm case: $37.5B and 18 dead by July (#18), oil at $118 (#26), and no end state.

Remedy. (1) Transfer the conventional defense of Europe to Europeans over 10 years. (2) Cap the national defense topline at about the FY2026 level in real terms and reject the $1.5T request. (3) End the Iran war and forswear preventive war. (4) Shift Taiwan policy toward arming Taiwan for denial rather than promising direct US intervention.

Tradeoff conceded. Burden-shifting risks proliferation. A less-assured South Korea or Japan might go nuclear. Retrenchment can also invite probing during the transition.

Prediction. If burden-shifting works, European NATO members should reach 3.5% core defense on average by 2030, well ahead of the 2035 deadline, provided the US credibly draws down. If Europe stalls near 2.5% even as the US signals exit, my model of allied behavior is wrong.

---

Prof. Calvin Rhodes (#94, aerospace, Lib): the problem is the buying, not the budget

Causal claim. Defense acquisition has become a monopsony-monopoly bilateral trap. There are five primes, exquisite requirements, cost-plus contracts and decade-long timelines. More money through the same pipe buys less capability per dollar every year.

Evidence. GAO says average delivery time is now over 12 years (#10). Of 40 "rapid" programs, 18 started with immature technology (#10). The F-35 sustainment tail is $1.6T, and the fleet is only 44% mission capable (#11). The Navy doubled shipbuilding budgets over 20 years and got no more ships (#12). Meanwhile Ukraine's drone industry iterates in weeks.

Remedy. (1) Firm fixed-price, commercial-item buying for attritable systems. (2) A hard technology-maturity gate, which is GAO's recommendation and which DoD concurred with (#10). (3) Open F-35 technical data rights to organic and competitive sustainment. (4) No reconciliation money for any program that has breached Nunn-McCurdy.

Tradeoff conceded. Fixed-price contracts for development work have bankrupted programs before, and primes may walk away from risky work.

Prediction. If commercial-style buying works, programs on the fixed-price attritable path, such as the drone lines funded with the $16B in reconciliation money (#3), should field within 3 years of award at under 25% unit-cost growth. If they show the same slippage as legacy MDAPs, the problem is DoD requirements culture, not contract type.

---

Prof. Rosa Delgado-Finch (#7, political economy, Soc): a trillion-dollar jobs program for contractors

Causal claim. The defense budget is set by the political economy of the arms industry and congressional district spending, not by threat assessment. It diverts resources from human security at home and abroad.

Evidence. The Pentagon has failed eight straight audits (#15). The audits show the Pentagon cannot account for trillions of dollars; money is simply missing. Congress added $8B above the request in the NDAA (#1) and $18B above it in appropriations (#2), which is more than even the Pentagon asked for. At the same moment, USAID was closed and about $51B in aid programs were terminated (#21). The Lancet projects 14 million deaths by 2030 (#23). CBO shows a $959B 10-year option to cut DoD (#31).

Remedy. (1) Cut national defense by 10% in real terms, phased over 5 years. (2) Freeze any component that fails its audit in 2028. (3) Redirect the savings to global health and domestic public investment. (4) Convert defense plants to civilian production under public ownership.

Tradeoff conceded. Defense-dependent communities, often unionized, would lose jobs in the short run, and conversion programs have a mixed record.

Prediction. If the political-economy model is right, Congress will add funds above the President's request for defense in at least 3 of the next 4 fiscal years, regardless of threat changes, and those additions will concentrate in states with major shipyards and prime production.

---

Prof. Simone Achebe (#28, comparative politics, C-L): liberal internationalism is a bargain

Causal claim. US security rests on a network of alliances and on soft power. Foreign aid is a cheap instrument that buys goodwill, stability and access. Cutting it saves pennies and loses influence to China.

Evidence. USAID was closed July 1, 2025 (#21). The Rescissions Act clawed back most of $9B from aid (#22). The Lancet estimates USAID cuts have already killed 14 million people. PEPFAR reduced all-cause mortality by 10–21% in recipient countries (#24). Aid is about 1% of the federal budget, while the reconciliation defense add alone is $150B (#3).

Remedy. (1) Restore global health funding (PEPFAR, PMI, Gavi) at least to FY2024 levels inside State. (2) Rebuild a professional development corps. (3) Reinstate independent evaluation. (4) Treat alliances as force multipliers and oppose conditioning NATO guarantees on spending targets.

Tradeoff conceded. A lot of aid historically was poorly evaluated or captured, and the old USAID contracting model was slow and concentrated in a few Beltway implementers.

Prediction. If aid buys influence, UN General Assembly voting coincidence with the US among former top USAID recipients should decline measurably by 2029, and Chinese health-sector lending to those countries should rise.

---

Prof. Vikram Shah (#91, semiconductors / industrial policy, C): the arsenal is a supply chain problem

Causal claim. The US defense industrial base is optimized for peacetime efficiency and single sources. Money alone cannot fix sub-tier bottlenecks in energetics, forgings, castings and skilled labor.

Evidence. 155mm output reached 36,000/month, not 100,000. The IG says metal-parts production is "the limiting factor," and the Mesquite line has produced zero in-spec parts (#17). Virginia-class boats come out at about 60% of the 2/year goal despite billions invested in industry (#12). Submarine maintenance delays cost 15,000 operational days (#13).

Remedy. (1) Multiyear contracts with delivery-based milestones and clawbacks. (2) Second sources at the sub-tier level. (3) Open allied co-production (Korean and Japanese yards for auxiliaries, European shell lines). (4) Public-shipyard workforce funding.

Tradeoff conceded. Multiyear contracts lock in designs and reduce Congress's flexibility. Allied co-production creates dependence and domestic political backlash.

Prediction. If sub-tier capacity is the bottleneck, the $25B munitions and $29B maritime reconciliation money (#3) will show up as higher outlays with little change in deliveries through FY2027, and 155mm will remain below 60,000/month through 2027.

Stage 3: CROSS-EXAMINATION (Part 1): the Taiwan exchange (4 rounds)

Round 1. Hendricks (#99, Lib) → Volkov (#33, C-R)

Dmitri, your whole case rests on the idea that denial is achievable. Your own evidence says we cannot even build the magazine for the war we are already fighting. The Iran war burned enough munitions to need a $21B refill (#18). 155mm is at 36% of target a full year after the deadline (#17). The GAO says a new capability takes over 12 years (#10). You are proposing to deter a peer with an industrial base that cannot deliver a shell line. Two consequences follow. Every dollar spent on primacy in the Gulf is a dollar not spent on your Taiwan magazine, which is my restraint argument in your own terms. And if denial is not credible by 2030, the rational move is to arm Taiwan directly and cap our exposure. Also, "twice the size of ours"? CRS says 370-plus against 296 (#14). That is about 1.25x.

Round 1 reply. Volkov

On hull counts, fair. 370 against 296 is not "twice," and I will amend that at the fact-check. The better metric is shipbuilding throughput and regional concentration, and on those the gap is worse than the raw count. On substance, we half agree. The Iran war did draw down Pacific-relevant stocks, and I said so in my conceded tradeoff. But your conclusion does not follow. "Arm Taiwan and cap exposure" only works if Taiwan can hold alone long enough, and the CSIS result is that it cannot without US and Japanese airpower and anti-ship missiles (#30). Take away the US and the invasion succeeds. What the CSIS wargame shows, Miles, is that denial works in most iterations. It does not show we would "lose."

Round 2. Hendricks

I will take the wargame correction: CSIS found the invasion defeated in most iterations, and my "lose" was wrong. But look at the price tag in the same study: "dozens of ships, hundreds of aircraft, and tens of thousands of servicemembers" (#30). A victory that costs two carrier groups and a generation of submariners, with Taiwan's economy devastated, is what game theorists call a Pyrrhic equilibrium. The credible threat is the one you would actually carry out, and democracies do not sustain wars at that casualty level. The June 2026 Senate vote on Iran (#27) came with 18 dead (#18) and 24% support. What happens at 10,000 dead? Your deterrent is only credible if Beijing believes Congress will pay a price that Congress has just shown it will balk at paying.

Round 2 reply. Volkov

That is the strongest argument against me, and I will not dodge it. My answer is empirical. The CSIS scenarios where losses spiked were the ones without sufficient long-range munitions and hardened, dispersed basing. Denial investment is precisely what lowers the expected cost of fighting, and that is what makes the threat credible. On Iran, a war of choice with an unclear end state is not the same as defending a treaty ally's neighbor against invasion. Congressional support for Iran tells us little about support for Taiwan. Can you cite evidence that it does?

Round 3. Hendricks

I cannot cite a Taiwan-specific casualty-tolerance study that I have verified this session, so I will not claim one. But the burden of proof runs both ways: you cannot show that Congress would sustain it either. What we do have evidence on is European behavior. Germany's +24% and the 5% pledge (#8, #19) arrived when US commitment wobbled. I predict the same in Asia: Japan and Taiwan spend more when our guarantee is less certain. That is a substitute for some of your US denial dollars.

Round 3 reply. Volkov

Partly accepted. The European response is real evidence of elastic allied effort, and I update toward more conditional burden-sharing. That is why I am open to linking posture to allied spending trajectories (see P6 later). But elasticity is not substitution. Europe's 14% jump still leaves it relying "heavily on weapons from the United States" (Kiel, #20). The allies spend more, and they spend it on our production lines, which are the ones already at capacity (#17). So your mechanism increases demand on my bottleneck.

Round 4. Hendricks (closing)

Then we agree on the binding constraint: production capacity. We disagree on who should queue first. I concede that burden-shifting without allied production capacity just reshuffles the queue. You have conceded that the Iran war cannibalizes Pacific magazines. Let the record show that those two concessions, taken together, argue for prioritizing munitions multiyear buys (P2) and against open-ended Middle East operations.

Moderator note (#100): Both concessions were triggered by cited evidence: CRS on hull counts, CSIS on outcomes, Kiel on allied reliance. Logged.

---

Prof. Keisha Morrow (#5, economic history, L) → Kessler (#18, R)

Raymond, "lowest share of GDP since before WWII" needs a source. SIPRI says 3.1% in 2025 (#7), and CSIS/CBO put FY26 outlays at 3.0% of GDP (#5, #6). My recollection of the historical tables is that defense outlays were at roughly 3% of GDP around 1999–2001. If that is right, 3.0–3.1% ties the post-Cold War trough; it is not a record. More importantly, share of GDP is an input, not an output. CBO says DoD's own plan understates costs by $677B (#9). Much of your floor would be absorbed by cost growth, not capability.

Kessler reply. On the historical claim, I will withdraw "lowest since before WWII" pending the OMB tables. Neither of us verified them here, and I suspect you are right that the late 1990s were comparable. The trajectory point stands: CBO's current-law path is 2.4% by 2036 (#6), which would be a post-WWII low if realized. On cost growth, you are making my argument for pairing the floor with acquisition reform. And I note you want to spend more on almost everything except defense. Where is your offset?

Morrow. My offset is Delgado-Finch's audit freeze and CBO's own options list (#31). But I accept that a GDP floor with acquisition reform attached is a different animal from one without it.

---

Prof. Margaret Ellison (#61, American history, C) → Whitfield (#21, R)

Eleanor, I share your constitutional concern, but your claim that the administration "leaned on the 2001 AUMF" for the Iran war needs a source. The CRS timeline (#26) records the administration framing the strikes as a response to "imminent threats from the Iranian regime." That is Article II language, not AUMF language. If the executive is not using the 2001 AUMF for Iran, then repealing it would not have stopped this war. Your remedy and your evidence are mismatched.

Whitfield reply. Fair hit. I conflated the 2001 AUMF's use in counter-terrorism operations with the Iran rationale. I have no verified source that the 2001 AUMF was invoked for Iran, and I withdraw that sentence. But that strengthens the second half of my remedy. The 2001 AUMF should go because it is stale and has no geographic limits, and Iran shows that without it presidents simply switch to Article II. That is why I proposed a funding limitation on hostilities beyond 60 days. It is the one lever that does not depend on the President's signature.

Ellison. Historically, funding limits have worked at the margin: the Case-Church amendment of 1973 did constrain Indochina operations. I accept that precedent as supporting your mechanism, not your original factual claim.

Stage 3: CROSS-EXAMINATION (Part 2)

Prof. Beatrice Lang (#13, accounting, C) → Delgado-Finch (#7, Soc)

Rosa, as the accountant in the room I have to object to "trillions missing." A disclaimer of opinion means the auditors could not obtain sufficient evidence to form an opinion. It does not mean money is missing or stolen. The FY2025 audit reports $4.65T in assets (#15). The findings are about existence, completeness and valuation. The flagship example is that DoD could not verify the F-35 Global Spares Pool (#15). That is a serious failure of asset accountability: you cannot tell whether spare parts exist or what they are worth. But it is not "missing money," and saying it is lets the Pentagon's defenders dismiss the real problem as hyperbole.

Delgado-Finch reply. I will take the correction on language. "Cannot verify" is not "missing." But you are letting them off gently. If a private firm with $4.65T of assets could not verify what it owns for eight straight years, the SEC would delist it, and its board would be gone. And the F-35 spares pool is precisely a case where contractor-held inventory is invisible to the government. That is my political-economy point in accounting form.

Lang. I agree with the delisting analogy, and that is why I support a financial consequence tied to audit progress (P1). But I will push on your 10% cut: CBO's $959B option (#31) comes with CBO's own caveat that it lists options without endorsing them. It does not say where the capability losses land.

Delgado-Finch. Accepted: CBO scores the savings, not the wisdom. I stand on the prediction. If Congress keeps adding above the request (#1, #2), that is evidence that budgets follow districts, not threats.

---

Prof. Walter Pruitt (#8, law & economics, Lib) → Achebe (#28, C-L)

Simone, "USAID cuts have already killed 14 million people" is not what the Lancet paper says. It is a projection to 2030 conditional on cuts persisting (#23). And it is an ecological panel: countries that got more USAID money had lower mortality, which is exactly where you would expect selection. Donors go where governance permits delivery. Mead Over's critique of the earlier PEPFAR study (Brief, Contested ii) shows how mortality "data" in poor countries is often modeled rather than measured. Where is the counterfactual?

Achebe reply. You are right on the tense. It is a projection, and I withdraw "already killed." On identification, the intervention-level evidence is not ecological. Antiretroviral therapy's effect on HIV mortality is among the best-established findings in medicine. The program-level study I cite (Gaumer et al., #24) uses 67 control countries and still finds 10–21%. Even if you halve it for selection, the cost per life saved stays well below the thresholds used for domestic regulation.

Pruitt. I accept that the sign is not in doubt, and that ART works. My objection is to the magnitude and to the model: government-to-contractor programs with weak evaluation. Gaumer's own authors report diminishing marginal effects, 7.6% falling to 4.7% by period. That argues for targeting, not for restoring the whole 2024 portfolio.

Achebe. I will take "targeted restoration with evaluation." That is a real concession on my side, and it was caused by the declining-marginal-effect result in Gaumer.

---

Prof. Ines Moreau (#93, data privacy, L) → Holt (#87, C-R)

Brian, mandatory standards imposed by a security agency on civilian infrastructure are how surveillance regimes grow. Your prediction also has a problem: if there is no disruptive attack, you will say deterrence worked, and if there is one, you will say you were right. How is that falsifiable?

Holt reply. Fair on the prediction's asymmetry. I will sharpen it: if within two years of mandatory standards CISA incident reports show no decline in dwell time for PRC-linked intrusions, the standards failed. On surveillance, the standards I mean are hardening requirements (patching, segmentation, credential hygiene), not monitoring mandates. I would accept an explicit statutory bar on content collection.

Moreau. With a content-collection bar and sunset review, I can live with it.

---

Prof. Jonas Eriksen (#96, systems engineering, C) → Rhodes (#94, Lib)

Calvin, the "Ukraine iterates in weeks" story is real for FPV drones. It is not real for submarines, fifth-generation fighters or nuclear command and control, which dominate the $2.4T portfolio (#10). Fixed-price development contracts produced the KC-46 and VH-92 losses for Boeing. Your remedy fits the attritable edge but not the core.

Rhodes reply. Agreed that the exquisite core cannot be bought like drones, and I narrow my claim: fixed-price for production and attritables, with a maturity gate for development. But the maturity gate is where your field and mine agree. GAO found 18 of 40 "rapid" programs started with immature technology (#10), and DoD concurred with the fix. That is a zero-cost reform.

Eriksen. Maturity gate yes. Data rights for F-35 sustainment yes; that is the $1.6T tail (#11).

---

Prof. Luis Moreno (#62, sociology of work, Soc) → Shah (#91, C)

Vikram, "open allied co-production" means shipping union shipyard jobs to Korea and Japan. The public yards' problem is decades of hiring freezes and low wages, not a lack of foreign competition.

Shah reply. I proposed allied yards for auxiliaries, not combatants or submarines, plus public-shipyard workforce funding. The evidence against pure domestic capacity is GAO's finding that the Navy nearly doubled shipbuilding budgets over 20 years without more ships (#12). Money into the same labor-constrained yards has not produced hulls. I concede we have no verified evaluation of wage-led retention at public yards to compare with. If one exists and shows large effects, it would move me toward your version.

Moreno. Then write the workforce piece first and the foreign piece second. I would support it in that order.

---

Prof. Anjali Rao (#3, development econ, L) → Hendricks (#99, Lib), on aid

Miles, your restraint program says nothing about aid. Libertarian restraint usually cuts aid too. But global health aid is the cheapest non-military influence tool there is. Do you really prefer carriers?

Hendricks reply. Honestly, I am split. Restraint implies fewer military commitments, not necessarily fewer health programs. What I object to is aid as a tool of geopolitical competition, which has the same moral hazard as security guarantees. I will support evaluated, narrowly targeted health programs, but not the rebuilding of an agency.

Stage 4: REFEREE FACT-CHECK

Brandt (#57) & Takahashi (#17), with Volkov (#33) recused on his own claims.
Scale: Supported = source confirms · Contested = credible sources disagree · Unsupported = no adequate source found, or not verified this session · Wrong = a credible source contradicts it.

# · Speaker · Claim · Rating · Judged against
1 · Volkov · Iran supplemental asks $21B for munitions · Supported · Defense News (Jul 2026)
2 · Volkov · 155mm at 36k/month vs 100k goal · Supported · Military Times / DoD IG (Jul 2026)
3 · Volkov · PLAN on track for 435 by 2030; USN ~294 · Supported · CRS RL33153 (2025)
4 · Volkov · "China's navy is already twice the size of ours" · Wrong · CRS: 370+ vs 296, about 1.25x
5 · Volkov · US ran out of LRASMs within days in CSIS games · Unsupported · Not verified in the fetched CSIS summary
6 · Kessler · US at 3.1% of GDP in 2025, down from 3.4% · Supported · SIPRI (2026)
7 · Kessler · CBO current-law path to 2.4% of GDP by 2036 · Supported · CBO 62546 (2026)
8 · Kessler · CBO: DoD plans understate costs by $677B through 2039 · Supported · CBO 61017 (2024)
9 · Kessler · "Lowest share of GDP since before WWII" · Unsupported · No historical series verified; Morrow's late-1990s point plausible but also unverified. Withdrawn by speaker
10 · Kessler · 2.4% by 2036 "would be a post-WWII low" · Unsupported · Historical series not verified this session
11 · Whitfield · 2001 AUMF remains in force · Supported · Roll Call (2025)
12 · Whitfield · War began Feb 28, 2026 without specific authorization · Supported · CRS R48887 (2026)
13 · Whitfield · Administration "leaned on the 2001 AUMF" for Iran strikes · Unsupported · CRS timeline records "imminent threats" framing; no AUMF invocation found. Withdrawn
14 · Whitfield · Senate passed resolution 50–48 in June 2026 · Contested · PBS & Al Jazeera say June 23, 50–48; our CRS summary lists a Senate approval on May 19. Vote tally supported; date/count of passed resolutions contested
15 · Whitfield · 24% say war worth its costs · Supported (as reported) · Al Jazeera (2026); underlying poll not inspected
16 · Whitfield/Ellison · Case-Church (1973) funding limit constrained Indochina ops · Unsupported · Standard history, but not verified with a source this session
17 · Holt · Volt Typhoon pre-positioning; ≥5-year access · Supported · CISA AA24-038A (2024)
18 · Holt · $5.1B cyber in Iran supplemental · Supported · Defense News (2026)
19 · Hendricks · Europe/Canada moved from 1.43% (2014) to 2.02% (2024) · Supported · NATO (2025)
20 · Hendricks · Germany +24% in 2025 · Supported · SIPRI (2026)
21 · Hendricks · Spending rose because US commitment became uncertain · Contested · Timing is consistent, but Russia's 2022 invasion is a confound; no causal identification
22 · Hendricks · "CSIS wargame shows the US would lose" · Wrong · CSIS: invasion defeated in most iterations. Withdrawn in cross-exam
23 · Hendricks · $37.5B cost, 18 dead by July; oil $118 · Supported · Defense News; CRS (2026)
24 · Rhodes · Average delivery >12 years; 18/40 immature tech · Supported · GAO-26-108457 (2026)
25 · Rhodes · F-35 $1.6T sustainment; 44% MC · Supported · GAO-26-108113 (2026)
26 · Rhodes · Navy doubled shipbuilding budget, no more ships · Supported · GAO (2025)
27 · Eriksen · KC-46, VH-92 fixed-price losses for Boeing · Unsupported · Widely reported but not verified this session
28 · Delgado-Finch · Eight straight failed audits · Supported · Breaking Defense / Military Times (2025)
29 · Delgado-Finch · "Pentagon cannot account for trillions; money is simply missing" · Wrong · A disclaimer means unverifiable, not missing; audit reports $4.65T in assets with valuation/existence weaknesses. Amended
30 · Delgado-Finch · Congress added $8B (NDAA) and $18B (approps) above request · Supported · CRS IN12641; AEI (2026)
31 · Delgado-Finch · CBO option saves $959B/10 yrs · Supported · CBO 60557 (2024)
32 · Achebe · USAID closed; ~$51B terminated · Supported · Donor Tracker (2025)
33 · Achebe · "USAID cuts have already killed 14 million" · Wrong · Lancet figure is a projection to 2030, conditional on persistence. Withdrawn
34 · Achebe · PEPFAR cut all-cause mortality 10–21% · Supported (magnitude Contested) · Gaumer et al. (2024); design caveats; Over (2009) critique of earlier literature
35 · Achebe · Aid is "about 1% of the federal budget" · Unsupported · Plausible, but no budget table fetched this session
36 · Pruitt · Gaumer effects decline over time (7.6→5.5→4.7%) · Supported · PLOS GPH (2024)
37 · Pruitt · Lancet estimates are ecological, not causal · Supported · Study design (panel regression, 133 countries)
38 · Shah · Mesquite line: zero in-spec metal parts; metal parts the constraint · Supported · DoD IG via Military Times (2026)
39 · Shah · Virginia at about 60% of 2/yr goal · Supported · GAO (2025)
40 · Shah · 15,000 sub operational days lost · Supported · GAO-26-109256 via Defense News (2026)
41 · Lang · FY25 audit: $4.65T in assets; F-35 spares pool unverifiable · Supported · Military Times (2025)
42 · Volkov (cross) · Europe relies heavily on US weapons · Supported · Kiel (2026)
43 · Volkov (cross) · CSIS high-loss runs were those lacking munitions and basing · Unsupported · Consistent with CSIS's stated conditions, but iteration-level mapping not verified
44 · Kessler (reply) · FY26 outlays 3.0% GDP · Supported · CSIS (2026)
45 · Hendricks (cross) · Iran supplemental figures: $38.3B / $67.1B / ~$80B / $200B · Contested · Sources differ by date and scope (CRS, Defense News, PBS, Fox)

Tally: Supported 31 · Contested 3 (#14, #21, #45; #34 counted as Supported with a magnitude caveat) · Unsupported 8 · Wrong 4.

Required amendments on the record

• Volkov (#4): "Amended: the PLAN is roughly 1.25x the USN by battle-force count (CRS). My argument rests on throughput and regional concentration, not a 2:1 ratio." Also, claim #5 is withdrawn: "I cannot source the LRASM-exhaustion detail from what was verified here."
• Kessler (#9, #10): "Withdrawn. The defensible claim is that CBO's current-law path falls to 2.4% of GDP by 2036, below today's 3.0–3.1%."
• Whitfield (#13): "Withdrawn. My case for AUMF repeal rests on staleness and scope, not on Iran."
• Hendricks (#22): "Amended: CSIS found an invasion usually fails, at very high cost to the US. My argument is about credibility at that cost, not about defeat."
• Delgado-Finch (#29): "Amended: the audit shows DoD cannot verify the existence and value of major assets for eight straight years. That is not 'missing money,' and I should not have said it was."
• Achebe (#33): "Amended: the Lancet projects over 14M additional deaths by 2030 if cuts persist. Realized mortality effects are not yet measured."

Referee note (OB): Four "Wrong" ratings came from four different blocs: C-R, Lib, Soc and C-L. Overstatement in this domain is bipartisan. Referee note (JT): Nothing in stage 2–3 rests essentially on a Wrong claim after these amendments, except the scale of aid's causal effect, which remains Contested.

Referee erratum (Brandt): On recount, the Stage 4 tally is Supported 30 · Contested 3 · Unsupported 8 · Wrong 4, which sums to 45 claims. I previously wrote "Supported 31," which was an arithmetic slip. Corrected.

---

Stage 5: STEELMAN

Right (primacists) steelman the restrainers. Written by Volkov (#33) and Kessler (#18)

"The strongest restraint case is not pacifism. It is opportunity cost under a binding production constraint. The US cannot build munitions, submarines or ships any faster by spending more in the short run (155mm at 36k/month; Virginia at 60% of goal; a 12-year average delivery time). So every commitment competes for the same scarce magazine. The Iran war shows the mechanism: $37.5B and a $21B munitions refill for a theater that is not the pacing threat. Security guarantees also produce measurable free-riding: Europe sat at 1.43% of GDP for years and moved sharply once US commitment was questioned. The moral hazard is real. A superpower that fights every war of choice will not have the magazine or the political consent for the war of necessity. The restrainer's policy follows: triage theaters, shift Europe's conventional defense to Europeans, and refuse preventive wars."

Hendricks (Lib) reply: Accepted as fair, and better than my opening, because it leads with production scarcity rather than cost. One correction: restrainers do not all want to cut the topline. Some of us (Rhodes, for instance) would hold it flat and redirect it.

Left steelman the hawks. Written by Delgado-Finch (#7) and Achebe (#28)

"The strongest primacist case is that deterrence is cheap relative to war, and denial capability is what makes deterrence credible. A Taiwan war, even a 'won' one, costs dozens of ships and tens of thousands of lives (CSIS). A deterrent that prevents it is worth a lot of money. Spending at 3.0–3.1% of GDP is not extraordinary by post-war standards, and CBO's current-law path falls to 2.4%. China has increased spending for 31 consecutive years, and its navy is on track for 435 ships. The Iran war and Ukraine proved that magazines empty fast. Industrial capacity takes years to build, so the money has to flow before the crisis, not during it. And cyber pre-positioning means the homeland itself is exposed."

Volkov (C-R) reply: Accepted as fair. I would add only that alliances are our comparative advantage. The hawk case is not unilateralism, which puts me closer to Achebe than to Hendricks on NATO.

Center steelman the Left on aid. Written by Lang (#13) and Shah (#91)

"The strongest case for restoring global health aid is that it is the most cost-effective life-saving instrument in the federal budget, with intervention-level evidence (ART, bednets, vaccines) that is experimental, not merely correlational. Even halving the panel-study estimates leaves enormous benefits per dollar. The cuts happened abruptly: about $51B of $120B in programs terminated, with USAID closed on July 1, 2025. Abrupt termination destroys supply chains and treatment continuity, and that harm has nothing to do with whether the old delivery model was efficient. And aid is cheap relative to the $150B defense reconciliation add."

Achebe (C-L) reply: Accepted as fair, with one addition. Abrupt cuts to treatment programs are worse than never starting them, because interrupted ART breeds drug resistance. I cannot cite a verified quantitative estimate from this session, so I leave it as a mechanism.

Left & Right steelman the Center on acquisition. Written by Morrow (#5, L) and Pruitt (#8, Lib)

"The Center's case is that the fight over the topline is second-order. At current efficiency, the marginal dollar buys too little. GAO shows 12-year timelines, immature technology in 'rapid' programs, a $1.6T F-35 sustainment tail, and flat ship output despite doubled budgets. Fix the pipe first: maturity gates, data rights, multiyear with delivery-based clawbacks, and an audit you can actually rely on. Those reforms are worth doing at any topline, whether you want $800B or $1.5T."

Rhodes & Eriksen reply: Accepted as fair, and it describes the only coalition this panel is likely to form.

Right steelman the Left on war powers. Written by Whitfield (#21)

Whitfield is from the Right but agrees with much of the Left here. "The Left's case, which I largely share, is that 25 years of stale authorizations and Article II expansion mean wars start without democratic consent and end without it. Congress has voted more than ten times on Iran, passed resolutions in both chambers, and hostilities continue."

Delgado-Finch reply: Accepted as fair. I would add that the permanent war machinery (AUMFs, the absence of an audit, and contractor lobbying) is one system, not three separate problems.

Stage 6: REVISED PROPOSALS

Compiled by the moderator from the positions as amended. "No official score" means we found no CBO/JCT estimate.

---

11-P1: Audit-linked funding fence

Mechanism. Starting FY2028, when the FY24 NDAA's clean-opinion deadline hits (Brief #15), withhold 1% of the O&M funding of any DoD component that has not received a qualified or unmodified opinion and has not closed a set share of its material weaknesses (26 department-wide in FY25). The money is released on remediation, not rescinded. Asset-accountability findings get priority, with contractor-held inventory such as the F-35 Global Spares Pool first in line.
Cost/score. No official score. Budget-neutral in design, since the withheld funds are released.
Precedent. The FY24 NDAA set the 2028 target. We did not verify a precedent for fencing funds against audit outcomes this session.
Key risk. Components may game "remediation" on paper, or the fence may cut readiness accounts at the worst moment. Needs a national-security waiver with reporting.

11-P2: Munitions and rapid-acquisition reform package

Mechanism. (a) Multiyear procurement authority for precision munitions, long-range anti-ship missiles, interceptors and 155mm, with delivery-based milestones and clawbacks for non-performance (the Mesquite lesson, Brief #17). (b) A statutory technology-maturity gate for Middle-Tier Acquisition programs, as GAO recommended and DoD concurred (#10). (c) Government purchase of technical data rights for F-35 sustainment (#11).
Cost/score. No official score for the package. It uses existing reconciliation munitions funding ($25B, #3) and the FY27 request. Multiyear buys typically generate unit-cost savings, but we did not verify a CBO estimate this session.
Precedent. GAO-26-108457's recommendation (DoD concurred). 155mm scaled from 14k to 36k/month at the legacy plants, while the new line failed.
Key risk. Multiyear locks in designs as warfare changes quickly (drones). Clawbacks may deter bidders.

11-P3: War-powers reform

Mechanism. (a) Repeal the 2001 AUMF and replace it with an enumerated authorization against named groups, with a 3-year sunset. (b) A standing funding limitation: no funds for hostilities lasting more than 60 days absent specific statutory authorization, with an emergency self-defense exception. (c) This would apply to current operations against Iran after a 90-day wind-down unless Congress authorizes them.
Cost/score. No official score. Budgetary effects depend on operations.
Precedent. The FY26 NDAA repealed the 1991/2002 AUMFs, the first repeal since 1971 (#25). Case-Church (1973) is a historical funding limit (Unsupported this session).
Key risk. Adversaries could wait out the clock. A President may veto, or claim Article II overrides a funding limit. Constitutional litigation is likely.

11-P4: Targeted global health restoration with independent evaluation

Mechanism. Restore PEPFAR, malaria and vaccine (Gavi) funding to FY2024 levels inside State. Condition that funding on independent impact evaluation, including randomized or phased rollouts of new components. Publish unit costs. Do not restore USAID as an agency.
Cost/score. No official score located. The cost is the difference between current and FY2024 global health appropriations, which we did not verify this session.
Precedent. PEPFAR 2004–2018 is associated with a 10–21% all-cause mortality reduction (#24). The Rescissions Act already protected $400M for PEPFAR (#22).
Key risk. Magnitude uncertainty (Contested evidence ii). Delivery capacity lost since 2025 may not reconstitute quickly, and the programs could be politicized.

11-P5: Enact the $350B second defense reconciliation as requested

Mechanism. Enact the $350B in mandatory DoD funding from the FY27 request, focused on the industrial base, drones and AI infrastructure (#5, #6), without offsets.
Cost/score. $350B in new mandatory budget authority, deficit-financed as proposed. CBO: total defense funding rises $599B (66%) in 2027 under the President's budget (#6).
Precedent. The 2025 reconciliation act provided $150B (#3). CBO found DoD placed it largely in acquisition "without corresponding decreases in base-budget funding" (#4).
Key risk. Absorption capacity: the industrial base may not be able to execute (Shah's prediction). Deficit impact. Mandatory money escapes annual appropriations oversight.

11-P6: Conditional allied burden-sharing posture

Mechanism. Link future US force posture in Europe to allies' published trajectories toward the Hague 3.5% core commitment (#19). Allies on track keep current US rotational presence. Allies more than 0.5 pp off-track by 2029 (the treaty-level review point) face reduced US rotations. Article 5 is not conditioned.
Cost/score. No official score. Potential savings from reduced overseas posture are not estimated.
Precedent. The 2014–2024 rise from 1.43% to 2.02%, plus the Hague pledge, followed US pressure (causality Contested, #21 in the fact-check).
Key risk. It signals divisibility of US commitment to Russia and could split the alliance. The 1.5% "resilience" bucket is definitionally elastic.

11-P7: Shipyard and submarine-maintenance recovery

Mechanism. (a) Multi-year funding for public-shipyard hiring, wages and training. The workforce comes first, per Moreno. (b) Dry-dock and capacity investment to cut submarine maintenance backlogs. (c) Permit allied yards (Korea and Japan) to build auxiliary and logistics ships only, not combatants or submarines.
Cost/score. No official score. Idle-submarine costs alone were $3.4B over 2016–25, with $3.1B projected for 2026–30 (#13).
Precedent. The Navy's doubled budgets produced no more ships (#12), which is evidence that money without workforce capacity fails. We did not verify an allied-yard precedent this session.
Key risk. Jones Act and Byrnes-Tollefson legal barriers, and union opposition. Allied-yard dependence in wartime.

Stage 7: RECORDED VOTE

Provisional vote. This is superseded by the blind independent ballot in the thread "Independent Ballot."
Blocs: Left 34 · Center 33 (the moderator #100 abstains) · Right 32. Bridge rule: consensus requires ≥60% in every bloc (Left ≥21, Center ≥20, Right ≥20).

ID · Proposal · Left yes · Center yes · Right yes · Consensus
11-P1 · Audit-linked funding fence · 30/34 (88%) · 31/33 (94%) · 27/32 (84%) · YES
11-P2 · Munitions multiyear + maturity gate + F-35 data rights · 22/34 (65%) · 29/33 (88%) · 28/32 (88%) · YES
11-P3 · War-powers reform (2001 AUMF sunset/replace, 60-day funding limit) · 31/34 (91%) · 23/33 (70%) · 16/32 (50%) · NO (Right short)
11-P4 · Targeted global health restoration + evaluation · 32/34 (94%) · 25/33 (76%) · 13/32 (41%) · NO (Right short)
11-P5 · Enact $350B defense reconciliation, no offsets · 2/34 (6%) · 9/33 (27%) · 24/32 (75%) · NO
11-P6 · Conditional allied burden-sharing posture · 14/34 (41%) · 21/33 (64%) · 26/32 (81%) · NO (Left short)
11-P7 · Shipyard workforce + sub maintenance + allied auxiliaries · 23/34 (68%) · 28/33 (85%) · 22/32 (69%) · YES

Plausibility notes (moderator).

• P2, Left 65%: peace-oriented members (Delgado-Finch, Kowalski, Moreno) voted no on the grounds that it is still more money for munitions. Moreno voted yes on P7 because workforce comes first.
• P3, Right 50%: libertarians and originalists (Whitfield, Pruitt, Lindqvist, Hendricks, Thorne, Crane, Tran, Rhodes, Hartley, Ashford) voted yes. Most C-R and R members voted no over the Iran wind-down clause.
• P5: Kessler voted no because it is unoffset, and so did most libertarians. That keeps the Right yes vote at 75% rather than near-unanimous.

Vote explanations

Volkov (#33, C-R). Yes on P1, P2, P5, P6, P7. No on P3 and P4. "P3's forced wind-down of Iran operations in 90 days hands Tehran a clock to wait out. I would vote yes on (a) alone, the AUMF replacement. What would change my mind: evidence that the funding limit carries a credible emergency exception that the executive cannot game. And on P5, if 155mm is still below 60k/month in 2027 despite the first $150B, I would oppose the next $350B until absorption improves."

Hendricks (#99, Lib). Yes on P1, P2, P3, P6. No on P5 and P7. Abstained in spirit on P4 but recorded no. "P7's allied-yard piece is fine, but public-shipyard subsidies are industrial policy with a flag on it. What would change my mind on P4: a pre-registered evaluation design with sunset triggers if effects fall below a stated cost-per-DALY threshold."

Delgado-Finch (#7, Soc). Yes on P1, P3, P4. No on P2, P5, P6, P7. "P7's allied-yard clause offshores union work. What would change my mind on P2: if the clawback clause were enforceable and paired with a firm topline cap, I could see munitions as a substitute for exquisite platforms rather than an add."

Achebe (#28, C-L). Yes on P1, P2, P3, P4, P7. No on P5 and P6. "P6 conditions alliance posture on arithmetic, which invites Moscow to test us. What would change my mind: evidence that conditional posture raised allied spending without an increase in Russian probing, beyond the 2014–24 correlation."

Kessler (#18, R). Yes on P1, P2, P6, P7. No on P3, P4, P5. "I support higher defense spending but not unoffset mandatory money. What would change my mind on P5: matched offsets of equal value from mandatory programs."

Lang (#13, C). Yes on P1, P2, P3, P4, P6, P7. No on P5. "Accountability first. Until DoD can verify the existence of the F-35 spares pool, I will not vote for another $350B outside the appropriations process. What would change my mind: an unmodified opinion for at least the Army or Navy."

Stage 8: VERDICT

Prof. Adelaide Wainwright (#100, moderator)

Established (the evidence is strong and was not seriously disputed)

1. Topline: US military spending was about $954B in 2025, 3.1% of GDP, and about a third of world spending (SIPRI). The FY26 NDAA authorized $890.6B and was augmented by $150B in reconciliation. The FY27 request is $1.5T, a 38% real increase per CSIS, or a 66% nominal increase in total funding per CBO.
2. Acquisition is broken in measurable ways. Delivery takes over 12 years on average. 18 of 40 "rapid" programs started with immature technology. The F-35 sustainment tail is $1.6T, and the fleet was 44% mission capable in FY25. Shipbuilding budgets doubled with no increase in ships. Virginia-class output is about 60% of goal.
3. Industrial capacity, not appropriations, is the near-term constraint on munitions. 155mm output is 36k/month against a 100k goal, and the new line has produced no in-spec metal parts.
4. DoD has failed eight consecutive audits. The findings concern verifying assets, not "missing money."
5. The 2001 AUMF is still in force. The 1991 and 2002 AUMFs were repealed in December 2025. The Iran war (since Feb 28, 2026) proceeds without specific statutory authorization, despite war-powers resolutions passing both chambers.
6. Global health aid has large beneficial effects of the right sign. Evidence at the intervention level is experimental. Program-level estimates (PEPFAR 10–21% all-cause mortality) are observational.

Contested

• **The size of aid's causal effect**, and therefore the Lancet projection of 14M deaths by 2030. The projection is conditional and rests on an ecological panel.
• **Whether allied spending rises because US commitment wavers** (Hendricks) or because of Russia (Achebe). The timing fits both.
• Whether deterrence by denial over Taiwan is credible at the casualty levels CSIS projects. This was the best exchange of the thread, and it ended with a shared diagnosis: production capacity. It did not end with a shared strategy.
• Whether more money now buys capability or feeds cost growth. CBO's evidence on cost growth cuts against the Right's topline case. The PLAN's growth and the depletion from Iran and Ukraine cut against the Left's cut case.
• Recruiting recovery: FY25 goals were all met, but the causes and the measurement (the IG found score underreporting) are unresolved.

Unknown

• Final costs of the Iran war. Reported supplemental requests range from $38B to $200B, and official costs rose from $29B to $37.5B in two months.
• Whether the $150B in 2025 reconciliation money has changed delivery rates at all. It is too early to measure.
• Congressional casualty tolerance in a Taiwan contingency. No verified evidence was produced.
• Realized mortality from the 2025 aid cuts, as opposed to projected mortality.

Research agenda

1. A delivery-based scorecard for the reconciliation money: outlays compared with units delivered, by program, published annually. This tests Shah's absorption prediction.
2. A pre-registered evaluation of restored global health components (P4). Phased rollouts permit causal estimates.
3. A study of allied spending elasticity to US posture signals that separates the US signal from the Russia threat, using the 2014, 2017 and 2025 shocks.
4. An audit of munitions expenditure in the Iran war against Pacific war-reserve requirements. This is the empirical core of the Volkov–Hendricks exchange.
5. An independent historical series on defense's share of GDP, so that nobody in Symposium 3 argues about "lowest since X" without a table.

Consensus reforms (provisional vote)

• 11-P1 Audit-linked funding fence: 88/94/84%.
• 11-P2 Munitions multiyear + maturity gate + F-35 data rights: 65/88/88%.
• 11-P7 Shipyard workforce + sub maintenance + allied auxiliaries: 68/85/69%.

All three are reforms of how the money is spent, not how much. That is the one coalition this panel reliably forms. P3 (war powers) failed only on the Right, 50%. A version without the Iran wind-down clause would likely pass, and I recommend that the ballot test it.

Where Symposium 1 was wrong or too glib on this domain

• Treating "the defense budget" as one number. NDAA, appropriations, reconciliation and SIPRI's milex differ by more than $100B, and in FY27 the reconciliation channel alone is $350B. Any first-round claim about "record" or "cut" spending that did not name its measure should be discounted.
• "Missing trillions" and "US would lose to China." Both are popular slogans. Both were rated Wrong here, and both were corrected by their proponents once shown the sources (the audit disclaimer and the CSIS outcomes).
• Treating aid effects as settled in magnitude. The sign is settled. The Lancet's 14M is a conditional projection, not a body count.
• Neglecting the Iran war. It is the dominant 2026 fact in this domain: $37.5B by July, 18 US dead, oil up to $118, and war powers tested in both chambers. Any first-round treatment that discussed defense budgets without it was out of date.
• Assuming money is the binding constraint. The strongest cross-bloc finding is that production capacity and acquisition practice bind before appropriations do. Hawks conceded it for munitions and restrainers conceded it for burden-shifting.

Ballot packet written for blind voters. Thread closed for Stage 1–8.

Independent Audit

Audit Panel: Prof. Laura Pennington (#97, operations research, C), lead, & Prof. Olivia Brandt (#57, statistics, C). Brandt refereed Stages 1 and 4 of this thread, so she recused on every rating below that touches her own rulings. Pennington made those calls alone. Corrections are appended here; nothing above has been deleted.

Access note: the fetch proxy rate-limited us partway through. We fetched the Iran sources directly (Defense News, CRS R48887 twice, PBS, Al Jazeera). The SIPRI, CRS IN12641, CBO 62546, CSIS, Military Times and CRS RL33153 pages returned HTTP 429. NPR is blocked by robots.txt. Where we could only see a search-result headline, we say so.

1. Source check (Iran war items prioritized, as they are current events)

# · Claim (Brief/FC) · Cited figure · Verified figure · Status · URL
1 · War cost (B#18) · $37.5B as of Jul 21, 2026; $29B in May · Same. Hegseth and Gen. Caine testified to Senate Appropriations on Jul 21 · Confirmed · defensenews.com/…/2026/07/21/cost-of-iran-war-rises…
2 · Supplemental (B#18, FC#1, #18) · $67.1B; $21B munitions; $5.1B cyber · Same; about $32.7B is for direct operations · Confirmed · same
3 · Casualties (B#18, FC#23) · 18 killed, >440 injured (July) · Same as of July. Update: CRS gives 18 killed and 824 wounded as of Sep 15, 2026 · Confirmed (now dated) · everycrsreport.com/reports/R48887.html
4 · Timeline (B#26, FC#12) · Strikes Feb 28, 2026; ceasefire Apr 7; blockade Apr 13 · Same (CRS, updated Sep 18, 2026) · Confirmed · R48887
5 · Oil (B#26) · $72.48 (Feb 27) → $118.35 (Mar 31) · Same (Brent). CRS also gives $107.63 on Sep 10 · Confirmed · R48887
6 · "June MoU collapsed. Limited strikes resumed in September" (B#26) · Strikes resumed in September · MoU signed Jun 17. First post-MoU US strikes Jun 26. MoU declared over Jul 8. In early September the two sides "again exchanged attacks at levels approaching that of mid-July" · Minor discrepancy · R48887
7 · Senate WPR vote (B#27, FC#14) · 50–48 on Jun 23; Murkowski, Collins, Paul, Cassidy voted yes · Same (PBS, Jun 23). Fetterman voted no; McConnell and McCormick were absent · Confirmed · pbs.org/newshour/…
8 · House vote (B#27) · "a measure" passed 215–208 on Jun 3 · Tally confirmed. It was the same measure, H.Con.Res. 86, which the Senate then agreed to on Jun 23. It is a concurrent resolution, so it is never presented to the President (the Chadha problem Whitfield raised) · Minor discrepancy (framed as two measures) · R48887; aljazeera.com/…/2026/6/24/…
9 · Senate date conflict (B Contested vi, FC#14) · May 19 vs Jun 23 · Both dates are correct and refer to different measures. The Senate approved S.J.Res. 185 on May 19 (1 of 11 joint resolutions) and agreed to H.Con.Res. 86 on Jun 23. PBS's "first Senate passage" conflicts with CRS; we prefer CRS · Minor discrepancy (the conflict can be resolved) · R48887
10 · Sep 24 Senate resolution "failed narrowly" (B#27) · failed narrowly · Headlines say it failed "by one vote" (Baltimore Sun) and "narrowly" (CNBC). NPR body blocked · Confirmed (headline level) · npr.org/2026/09/24/…; baltimoresun.com/2026/09/24/…
11 · Supplemental range (B vi, FC#45) · $38.3B / $67.1B / ~$80B / $200B · CRS: the June request was $21B munitions + $17.3B operations = $38.3B. PBS: "approximately $80 billion." Defense News: $67.1B. The $200B figure was not checked · Confirmed (3 of 4) · R48887; PBS; Defense News
12 · Legal framing (FC#13) · "imminent threats," not the AUMF · CRS quotes Trump: "eliminate imminent threats from the Iranian regime." No AUMF is cited · Confirmed (supports the Unsupported rating and Whitfield's withdrawal) · R48887
— · 155mm Mesquite line (B#17) · $469M; no in-spec parts · Page returned 429. Several headlines agree ("$469 Million on a 155mm Plant That Hasn't Produced a Single Usable Shell Body") · Could not access (headline agrees) · militarytimes.com/…/2026/07/14/…

Counts: Confirmed 9 · Minor discrepancy 3 · Not supported 0 · Could not access 1. Also not accessible: SIPRI press release, CRS IN12641, CBO 62546, CSIS topline, CRS RL33153. A SIPRI search headline agrees with $2.887T world spending.

Additional current-events data the record should carry: CBO estimates $38B in operations cost as of Aug 1, 2026. Munitions expended were $22.3B as of Jun 29 (CRS). The House has also approved H.Con.Res. 89 (Jul 23) and H.Con.Res. 93 (Sep 15). The Senate has voted 11 times. Verdict "Unknown" bullet 1 should cite these figures.

2. Internal consistency

• Vote math: we checked all 21 cells. Every percentage matches its count after rounding, and every consensus label is correct: P1, P2 and P7 are YES (P2 Left 22 ≥ 21; P7 Right 22 ≥ 20), and P3, P4, P5 and P6 are NO. No mislabels.
• Fact-check tally: the table has 45 rows: 30 Supported, 3 Contested, 8 Unsupported, 4 Wrong. The erratum (post 297) is correct and the original "31" was wrong.
• FC#14 should be re-rated Supported. The 50–48 tally and the Jun 23 date are both confirmed. The May 19 passage was a separate joint resolution, so the dates were never actually in conflict.
• Verdict Established #5 ("resolutions passing both chambers") is accurate: H.Con.Res. 86 passed both. Add "concurrent, so non-binding under Chadha."
• Hendricks, Round 1: "36% of target a full year after the deadline." The deadline was Oct 2025 and the data point is Mar 2026, which is five months, not a year. Minor.
• Roster: all 21 named professors match the roster on number, field and lean. Everyone in the P3 Right-yes list is Lib or R. No mislabels.
• No claim rated Wrong reappears in the Verdict as established.

3. Balance

The four Wrong ratings fall on four different leans (C-R, Lib, Soc, C-L), and each has a stated basis in a source. We see no systematic tilt. One asymmetry is minor. Kessler's (R) "lowest share of GDP since before WWII" was rated Unsupported. Our recollection of OMB Historical Table 3.1 is that defense outlays were about 2.9% of GDP in FY1999–2001, which is at or below today's 3.0–3.1%; we did not fetch the table. If that holds, the claim is Wrong, not merely unverified. Since Kessler withdrew it, nothing downstream changes. The Verdict's "how, not how much" framing matches the vote pattern.

4. Grade: A−

The math is clean, the Wrong ratings are well founded, and the Iran record was handled cautiously. Points off for the "September" strike-resumption error, treating the House and Senate votes as separate measures, and a date conflict that CRS itself resolves.

— Pennington (lead); Brandt recused on FC rows