q7-mass-driver-feasibility

What engineering and capital work remains before a lunar mass driver is operational, how compressible is each step under industrial-explosion or TAI-level automation, and how much does the root answer depend on its availability vs. chemical-rocket-only delivery?

First-principles BAU capex $1,242B over 20-30 years (96% track + drive coils); industrial-explosion $127B / ~6 years; TAI-grade $13.3B / ~1.5 years (converges with q2's $10B assumption and NASA 1979 SP-428 inflated). Stall never reached. Cycle-life gap (10⁶-10⁹ shots required vs ~10³ demonstrated; Navy EMRG cancelled 2021) is the binding engineering risk independent of capital and TAI compression — requires materials-science breakthrough. q2's $50/kg mass-driver headline is TAI-conditional.

Confidence: medium-low

Q7 — Lunar Mass Driver Feasibility

Motivation

The question of when lunar manufacturing beats Earth launch for orbital infrastructure turns, in the late era, on a single architectural choice. Chemical rockets ascending from the lunar surface are punished by a propellant-use ratio of approximately fourteen at LEO (q4.c1) — meaning fourteen kilograms of propellant must move for each kilogram of product delivered. The mass driver was proposed in the 1970s by Gerard O'Neill and Henry Kolm as the way around that arithmetic: an electromagnetic launcher fixed to the lunar surface that fires bulk material into cislunar space using solar or fission electricity rather than imported chemical fuel. The 1979 NASA SP-428 Phase II reference design (Kolm, Fine, Mongeau, Williams) projected $3.137 billion 1979 dollars for operational capability by mid-1985. That target was not met. Forty years later, no operational lunar mass driver exists.

In 2025-2026 the concept has returned to public discourse: Elon Musk's "Project TERAFAB" announcements, Casey Handmer's May 2026 engineering brief, Phil Metzger's affirmative reception of the strategic pivot (q7.c14, q7.c10 — for the engineering specifics — and the broader Tier-B source-review). Whether that revival reflects a genuine engineering possibility or another false dawn determines whether the report's headline crossover — q2.c5's $50/kg late-era lunar-to-LEO via mass driver + SEP — is the realistic architecture or an aspirational anchor.

Where it fits

This leaf provides the engineering feasibility analysis for the mass-driver line in q2's lunar-ascent cost spectrum. Three sibling leaves feed in: q4 (gear-ratio) establishes that mass driver + SEP transfer is structurally required to reach LEO economically; q3 (ISRU feasibility) confirms that raw basalt projectiles do not depend on refined-product maturity, so the mass driver's projectile material is high-TRL; q2 (lunar-ascent-cost) integrates the mass-driver line into the overall lunar-to-LEO cost spectrum. The downstream synthesis is q8 (crossover): when does lunar manufacturing beat Earth launch? The mass driver's availability — its capital, throughput, cycle life, and the regime of acceleration governing its construction — sets when (or whether) the crossover lands in the report's planning window.

Headline answer

A lunar mass driver lands in the report's headline architecture only under industrial-explosion or TAI-grade compression of capital and engineering timelines. Under business-as-usual (BAU), the first-principles capital cost for a 1 Mt/yr operational system lands at approximately $1,242 billion dominated by the multi-kilometer track at $800M/m lunar-construction-multiplied EMALS pricing. Time to Mt-scale operational under BAU is 20-30 years, with sequential engineering milestones (Earth-based 100-launch demonstrator → sub-scale lunar prototype → kt-scale pilot → Mt-scale operational) each requiring physical lead-times that BAU compressibility cannot collapse. Under industrial explosion (~10× capital compression), the figures land at $127B / ~6 years. Under TAI-grade automation (~100× compression), the figures are $13.3B / ~1.5 years — within striking distance of q2.c5's $10B capital assumption.

Confidence: medium-low. The kinematic and energy-budget derivations are high-confidence physics; the capital cost components are low-confidence sourced-prior estimates. The regime-compression factors are speculative scenario priors. The Codex pass-2 audit appropriately downgraded the calc's confidence on these terms.

The implication for the report: q2.c5's $50/kg late-era headline is conditional on TAI-grade economics, not derivable under BAU. In any time horizon shorter than ~25 years under BAU progress, chemical rockets dominate the lunar-to-LEO architecture regardless of mass-driver availability. The "crossover happens late" framing in the synthesis must specify the acceleration regime it assumes.

Body

Kinematics and energy budget

Lunar escape velocity is 2,375 m/s; lunar low orbit (LLO) circular speed is 1,679 m/s — both derived from the lunar mass (7.342×10²² kg) and radius (1,737.4 km) via $v_{\text{esc}} = \sqrt{2GM/R}$ and $v_{\text{LLO}} = \sqrt{GM/R}$. Janhunen's 2024 result on mass-concentration gravity anomalies (arXiv:2410.09616) shows that projectiles launched at LLO speed can be caught from stable nine-day mascon-assisted orbits at equatorial nearside sites, eliminating the need to reach escape velocity and reducing the kinetic energy per kilogram by approximately fifty percent (1.41 MJ/kg at LLO vs 2.82 MJ/kg at escape).

The track length-acceleration trade is hard kinematics. For $v^2 = 2aL$, a muzzle velocity of 1.7 km/s requires 144 m of track at 1000 g, 1.4 km at 100 g, or 7.2 km at 20 g. The Handmer 2026 design lands at the 1000 g / 144 m / LLO-speed end of this envelope. Modern engineering work has converged toward sub-escape muzzle velocities with orbital tug catching, away from the 1970s O'Neill / NASA SP-428 escape-velocity-to-L5 architecture (q7.c1, q7.c2).

System efficiency and power

Net system efficiency spans a substantial range. The DARPA 45-stage coilgun, the most-recently-documented multi-stage coilgun in the engineering literature, achieved 22 percent efficiency. Wright, Kuznetsov, and Kloesel (2011 IEEE / NASA, the load-bearing modern peer-reviewed reanalysis) used 33 percent as TRL-grounded. Handmer's 2026 engineer-advocate brief assumes 90 percent driver efficiency. The 1979 NASA SP-428 chapter claimed 96.4 percent system efficiency — a figure modern reviewers interpret as electromagnetic-conversion-only, not net system efficiency, though the SP-428 chapter itself does not explicitly limit its scope (q7.c3, source-review of nasa-mass-drivers-iii-1979 with the "Merits investigation" flag).

At 50 percent net efficiency (a conservative low-end-of-aspirational assumption — not the arithmetic midpoint), the wall energy per kg is 2.82 MJ/kg at LLO speed. At one Mt/yr operational throughput the average wall power is approximately 89 MW; at the 10 Mt/yr Handmer aspirational nameplate it rises to 894 MW. Peak instantaneous power per 200 kg shot at 1000 g is approximately 6.6 GW over 171 milliseconds — Handmer's 16 GW figure includes the loaded sled at 1000 kg total mass rather than the unencumbered 200 kg projectile baseline. Both are internally consistent within their architectural framing (q7.c4).

Pulsed-power and capacitor bank

Energy storage per shot at 200 kg / 1.7 km/s / 90 percent driver efficiency is approximately 315 MJ. At defense-grade pulsed-power capacitor energy density (~2 kJ/kg from the broader pulsed-power engineering literature, not specifically the DSIAC source despite the qualitative pulsed-power-bottleneck framing), the per-shot capacitor mass is approximately 157 metric tonnes; with 5× redundancy and $10/kJ at scale plus Earth-to-Moon shipping, the bank capex is approximately $0.8 billion. This is a sub-dominant cost line. At one Mt/yr throughput (5 million shots per year, one shot every 6.3 seconds) the capacitor bank must recycle within ~6 seconds — well within state-of-art power-electronics switching, though thermal load, pulse shaping, and the cycle-life requirement (separate consideration below) are not addressed by switching speed alone (q7.c5).

Capital cost decomposition (BAU)

Component-by-component first-principles estimate for a 1 Mt/yr, 100 g track (1.5 km), η=0.50 architecture. Lunar construction multiplier is 200× Earth-construction baseline (a sourced-prior with wide uncertainty range: plausibly 50× to 1000×).

Component	Capex (BAU)
Track + drive coils (1500 m × $800M/m)	$1,200B
Reactor (200 MW class, Earth-built + shipping)	$3B
Capacitor bank (5× redundancy)	$0.8B
Projectile manufacturing plant (ISRU sintering)	$6B
Catcher / orbital tug	$0.15B
Foundation / anchor / regolith burial	$2B
Control / instrumentation	$0.5B
Construction labor (3-year build, $10B/yr peak)	$30B
Total	$1,242B

Track plus drive coils dominate at approximately 96 percent of total. This single component is the load-bearing variable in the BAU capex — and the primary axis along which TAI compression is expected to act (modular pre-fab assembly + ISRU-built drive coils + reduced human-hours bring the per-meter cost down by approximately two orders of magnitude under TAI; ~10× under industrial explosion) (q7.c6).

Per-kg amortized cost

At 20-year operational lifetime and 1 Mt/yr throughput (20 Mt total), the per-kg amortized cost to LLO under BAU decomposes:

Component	Per-kg, BAU
Capex amortized over 20 Mt lifetime throughput	$62.12/kg
Annual power consumption × $0.50/kWh lunar electricity	$0.39/kg
Operations and maintenance (5% of capex per year)	$62.12/kg
Total	$124.64/kg

This is the BAU first-principles per-kg cost for delivering bulk mass to LLO. It does not include the SEP transfer leg from LLO to LEO that q2.c5 includes in its $50/kg headline — the reconcile pass deferred that bridge as an explicit reconciliation rather than an extrapolation. The figure substantially diverges from q2's $50/kg only under BAU; under TAI-grade compression the two converge structurally (q7.c7).

Regime sensitivity

The four-regime framework (q7.c8; consistent with q1.c7, q3.c8, q4 on cross-leaf regime treatment):

Regime	Capex compression	Total capex	Time to Mt-scale	Per-kg amortized capex
Business-as-usual	1.0× baseline	$1,242B	20-30 yr	$62/kg
Industrial explosion	~10×	$127B	~6 yr	$6.3/kg
TAI-grade	~100×	$13.3B	~1.5 yr	$0.66/kg
Stall	1.0× (never built)	—	never	N/A

The TAI-grade endpoint converges with two independent reference points: q7.c8's $13.3B sits within an order of magnitude of q2.c5's $10B capital assumption (q7.c8, q7.c11) and matches the 1979 NASA SP-428 figure inflated to 2026 dollars ($3.137B 1979 ≈ $13B 2026, per q7.c12). This convergence is consistent with the interpretation that the 1979 baseline projection implicitly assumed engineering compression that the 1979-2025 historical record did not realize. Wright et al. 2011 provided the TRL-grounded reanalysis with the negative cost-benefit verdict ("the conclusion, however, is not as favorable for LEML as originally suggested") — this is the load-bearing modern primary source supporting the BAU pessimism (q7.c12).

Engineering milestones

Five sequential physical milestones must be crossed before operational deployment. Timelines per regime (scenario priors, not externally validated):

Milestone	Specification	BAU	IE	TAI
M1 — Earth-based demonstrator	1.7 km/s muzzle, 100 launches without component replacement	2-4 yr	6 mo	weeks
M2 — Sub-scale lunar prototype	5-50 t mass, 100 m track, 100 kg payload, 10 launches	5-8 yr	18 mo	6 mo
M3 — Pilot kt-scale operational	1-10 kt/yr, 10⁴-10⁵ cycle life	10-15 yr	3 yr	1 yr
M4 — Mt-scale operational	1 Mt/yr, ISRU-supplied projectiles	20-30 yr	6 yr	1.5 yr
M5 — 10 Mt-scale nameplate	Handmer aspirational, Project TERAFAB-class	40-60+ yr	10 yr	2-3 yr

Peterkin (General Atomics, EMALS builder) identified M1 as the next required demonstration: "to prove viability, we need to demonstrate that this approach can achieve a lunar orbit speed for at least 100 launches without needing to replace launcher components." The closest historical analog is the Navy EMRG program (32 MJ muzzle energy, Mach 7.5, cancelled 2021 after approximately $500M over cycle-life and barrel-wear issues) — a sobering precedent. SpinLaunch, the most-funded modern private Earth-based kinetic launcher, has demonstrated 10,000 g acceleration survival for a 1U CubeSat (August 2025) after $150M+ and a decade of development, then strategically pivoted to chemical-rocket delivery for its Meridian constellation. These historical analogs are consistent with q7's BAU M2-M3 estimate of 10-15 years from program start to pilot-scale operational; political will, capital scale, and engineering closure are all binding (q7.c9, q7.c15).

Cycle-life gap — the binding engineering risk

The single most under-discussed engineering risk in the mass-driver literature is the cycle-life gap between demonstrated and operational requirements. At 1 Mt/yr × 200 kg/shot × 20 yr lifetime, the system requires approximately 100 million shots. At 10 Mt/yr (q2's nameplate assumption), approximately one billion shots over 20 years. State-of-art demonstrated cycle life for EM launchers is approximately 100-1,000 shots — the Navy EMRG cancellation rationale cited cycle-life and barrel-wear as the binding failure modes (DSIAC 2015). Handmer's 2026 engineering brief contains the strongest single-source corroboration: "sintered magnet blocks 9 m long under 1000 g shear and oscillating tension fatigue is not obviously feasible."

This 5-7 orders of magnitude gap between demonstrated and required cycle life is independent of capital cost and TAI compression. No level of capital allocation builds an operational mass driver if the underlying engineering does not close a path to 10⁶-10⁹-shot cycle life. This gap is the most critical engineering uncertainty in the feasibility analysis, and it would benefit from a dedicated q7.1 sub-leaf in a future tree-pass (q7.c10).

Alternative architecture: the lunar space elevator

The root question — when does lunar manufacturing beat Earth launch? — does not strictly depend on mass-driver availability. Pearson, Levin, Oldson, and Wykes (NIAC Phase II Final Report, 2005) demonstrated that a lunar space elevator using commercially available M5 fiber (Kevlar-class para-aramid) is technically feasible without requiring carbon nanotubes. A 30 mm × 0.023 mm M5 ribbon supports 2,000 kg at the lunar surface and 100 cargo vehicles of 580 kg each distributed along the length. Climbing-vehicle energy demand drops from approximately 10 kW at the surface to under 100 W at seven percent of the distance to L1. The 2005-2025 historical record — zero commercial follow-on after NIAC funding ended, LiftPort 2019 reporting "no progress beyond the lunar elevator company's conceptualized design" — is consistent with the binding constraint being capital and program commitment rather than physics (q7.c13). The lunar elevator avoids the cyclic-load fatigue, GW-scale pulsed power, and precision projectile-capture problems of a mass driver; it is the canonical alternative architecture for cis-lunar bulk transport.

Public-figure positions (tier-B summary)

Modern public-figure positions on lunar mass drivers (mid-2020s), per q7.c14 and the source-review tier-B treatment:

Robert Peterkin (General Atomics EMS, EMALS builder): credentialed industry-engineering voice; identifies the M1 milestone explicitly; advocates evolution of EMALS toward lunar EM launch as the next funded R&D step.
Elon Musk: SpaceX/Tesla/xAI strategic commitment via "Project TERAFAB" (Feb-May 2026), but no funded engineering program. Aspirational figures: $500/kg, 500-1000 TW/yr of AI satellite deployment, "Path to Petawatts." Musk's personal lifetime-horizon framing ("I want to live long enough to see it") is consistent with q7's BAU decades-scale assessment. The 500-1000 TW/yr capacity claim sits 100-1000× above Handmer's 10 Mt/yr nameplate (under reasonable mass intensity assumptions) and is contradicted by q7's BAU and IE regimes; under aggressively low kg-per-kW assumptions for ultra-thin satellites the gap closes, but the figure is not engineering-grounded.
Phil Metzger (UCF, canonical academic on lunar manufacturing economics, q4 anchor source): qualitative paradigm-level affirmation of Musk's pivot. Does not address engineering feasibility timeline or capital. Meaningful as a signal that the strategic pivot has academic validation.
Casey Handmer: load-bearing modern engineer-advocate. The May 2026 engineering brief is the closest thing the literature has to a contemporary technical design. Headline economics are aspirational (10 Mt/yr nameplate, $10/kg "rocks in lunar orbit") but the engineering caveats are explicit ("not obviously feasible" magnet fatigue).

Dependence of the root question

The root question's answer — when does lunar manufacturing beat Earth launch for orbital infrastructure? — depends on mass-driver availability only in the late-era / TAI-grade regimes. Under BAU and any time horizon shorter than approximately 25 years, the engineering milestones M2-M5 require sequential physical lead-times that BAU compressibility cannot collapse, so mass drivers do not reach Mt-scale operational regardless of investment level. In that case chemical rockets dominate the lunar-to-LEO architecture and the root question's answer collapses to whether chemical-aggressive-ISRU late-era ($994/kg per q2.c3) can beat Starship-optimistic-late ($59/kg per q1) — which it cannot at scale.

Under industrial explosion, mass drivers reach Mt-scale operational in approximately six years; per-kg amortized capex drops to approximately $6/kg at LLO; with the q2-derived SEP transfer leg added, the lunar-to-LEO figure is competitive with optimistic Starship and beats partial-reuse Starship. Lunar manufacturing for orbital infrastructure becomes economically rational. Under TAI-grade compression, the same milestones complete in approximately 1.5 years and the per-kg figure drops to approximately $0.66/kg amortized. Under stall, the question is moot.

The mass driver is therefore a scenario-conditional load-bearing variable for the report's headline crossover. The report's central architectural argument (q4 → q7 → q2 → q8) is internally consistent, but its quantitative headline is conditional on the acceleration regime the synthesis assumes (q7.c11).

Confidence per finding

Claim	Confidence	Reasoning
q7.c1: kinematic constants (escape, LLO, energy per kg)	high	First-principles physics
q7.c2: kinematic track-acceleration trade	high	First-principles physics
q7.c3: efficiency range (22-90%)	medium	Engineering ceiling sourced-priors
q7.c4: average and peak power	high	First-principles given efficiency assumption
q7.c5: capacitor bank parameters	medium	Engineering priors + first-principles arithmetic
q7.c6: BAU capex decomposition ($1,242B)	low	Track-per-meter sourced-prior has wide uncertainty (50×-1000×)
q7.c7: BAU per-kg amortized ($125/kg)	low	Inherits q7.c6 uncertainty plus O&M scenario assumption
q7.c8: regime sensitivity capex / time-to-ops	speculative	Compression factors are scenario priors per anti-pattern #11
q7.c9: M1-M5 milestones	speculative	Scenario construction; cycle-life-gap caveat applies
q7.c10: cycle-life gap (100M-1B shots required, ~1000 demonstrated)	medium-high	Codex audit-surfaced; strong source corroboration (Handmer, DSIAC, Navy EMRG cancellation)
q7.c11: root-question dependence (regime-conditional)	medium	Structural finding; survives source-review
q7.c12: historical 1979-2011 NASA arc	high	Documented in primary sources
q7.c13: lunar elevator alternative architecture	high	Documented in NIAC Phase II report
q7.c14: public-figure positions	medium-high	Quoted statements verified in extracts
q7.c15: SpinLaunch analog	high	Documented in trade press

Limitations

Numerical:

Track per-meter cost (200× lunar multiplier on EMALS Earth pricing) is the single most uncertain input. Plausible range 50×-1000×; the BAU capex swings from $300B to $6T across this range.
Regime-compression factors (track 100× under TAI, construction 50×) are scenario priors with no external calibration. If TAI compression is only 10×, the TAI capex rises to $100B and the q2.c5 headline collapses.
The Acta Astronautica 1980 O'Neill-Kolm paper and AIAA 2025-4123 paper were not fully fetched in this iteration; the latter in particular would provide the most recent peer-reviewed sizing anchor.

Engineering:

The cycle-life gap (10⁶-10⁹ shots required vs ~10³ demonstrated) is the binding constraint. q7 does not derive a path to closing this gap — that should be a dedicated q7.1 sub-leaf in a future tree-pass.
The calc derives muzzle velocity and energy budget, not full orbit-insertion engineering. Launch angle dispersion, abort handling, catcher ΔV requirements, and trajectory precision (Janhunen mascon-orbit + tug catching architecture) are out of scope here. The Handmer 2026 design flag on "velocity dispersion >10⁻⁴ makes projectiles orbital debris" requires dedicated treatment.

Architectural:

The lunar space elevator alternative (Pearson NIAC 2005) is not deeply explored in this leaf. If the elevator is more easily engineered than the mass driver, the q4 → q7 → q2 chain has a competing branch.
The mass-driver-vs-chemical-only choice in q2 is itself conditional on the regime: under BAU neither path delivers a Mt-scale lunar industry within the planning horizon, and the report's central question collapses to "Earth launch wins indefinitely."

Methodological:

The "first-principles" framing in the original calc was overstated; many inputs are sourced priors. Codex pass-2 audit accepted this; the calc and write are now properly labeled.
One Codex-flagged hallucination on a Wright efficiency quote was caught and corrected in source-review.
The Acta Astronautica 1980 and AIAA 2025-4123 PDF fetches are pending; bibliographic anchors are in place but full Tier-S claim-by-claim review awaits.

What could change the answer: A breakthrough in cycle-life engineering (specifically the magnet-fatigue and barrel-wear problems flagged by Handmer 2026 and DSIAC 2015) would substantially upgrade the BAU feasibility verdict — possibly by an order of magnitude. A demonstrated kg-per-kW reduction for AI satellites to ≤0.1 kg/kW would partially reconcile Musk's 500-1000 TW/yr claim with q7's BAU envelope. A funded national-program-scale capital commitment to mass-driver engineering (comparable to the Apollo program's $250B equivalent 2026 dollars, not the EMRG program's $500M) would shift the BAU timeline toward the IE regime. None of these conditions is currently met.

Flag for Avi

q2-q7 capital contradiction (load-bearing): q2.c5's $10B mass-driver capital implicitly assumes TAI-grade engineering compression; q7.c6 explicitly enumerates the $1,242B BAU baseline. The cross-consistency / synthesis pass should foreground this. The report's headline crossover ($50/kg late-era via mass driver + SEP) is conditional on the acceleration regime — under BAU the crossover does not happen, period.
Cycle-life gap as candidate sub-leaf: q7.1 should investigate specific engineering paths to 10⁶-10⁹-shot cycle life. This is the binding constraint and the most under-discussed risk.
Missing primary sources: O'Neill-Kolm Acta Astronautica 1980 PDF; AIAA 2025-4123 full PDF; SP-428 efficiency-claim subsystem breakdown. All worth fetching in a future iteration.
Public-figure calibration: Musk's 500-1000 TW/yr / $500/kg figures are aspirational, not engineering-grounded; the report should distinguish strategic-direction-from-principal vs. engineering-feasibility-claim.

Evidence agreement

supports partial contradicts none

q7.c1 Lunar escape velocity is 2,375 m/s and lunar low orbit (LLO) circular speed is 1,679 m/s, derived from the lunar mass (7.342e22 kg) and radius (1,737.4 km) via v_esc = sqrt(2GM/R) and v_orbit = sqrt(GM/R). Mascon-anomaly-assisted designs (Janhunen 2024) can catch projectiles launched at LLO speed (1.7 km/s) rather than escape (2.4 km/s), reducing kinetic energy per kg by ~50% (1.41 MJ/kg vs 2.82 MJ/kg).

first
janhunen
lunarpedia
handmer
q7.c2 Constant-acceleration kinematics fix the track-length × acceleration trade: for v=1.7 km/s, a 1000g profile requires 144 m, 100g requires 1.4 km, 20g requires 7.2 km; for v=2.4 km/s, lengths roughly double. This is the inescapable physical trade between track length and projectile g-loading.

first
q7.c3 Net system efficiency for a lunar mass driver is bounded by the demonstrated range: 22% (DARPA 45-stage coilgun, modern Earth-tested) at the low end to 90% (Handmer 2026 engineer-advocate optimistic). The 96% claimed in 1979 NASA SP-428 is electromagnetic-conversion-only, not net system efficiency. A conservative low-end-of-aspirational midpoint of 50% gives wall energy 2.82 MJ/kg at 1.7 km/s and 5.64 MJ/kg at 2.4 km/s.

first
coilgun
wright
nasa
handmer
q7.c4 At operational throughput of 1 Mt/yr, average wall power is approximately 89 MW (η=0.50, v=1.7 km/s). Peak instantaneous power per 200-kg shot at 1000g is approximately 6.6 GW over 171 ms. This peak-to-average ratio (~75x) sets the pulsed-power capacitor bank / flywheel requirement. At 10 Mt/yr (Handmer aspirational nameplate), average wall power rises to 894 MW (consistent with Handmer's 450 MW at his higher 90% efficiency assumption).

first
handmer
q7.c5 Energy storage per shot at 200 kg, 1.7 km/s, η=0.90 is approximately 315 MJ. At defense-grade pulsed-power capacitor energy density (~2 kJ/kg) this requires approximately 157 t of capacitors per shot; with 5× redundancy and $10/kJ at scale, the per-shot capacitor capex is approximately $3.1M and the bank capex is approximately $0.8B including Earth-to-Moon shipping. At 1 Mt/yr throughput (5M shots/yr, 6.3 s interval), recycle time is well within state-of-art switching.

first
q7.c6 Independent component-by-component BAU capital decomposition for a 1 Mt/yr, 100g track (1.5 km), η=0.50 architecture lands at approximately $1,242B total. Track + drive coils dominate ($1,200B, ~96% of total), driven by a sourced-prior 200× lunar construction multiplier applied to EMALS per-meter cost. Other line items: reactor $3B, capacitor bank $0.8B, projectile manufacturing plant $6B, catcher $0.15B, foundation $2B, control $0.5B, construction labor $30B.

first
q7.c7 Per-kg amortized cost to LLO under BAU (1 Mt/yr × 20 yr, $1,242B capex, $0.50/kWh lunar electricity, 5%/yr O&M) is approximately $125/kg (capex $62 + O&M $62 + power $0.40). This does NOT include the SEP transfer leg from LLO to LEO (which q2.c5 includes in its $50/kg headline). The BAU result substantially diverges from q2's $50/kg figure; the divergence collapses only under TAI-grade compression.

first
q7.c8 Regime sensitivity for capex and time-to-Mt-scale-operational: BAU ($1,242B / 20-30 yr), Industrial Explosion ($127B / 6 yr), TAI-grade ($13.3B / 1.5 yr), Stall (never). The TAI-grade figure is within striking distance of q2.c5's $10B capital assumption, meaning q2's $50/kg headline is achievable only under TAI-grade compression of the BAU baseline; under any time horizon shorter than ~25 years BAU, chemical rockets dominate the architecture regardless of mass-driver availability.

first
q7.c9 Discrete engineering milestones to Mt-scale operational, with regime-dependent timelines (scenario priors per anti-pattern #11): M1 (Earth-based demo at 1.7 km/s with 100-launch cycle life) 2-4 yr BAU / 6 mo IE / weeks TAI; M2 (sub-scale lunar prototype, 100 m track, 100 kg payload, 10 launches) 5-8 yr / 18 mo / 6 mo; M3 (pilot kt-scale operational, 10^4-10^5 cycle life) 10-15 yr / 3 yr / 1 yr; M4 (Mt-scale operational, ISRU-supplied projectiles) 20-30 yr / 6 yr / 1.5 yr; M5 (10 Mt-scale Handmer nameplate) 40-60+ yr / 10 yr / 2-3 yr.

first
q7.c10 Cycle-life gap is the most under-discussed engineering risk in the mass-driver literature. At 1 Mt/yr × 200 kg/shot × 20 yr the system requires approximately 100 million shots; at 10 Mt/yr (q2's nameplate) approximately 1 billion shots. State-of-art demonstrated cycle life for EM launchers is approximately 100-1,000 shots; the Navy EMRG program was cancelled in 2021 after roughly $500M investment with cycle-life and barrel-wear as the binding failure modes. The 5-7 orders of magnitude gap between demonstrated and required cycle life is independent of capital cost and TAI compression; it requires an engineering breakthrough to close.

first
dsiac
coilgun
handmer
q7.c12 The 1979 NASA SP-428 Mass Drivers III chapter projected $3.137 billion (1979 dollars) for a baseline operational lunar mass driver by mid-1985, with 42 kg/s throughput, 96.4% claimed system efficiency, and 488 m total length. Inflated to 2026 dollars, this is approximately $13 billion. The 1985 operational target was not met; no operational lunar mass driver has been built in the intervening 40 years. The 2011 Wright et al. NASA/IEEE peer-reviewed reanalysis with TRL-grounded subsystems reached a negative cost-benefit conclusion.

nasa
nasa
wright
kolm
q7.c13 Pearson, Levin, Oldson, Wykes (NIAC Phase II Final Report, 2005) demonstrated that a lunar space elevator using commercially available M5 fiber is technically feasible without carbon nanotubes. A 30mm × 0.023mm M5 ribbon supports 2,000 kg at the lunar surface and 100 cargo vehicles of 580 kg each distributed along the length. This is the canonical non-mass-driver architecture for cis-lunar bulk transport. From 2005 NIAC funding to 2019 LiftPort status update, no progress was made beyond conceptualized design — the bottleneck for this alternative is capital and program commitment, not physics.

pearson
pearson
pearson
q7.c14 Modern public-figure positions on lunar mass drivers (mid-2020s): Elon Musk publicly committed strategic intent through 'Project TERAFAB' announcements in Feb-May 2026, with aspirational figures of $500/kg and 500-1000 TW/yr of AI satellite deployment, but no funded engineering program. Phil Metzger treated Musk's pivot as a vindication of the O'Neill paradigm. Robert Peterkin (General Atomics EMS, EMALS builder) advocates evolutionary path from EMALS to lunar EM launch, with the next demonstration milestone defined as '100 launches at lunar orbit speed without needing to replace launcher components.'

musk
musk
musk
musk
metzger
peterkin
q7.c15 SpinLaunch is the most-funded modern Earth-based kinetic launch development. Since 2014, approximately $150M private capital has been deployed; in August 2025 the company demonstrated 10,000g acceleration survival for a 1U CubeSat. Full operational system design targets 2.1 km/s muzzle velocity with 400 kg payload. April 2025 strategic pivot: announced Meridian (280-satellite constellation) launched on conventional chemical rockets — implying the operational kinetic system is not on a near-term schedule. The trajectory is consistent with q7's BAU M2-M3 estimate of 10-15 years from program start to pilot-scale operational; political will and capital scale (not physics) are the binding constraints on Earth-based kinetic launch.

spinlaunch
spinlaunch
spinlaunch
q7.c11 The root question's answer — when does lunar manufacturing beat Earth launch for orbital infrastructure? — depends on mass driver availability ONLY in the late-era / TAI-grade regimes. Under BAU and any time horizon shorter than ~25 years, mass drivers do not reach Mt-scale operational regardless of investment level (the engineering milestones M2-M5 require sequential physical lead-times that BAU compressibility cannot collapse). In the early-mid BAU eras chemical rockets dominate the architecture by an order of magnitude regardless of mass-driver availability. Under Industrial Explosion or TAI-grade compression, mass drivers reach operational scale within the report's planning horizon and become the dominant architecture.

first

Pass status

✓ research ✓ calc ✓ reconcile ✓ source-review ✓ consistency ✓ write

Claims (15) · evidence + audit status

q7.c1 Lunar escape velocity is 2,375 m/s and lunar low orbit (LLO) circular speed is 1,679 m/s, derived from the lunar mass (7.342e22 kg) and radius (1,737.4 km) via v_esc = sqrt(2GM/R) and v_orbit = sqrt(GM/R). Mascon-anomaly-assisted designs (Janhunen 2024) can catch projectiles launched at LLO speed (1.7 km/s) rather than escape (2.4 km/s), reducing kinetic energy per kg by ~50% (1.41 MJ/kg vs 2.82 MJ/kg).

high ✓

first-principles-calc ✓ janhunen-2024-gravity-anomalies ✓ lunarpedia-mass-drivers ✓ handmer-mass-driver-2026 ✓
q7.c2 Constant-acceleration kinematics fix the track-length × acceleration trade: for v=1.7 km/s, a 1000g profile requires 144 m, 100g requires 1.4 km, 20g requires 7.2 km; for v=2.4 km/s, lengths roughly double. This is the inescapable physical trade between track length and projectile g-loading.

high ✓

first-principles-calc ✓
q7.c3 Net system efficiency for a lunar mass driver is bounded by the demonstrated range: 22% (DARPA 45-stage coilgun, modern Earth-tested) at the low end to 90% (Handmer 2026 engineer-advocate optimistic). The 96% claimed in 1979 NASA SP-428 is electromagnetic-conversion-only, not net system efficiency. A conservative low-end-of-aspirational midpoint of 50% gives wall energy 2.82 MJ/kg at 1.7 km/s and 5.64 MJ/kg at 2.4 km/s.

medium ⚠

first-principles-calc ✓ coilgun-wiki-engineering ✓ wright-kuznetsov-kloesel-2011 ✓ nasa-mass-drivers-iii-1979 handmer-mass-driver-2026 ✓
q7.c4 At operational throughput of 1 Mt/yr, average wall power is approximately 89 MW (η=0.50, v=1.7 km/s). Peak instantaneous power per 200-kg shot at 1000g is approximately 6.6 GW over 171 ms. This peak-to-average ratio (~75x) sets the pulsed-power capacitor bank / flywheel requirement. At 10 Mt/yr (Handmer aspirational nameplate), average wall power rises to 894 MW (consistent with Handmer's 450 MW at his higher 90% efficiency assumption).

high ⚠

first-principles-calc ✓ handmer-mass-driver-2026
q7.c5 Energy storage per shot at 200 kg, 1.7 km/s, η=0.90 is approximately 315 MJ. At defense-grade pulsed-power capacitor energy density (~2 kJ/kg) this requires approximately 157 t of capacitors per shot; with 5× redundancy and $10/kJ at scale, the per-shot capacitor capex is approximately $3.1M and the bank capex is approximately $0.8B including Earth-to-Moon shipping. At 1 Mt/yr throughput (5M shots/yr, 6.3 s interval), recycle time is well within state-of-art switching.

medium ⚠

first-principles-calc ✓
q7.c6 Independent component-by-component BAU capital decomposition for a 1 Mt/yr, 100g track (1.5 km), η=0.50 architecture lands at approximately $1,242B total. Track + drive coils dominate ($1,200B, ~96% of total), driven by a sourced-prior 200× lunar construction multiplier applied to EMALS per-meter cost. Other line items: reactor $3B, capacitor bank $0.8B, projectile manufacturing plant $6B, catcher $0.15B, foundation $2B, control $0.5B, construction labor $30B.

low ⚠

first-principles-calc ✓
q7.c7 Per-kg amortized cost to LLO under BAU (1 Mt/yr × 20 yr, $1,242B capex, $0.50/kWh lunar electricity, 5%/yr O&M) is approximately $125/kg (capex $62 + O&M $62 + power $0.40). This does NOT include the SEP transfer leg from LLO to LEO (which q2.c5 includes in its $50/kg headline). The BAU result substantially diverges from q2's $50/kg figure; the divergence collapses only under TAI-grade compression.

low ⚠

first-principles-calc ✓
q7.c8 Regime sensitivity for capex and time-to-Mt-scale-operational: BAU ($1,242B / 20-30 yr), Industrial Explosion ($127B / 6 yr), TAI-grade ($13.3B / 1.5 yr), Stall (never). The TAI-grade figure is within striking distance of q2.c5's $10B capital assumption, meaning q2's $50/kg headline is achievable only under TAI-grade compression of the BAU baseline; under any time horizon shorter than ~25 years BAU, chemical rockets dominate the architecture regardless of mass-driver availability.

speculative ⚠

first-principles-calc ✓
q7.c9 Discrete engineering milestones to Mt-scale operational, with regime-dependent timelines (scenario priors per anti-pattern #11): M1 (Earth-based demo at 1.7 km/s with 100-launch cycle life) 2-4 yr BAU / 6 mo IE / weeks TAI; M2 (sub-scale lunar prototype, 100 m track, 100 kg payload, 10 launches) 5-8 yr / 18 mo / 6 mo; M3 (pilot kt-scale operational, 10^4-10^5 cycle life) 10-15 yr / 3 yr / 1 yr; M4 (Mt-scale operational, ISRU-supplied projectiles) 20-30 yr / 6 yr / 1.5 yr; M5 (10 Mt-scale Handmer nameplate) 40-60+ yr / 10 yr / 2-3 yr.

speculative ⚠

first-principles-calc ✓
q7.c10 Cycle-life gap is the most under-discussed engineering risk in the mass-driver literature. At 1 Mt/yr × 200 kg/shot × 20 yr the system requires approximately 100 million shots; at 10 Mt/yr (q2's nameplate) approximately 1 billion shots. State-of-art demonstrated cycle life for EM launchers is approximately 100-1,000 shots; the Navy EMRG program was cancelled in 2021 after roughly $500M investment with cycle-life and barrel-wear as the binding failure modes. The 5-7 orders of magnitude gap between demonstrated and required cycle life is independent of capital cost and TAI compression; it requires an engineering breakthrough to close.

medium-high ✓

first-principles-calc ✓ dsiac-hypervelocity-2015 ✓ coilgun-wiki-engineering ✓ handmer-mass-driver-2026 ✓
q7.c12 The 1979 NASA SP-428 Mass Drivers III chapter projected $3.137 billion (1979 dollars) for a baseline operational lunar mass driver by mid-1985, with 42 kg/s throughput, 96.4% claimed system efficiency, and 488 m total length. Inflated to 2026 dollars, this is approximately $13 billion. The 1985 operational target was not met; no operational lunar mass driver has been built in the intervening 40 years. The 2011 Wright et al. NASA/IEEE peer-reviewed reanalysis with TRL-grounded subsystems reached a negative cost-benefit conclusion.

high ○

nasa-mass-drivers-iii-1979 ✓ nasa-mass-drivers-iii-1979 ✓ wright-kuznetsov-kloesel-2011 ✓ kolm-l5-news-1980 ✓
q7.c13 Pearson, Levin, Oldson, Wykes (NIAC Phase II Final Report, 2005) demonstrated that a lunar space elevator using commercially available M5 fiber is technically feasible without carbon nanotubes. A 30mm × 0.023mm M5 ribbon supports 2,000 kg at the lunar surface and 100 cargo vehicles of 580 kg each distributed along the length. This is the canonical non-mass-driver architecture for cis-lunar bulk transport. From 2005 NIAC funding to 2019 LiftPort status update, no progress was made beyond conceptualized design — the bottleneck for this alternative is capital and program commitment, not physics.

high ○

pearson-lunar-elevator-2005 ✓ pearson-lunar-elevator-2005 ✓ pearson-lunar-elevator-2005 ✓
q7.c14 Modern public-figure positions on lunar mass drivers (mid-2020s): Elon Musk publicly committed strategic intent through 'Project TERAFAB' announcements in Feb-May 2026, with aspirational figures of $500/kg and 500-1000 TW/yr of AI satellite deployment, but no funded engineering program. Phil Metzger treated Musk's pivot as a vindication of the O'Neill paradigm. Robert Peterkin (General Atomics EMS, EMALS builder) advocates evolutionary path from EMALS to lunar EM launch, with the next demonstration milestone defined as '100 launches at lunar orbit speed without needing to replace launcher components.'

medium-high ○

musk-mass-driver-tweet-2026 ✓ musk-mass-driver-tweet-2026 ✓ musk-mass-driver-tweet-2026 ✓ musk-mass-driver-tweet-2026 ✓ metzger-twitter-2025 ✓ peterkin-general-atomics-2023 ✓
q7.c15 SpinLaunch is the most-funded modern Earth-based kinetic launch development. Since 2014, approximately $150M private capital has been deployed; in August 2025 the company demonstrated 10,000g acceleration survival for a 1U CubeSat. Full operational system design targets 2.1 km/s muzzle velocity with 400 kg payload. April 2025 strategic pivot: announced Meridian (280-satellite constellation) launched on conventional chemical rockets — implying the operational kinetic system is not on a near-term schedule. The trajectory is consistent with q7's BAU M2-M3 estimate of 10-15 years from program start to pilot-scale operational; political will and capital scale (not physics) are the binding constraints on Earth-based kinetic launch.

high ○

spinlaunch-status-2025 ✓ spinlaunch-status-2025 ✓ spinlaunch-status-2025 ✓
q7.c11 The root question's answer — when does lunar manufacturing beat Earth launch for orbital infrastructure? — depends on mass driver availability ONLY in the late-era / TAI-grade regimes. Under BAU and any time horizon shorter than ~25 years, mass drivers do not reach Mt-scale operational regardless of investment level (the engineering milestones M2-M5 require sequential physical lead-times that BAU compressibility cannot collapse). In the early-mid BAU eras chemical rockets dominate the architecture by an order of magnitude regardless of mass-driver availability. Under Industrial Explosion or TAI-grade compression, mass drivers reach operational scale within the report's planning horizon and become the dominant architecture.

medium ⚠

first-principles-calc ✓

Sources cited (18) · expandable

aiaa-2025-4123-mass-driver-tech Cost-Benefit Analysis of Lunar Mass Driver Technologies

unknown — paywalled, abstract only 2025 paper

https://arc.aiaa.org/doi/abs/10.2514/6.2025-4123
show extract (2.6k chars)
Abstract

AIAA paper 2025-4123 "Cost-Benefit Analysis of Lunar Mass Driver Technologies" (AIAA Aviation Forum and ASCEND 2025) develops parametric sizing models for railgun and coilgun lunar mass-driver designs. The abstract describes the work as outputting "the overall system mass and size, and the required power and energy to launch a specific payload" for the trade space of mission requirements. Two primary architectures compared: railgun (high-burst current, single force pulse, simpler) and coilgun (sequential staged coils, gentler acceleration, preferred for sensitive payloads). The paper is paywalled at arc.aiaa.org and the full content was not fetched in this session — only the abstract and search-summarized findings are captured here. Triangulation against Handmer 2026 and Wright et al. 2011 suggests this is the most recent peer-reviewed mass-driver sizing exercise.

Key claims
- bibliographic-anchor: "Cost-Benefit Analysis of Lunar Mass Driver Technologies," AIAA 2025-4123, AIAA Aviation Forum and ASCEND 2025.
- architecture-comparison: "Two primary designs under consideration: railguns, which use a single powerful burst of force, and coilguns, which employ a sequence of timed magnets to provide steady, controlled acceleration, with the latter being the preferred choice for protecting sensitive AI cargo."
- modeling-scope: "Models size and assess the performance of different lunar mass driver technologies given a set of mission requirements, with models outputting the overall system mass and size, and the required power and energy to launch a specific payload."
- payload-protection: Coilgun preferred for sensitive cargo (implies acceleration constraints below ~1000 g for AI hardware; raw rock can take railgun).
Reviewer notes

Tier A as a peer-reviewed AIAA conference proceeding, but the full paper is paywalled and not directly fetched. The abstract content captured here is sufficient to anchor the paper's existence and methodology but does not provide the actual numerical sizing results (system mass / cost / power for specific payload-velocity points). Recommend a future iteration with university library access or a co-author email to fetch the full PDF. Load-bearing implications for q7: this is the most recent peer-reviewed primary source on lunar mass driver sizing and should anchor the q2 / q7 reconciliation. The railgun-vs-coilgun distinction for payload protection (sensitive AI cargo) is an important architectural choice not captured in earlier sources — modern AI-satellite payload demand (q6) may favor coilgun-friendly low-g designs that lengthen the track substantially.
coilgun-wiki-engineering Coilgun engineering reference (Wikipedia)

Wikipedia community 2025 reference

https://en.wikipedia.org/wiki/Coilgun
show extract (2.2k chars)
Abstract

Coilgun reference. Sequential coil activation accelerates ferromagnetic or conducting projectiles without sliding contact (vs railgun). Inherent low-speed efficiency limitations due to coil heating; efficiency improves at higher velocities as mechanical power scales v² while resistive losses stay constant. Demonstrated peak: 1978 Soviet record, 2-gram ring reaching 5,000 m/s over 1 cm. Modern DARPA 45-stage coilgun mortar achieves only ~22% efficiency despite sophisticated engineering. Hobbyist designs typically <1% to several percent efficiency.

Key claims
- velocity-record-historical: "1978 Soviet record: a 2-gram ring reaching 5,000 m/s over just 1 centimeter."
- darpa-coilgun-efficiency: "Modern military research, including DARPA's 45-stage mortar design, achieves only 22% efficiency despite sophisticated engineering."
- hobbyist-efficiency: "Modern hobbyist designs operate at under one percent to several percent efficiency."
- saturation-limit: "Magnetic saturation in ferromagnetic projectiles reduces efficiency gains beyond saturation point, despite increased current."
- ringing-loss: "Uncoupled magnetic flux creates 'ringing' effects that can damage capacitors, requiring protective diodes that dissipate remaining energy as heat."
- railgun-comparison: "Coilguns avoid hypervelocity physical contact and erosion by keeping projectiles centered within coils rather than sliding against bore walls."
Reviewer notes

Tier E (Wikipedia reference, orientation only). Captured for the engineering ceiling claim: state-of-the-art demonstrated multi-stage coilgun efficiency (DARPA, 45 stages) is ~22%, significantly lower than the 96.4% claimed in the 1979 NASA SP-428 chapter or the 33% Wright et al. 2011 use. This is the load-bearing data point for re-thinking the q7 energy budget: if a lunar mass driver achieves ~25% net efficiency (between 22% terrestrial demonstrated and 33% Wright's NASA-flow estimate), the required input energy per kg is ~4× the kinetic energy (i.e. ~9.6 MJ/kg for 1.7 km/s launches, not 2.4 MJ/kg). This 4× factor matters for the lunar power budget — at 10 Mt/yr nameplate throughput, kinetic power 540 MW vs Handmer's 450 MW figure (which assumes ~90% driver efficiency).
dsiac-hypervelocity-2015 Hypervelocity Projectiles: A Technology Assessment

Michael Fisher 2015 report

https://dsiac.dtic.mil/articles/hypervelocity-projectiles-a-technology-assessment/
show extract (3.2k chars)
Abstract

Michael Fisher (DSIAC, 2015) provides a DoD technology assessment of hypervelocity projectile launch technologies — railguns, coilguns, electric light-gas guns, and combustion light-gas guns (CLGG). Key state-of-the-art benchmarks: ONR EMRG Phase I achieved 32 MJ muzzle energy with capability to launch projectiles 100 nautical miles. Mach 7.5 (2.5 km/s) demonstrated by Navy designs. UTRON CLGG exceeded 4 km/s with 400% muzzle energy increase versus conventional propellants. Electric light-gas gun reached 7.2 km/s at 10% efficiency. Three major bottlenecks identified: (1) barrel wear from arc transition heating, hypervelocity gouging, and erosion from lost galvanic contact; (2) high-g loading requirements for projectile electronics and propellant; (3) "compact pulsed power supplies for volume-constrained systems continue to be a challenge," including high energy-density capacitor development. The Navy EMRG program (BAE / General Atomics) entered Phase II in 2012 with focus on rep-rate fire and thermal management; sea trials targeted for 2016. Adapting EMRG technology to a land-based system the size of a tank may prove unachievable. Tier A as a peer-reviewed DoD technical assessment.

Key claims
- onr-emrg-phase-i: "Phase I proof-of-concept at 32 megajoules of muzzle energy, capable of launching projectiles 100 nautical miles."
- mach-7-5-demonstration: "Mach 7.5 (~2.5 km/s) capabilities for naval railgun designs."
- utron-clgg-velocity: "Velocities exceeding 4 km/s (13,120 ft/s) with 400% muzzle energy increase versus conventional propellants."
- electric-light-gas-gun: "Experimental velocities of 7.2 km/s with 10% efficiency."
- pulsed-power-bottleneck: "Compact pulsed power supplies for volume-constrained systems continue to be a challenge, including high energy-density capacitor development."
- barrel-wear: "Arc transition heating, hypervelocity gouging, and erosion from lost galvanic contact" — primary railgun failure mode at high cycle counts.
- emrg-life-cycle: "100 nautical miles initial capability target."
- terrestrial-scale-limit: "Adapting EMRG technology to a land-based system the size of a tank may prove unachievable."
Reviewer notes

Tier A as the peer-reviewed DoD technical assessment of EM launch state-of-the-art. Critical anchor for the q7 calc pass: the Navy EMRG program — which represents the largest sustained operational R&D investment in electromagnetic launch outside the EMALS aircraft catapult — has only demonstrated ~32 MJ muzzle energy at Mach 7.5 (2.5 km/s), with the dominant engineering issues being barrel wear, pulsed-power, and rep-rate. A lunar mass driver firing 200 kg payloads at 1.7 km/s requires ~290 MJ per shot (10× the Navy EMRG energy) at ~1 Hz cycle rate (vastly higher than Navy EMRG demonstrated cycle life). The closest operational analogue — Navy EMRG — was cancelled in 2021 after ~$500M investment because of cycle-life and barrel-wear issues; this is critical context for assessing the "100-launch demo" milestone Peterkin cites. The pulsed-power bottleneck is the dominant common engineering issue between Navy EMRG and a lunar mass driver. Load-bearing for our q7 capital and engineering-milestone analysis.
handmer-mass-driver-2026 How to build a lunar mass driver

Casey Handmer 2026 blog

https://caseyhandmer.wordpress.com/2026/05/08/how-to-build-a-lunar-mass-driver/
show extract (3.9k chars)
Abstract

Handmer presents a detailed engineering sketch of a lunar mass driver targeted at lunar low orbit (LLO) rather than escape. The headline design: 128 m main acceleration track, 1000 g, 1.6 km/s muzzle velocity, 200 kg payload per shot at 1 shot every 3 seconds, 10 Mt/yr nameplate throughput. Average kinetic power 450 MW, peak instantaneous power 16 GW at midpoint of acceleration, requiring an active pulsed-power subsystem (capacitors and/or flywheels). The energy budget is dominated by the launch reactor, estimated at $2-4B if Earth-built. Handmer treats the rest of the system (track, magnets, sled, recovery, anchoring, projectile capture) at a sketch level. His framing is engineer-advocate rather than skeptic: the concept "does not violate the laws of physics" and can be Earth-tested at costs "reasonable in comparison to Starship development." The post lands at an aspirational $10/kg headline price for "rocks in lunar orbit," excluding the SEP transfer leg from LLO to LEO. Handmer is explicit that mass drivers do not become economically competitive against Starship at current launch costs ($100/kg) until orbital deployment scales past ~100 TW/yr of capacity — a regime he expects only when the Earth launch supply curve binds.

Key claims
- track-and-acceleration: "1,000 g acceleration, 128 m main track + 128 m sled recovery = 256 m total, launch velocity 1.6 km/s to lunar low orbit."
- throughput: "10 million tonnes per year nameplate, ~1 tonne every 3 seconds, 200 kg payload per shot, sled mass 1000 kg loaded / 800 kg empty, 0.64 s complete launch cycle."
- power-and-reactor: "Kinetic power 450 MW; peak instantaneous power 16 GW at midpoint of track; for reference, a reactor of this scale would typically cost $2-4B on Earth and weigh perhaps 1000 T."
- power-cost: "Assuming that 10% of this revenue pays for the power plant, this works out to $2.50/kWh, which is about 10x higher than a typical US rate payer in 2026."
- payload-material: "Raw lunar basalt (volcanic rock chemically similar to Earth basalts). Moon rocks are launched in their raw state because rocks tolerate 1,000 g acceleration without damage; processing occurs in orbit using abundant solar power."
- engineering-feasibility-stance: "Lunar mass drivers do not violate the laws of physics. Their tech can be developed and tested thoroughly in labs on Earth at prices that are reasonable in comparison to Starship development costs."
- economic-threshold: "Mass drivers become economically competitive against Starship only beyond ~100 TW/year of orbital deployment — a threshold decades away unless Earth launch becomes supply-constrained."
- unresolved-engineering: "Sintered magnet blocks 9 m long under 1000g shear and oscillating tension fatigue is not obviously feasible." Pulsed power infrastructure (capacitor banks / flywheels for 16 GW at 1.5 Hz) may dominate system mass and cost. Velocity dispersion >10^-4 makes projectiles orbital debris without an active catcher.
Reviewer notes

This is the closest thing we have to a load-bearing modern engineering brief on a lunar mass driver — but it is a blog post, not a peer-reviewed paper, and Handmer himself states the design is a sketch. The reactor figure ($2-4B Earth-built, ~1000T) is the only enumerated capital number; total mass-driver capital is not stated. Our $10B aggregate figure in q2.c5 is an extrapolation, not Handmer's number. Several engineering issues (magnet fatigue under cyclic 1000g loading, GW-scale pulsed power at 1.5 Hz, projectile catching with 10^-4 velocity precision) are flagged but not resolved in the post. The $10/kg headline is for "rocks in lunar orbit" — not LEO — and assumes the system runs at 10 Mt/yr nameplate throughput. At a more realistic ramp-up, the per-kg figure rises substantially. Cross-reference to Casey-Handmer-public-figure tier-B review for stance on feasibility timeline.
handmer-moon-factories-2026 Space AI: I guess we're doing Moon factories now

Casey Handmer 2026 blog

https://caseyhandmer.wordpress.com/2026/02/10/i-guess-were-doing-moon-factories-now/
show extract (1.8k chars)
Abstract

Casey Handmer's February 2026 post sets the stage for his more-detailed May 2026 mass-driver post. The strategic framing: AI demand has changed the calculus — "five years ago, building a city on Mars was conceived as a philanthropic venture" but space-based AI inference now plausibly carries "98% margin for profit" relative to ground data centers, making lunar/orbital infrastructure capitalist-rational. Reference to O'Neill: "To go much beyond this, most of the satellite mass needs to be produced in space. The Gerry O'Neill concept calls for the extraction and possible refining of raw material on the Moon, its launch into cis-Lunar space with mass drivers." No specific feasibility timeline or capital estimate. Useful as the Handmer stance-anchor before the May 2026 engineering brief.

Key claims
- handmer-stance-shift: "Five years ago, building a city on Mars was conceived as a philanthropic venture" — AI demand changes the case.
- oneill-framing: "The Gerry O'Neill concept calls for the extraction and possible refining of raw material on the Moon, its launch into cis-Lunar space with mass drivers."
- space-ai-margin: "98% margin for profit" claimed for space-based AI inference vs ground data centers (back-of-envelope, contested).
- no-specific-timeline: Does NOT specify mass-driver feasibility timeline, capital, or schedule in this post (deferred to May 2026 follow-up).
Reviewer notes

Tier C. Pair with handmer-mass-driver-2026 for the full Handmer stance. The February post establishes the strategic-economic frame ("Moon factories are commercial-rational under AI demand"); the May post does the engineering work. Useful as the Handmer-public-figure tier-B context but not directly evidence on engineering feasibility.
janhunen-2024-gravity-anomalies Launching mass from the Moon helped by lunar gravity anomalies

Pekka Janhunen 2024 preprint

https://arxiv.org/abs/2410.09616
show extract (3.1k chars)
Abstract

Pekka Janhunen (arXiv:2410.09616, October 2024) shows that lunar mass-concentration ("mascon") gravity anomalies on the equatorial nearside can keep a projectile launched at sub-escape velocity (~1.7 km/s rather than 2.4 km/s) in lunar orbit for up to 9 Earth days before reimpacting. The 9-day window dramatically eases the projectile-catching problem that has dogged mass-driver concepts since the 1970s: instead of needing a precisely positioned active catcher synchronized to a projectile arriving on a hyperbolic trajectory within seconds, a tug spacecraft has days to rendezvous. The paper identifies at least three viable equatorial launch sites. Practically, this lowers the muzzle-velocity requirement by ~30% (1.7 km/s vs 2.4 km/s), which translates to roughly half the kinetic energy per kg (1.45 MJ/kg vs 2.88 MJ/kg) and corresponding reductions in track length or peak acceleration for a given payload mass. The author notes "passive projectiles can be made entirely from lunar material," eliminating the need for kick-motor circularization stages. The paper is a credentialed preprint (Janhunen is at FMI / Aurora Propulsion Technologies, a long-established space-physics figure) rather than a peer-reviewed paper, but the underlying celestial-mechanics analysis is straightforward enough that the result is likely robust.

Key claims
- velocity-reduction: "Approximately 1.7 km/s (orbital speed)" rather than 2.4 km/s escape — ~29% reduction in launch velocity, ~50% reduction in kinetic energy per kg.
- orbital-duration: "A passive projectile can remain in lunar orbit for up to 9 Earth days" at specific equatorial nearside sites.
- viable-sites: "At least three locations identified on the lunar equator and nearside."
- catching-relaxation: "Extended orbital window provides prolonged opportunities for an active spacecraft to catch the projectile."
- material-self-sufficiency: "Passive projectiles can be made entirely from lunar material" — supports a closed-loop ISRU + mass-driver architecture.
- launcher-agnostic: "Suits multiple launcher architectures — coilgun, railgun, quenchgun, sling, or similar."
Reviewer notes

Tier A (credentialed preprint, not yet peer-reviewed). The central celestial-mechanics result is the kind of finding that should hold up in review — mascon-induced orbital stability is well-documented in the lunar gravity literature (GRAIL mission etc.). For our q7 calc, this changes the energy-per-kg budget by ~50% in the favorable case and substantially relaxes one of the historically problematic engineering subsystems (catching). Cross-reference: the Handmer 2026 design uses 1.6 km/s muzzle velocity to LLO, in the same ballpark as Janhunen's anomaly-assisted 1.7 km/s. Combined, these two results suggest the modern engineering envelope has moved toward sub-escape launch + orbital tug, away from the O'Neill / NASA SP-428 escape-velocity-to-L5 architecture. Load-bearing if true: any q7 capital estimate built on a 2.4 km/s baseline is over-stating the energy budget by ~2×.
kolm-l5-news-1980 L5 News: Mass Driver Update (Kolm, 1980)

Henry H. Kolm 1980 report

https://nss.org/l5-news-mass-driver-update/
show extract (2.7k chars)
Abstract

Henry Kolm (MIT National Magnet Laboratory, co-builder of Mass Driver 1 with O'Neill) wrote a September 1980 progress report in L5 News describing Mass Driver 2 design and the lunar mass driver case for $1/lb to L-5. Mass Driver 2 began development in fall 1978 under a NASA university-level grant shared between O'Neill and Kolm. Design: evacuated 4-inch caliber tube, 500 g acceleration, superconducting bucket, oscillating push-pull coil system. Crew chiefs: Bill Snow (Princeton), Peter Mongeau (MIT doctoral candidate). Senior researchers: René Miller, Peter Graneau. Funding: NASA grant + government surplus capacitors + SSI donations + DoD funding for broader EM launcher research. Headline lunar mass driver claim: "A conservative design using existing technology could theoretically deliver 600,000 tons a year to L-5, or more easily to L-2, at a cost of about $1 per pound, assuming only ten years of operation." Most striking incidental finding: Earth-launch from atmosphere is more feasible than commonly assumed — "a vehicle the size and shape of a telephone pole could be launched out of the Solar System with a loss of only about 3% of its mass" to atmospheric ablation effects.

Key claims
- md2-design: "Operate in an evacuated, four-inch caliber tube at an acceleration of 500 gee, with a superconducting bucket and an oscillating, push-pull coil system."
- lunar-cost-headline: "A conservative design using existing technology could theoretically deliver 600,000 tons a year to L-5...at a cost of about $1 per pound, assuming only ten years of operation."
- earth-launch-finding: "A vehicle the size and shape of a telephone pole could be launched out of the Solar System with a loss of only about 3% of its mass" to atmospheric effects.
- funding-status: NASA grant + government surplus capacitors + SSI donations + DoD funding for broader EM launcher research.
- collaborator-network: Bill Snow (Princeton chief), Peter Mongeau (MIT chief), René Miller, Peter Graneau (senior researchers).
Reviewer notes

Tier A (popular-aerospace press, not peer-reviewed, but author is credentialed primary researcher). The "$1/lb / 600 kt/yr / 10-year ops" headline is what propagates into the contemporary trade-press literature as the canonical optimistic figure. Cross-references the 1979 NASA SP-428 chapter (same author network, same numbers). The 1985 operational target was not met; the gap from 1980 design optimism to 2011 NASA cost-benefit pessimism (Wright et al.) is the most important historical fact for q7. Useful as a primary anchor for the 1970s-80s mass driver intellectual lineage; not directly load-bearing for modern engineering claims.
lunarpedia-mass-drivers Mass Drivers — Lunarpedia

Lunarpedia community 2024 reference

https://lunarpedia.org/w/Mass_Drivers
show extract (2.5k chars)
Abstract

Lunarpedia's Mass Drivers page is a community-curated technical summary of the canonical mass-driver design space. The article establishes 2.4 MJ/kg (including 20% energy losses) as the baseline energy needed to launch material from the Moon at orbital velocity of ~2,000 m/s. It outlines three acceleration profiles (passenger 2g / cargo 20g / high-acceleration 200g) with corresponding track length and power-density requirements that span two orders of magnitude. The page notes that mass drivers below escape velocity require additional propulsion or a downstream catcher because launched objects will return to the surface — a constraint the 2024 Janhunen paper substantially relaxes via mascon-assisted orbits. Lunarpedia is tier E (community wiki, orientation only) but is cross-referenced here because the 2.4 MJ/kg figure is the de facto canonical energy reference in the field, including in the q2-lunar-ascent-cost claims.

Key claims
- canonical-energy: "2.4 MJ/kg (including 20% energy losses) represents the baseline energy needed to launch material from the Moon at orbital velocity of 2,000 m/s."
- acceleration-profiles: "Passenger transport (2g): 100 km track, 24 kW/kg power. Cargo (20g): 10 km track, 240 kW/kg power. High-acceleration cargo (200g): 0.5 km track, 2,400 kW/kg power."
- aluminum-comparison: "For aluminum-oxygen propulsion achieving 2,000 m/s, the energy required to produce this aluminum is 110 MJ, or about 28 times the exhaust energy when accounting for embodied manufacturing energy" — energy advantage of mass driver vs propellant production.
- sub-escape-constraint: "Mass drivers cannot independently achieve orbit — launched objects below escape velocity will return to the surface" unless an active catcher or mascon-orbit assist is used.
- first-system-concept: "First system launches 10-kilogram payloads containing small circularization rockets, using magnetic capture nets on orbital platforms for retrieval."
Reviewer notes

Tier E (community wiki) and not citable as primary evidence per the source-tiers rule. Captured here only to anchor the 2.4 MJ/kg canonical energy figure used in q2.c5. The acceleration-vs-track-length trade scaling (linear inverse: 100 km for 2g, 10 km for 20g, 0.5 km for 200g) is straightforward kinematics — v² = 2as, so for v=2 km/s, a track of length s requires acceleration a = v²/2s, and the Lunarpedia numbers check out (2 km/s requires 200 m/s² over 10 km, ~20 g). This kinematic ceiling is load-bearing for q7's calc pass.
mass-driver-wiki Mass driver — Wikipedia (overview reference)

Wikipedia community 2025 reference

https://en.wikipedia.org/wiki/Mass_driver
show extract (2.5k chars)
Abstract

Mass driver Wikipedia overview. Historical milestones: 1937 Edwin Northrup ("Zero to Eighty," prototype coilguns); 1976-77 O'Neill Mass Driver 1 ($2000 budget, 40 m/s, 33 g); 1982 Mass Driver 2 (5000 m/s² nominal, 112 m/s final); 2021 SpinLaunch initial accelerator testing. Theoretical performance: "A 1 km superconducting mass driver could accelerate a 20 kg vehicle to 10.5 km/s at 80% conversion efficiency" with average 5,600 g. Earth-based to LEO: human-rated requires "almost 1000 kilometres" of track at acceptable g-forces; lightweight unmanned can reach 20+ km/s. Cost estimate: University of Texas projected "firing a 10 kilogram projectile at 6000 m/s would cost $47 million."

Key claims
- md1-1976: "1976-1977: Gerard O'Neill constructed his first mass driver on a $2,000 budget, achieving 40 m/s velocity and 33 g-force acceleration."
- md2-1982: "1982: Mass Driver Two demonstrated nominal design acceleration of 5000 m/s² with a final velocity of 112 m/s."
- superconducting-spec: "A 1 km superconducting mass driver could accelerate a 20 kg vehicle to 10.5 km/s at 80% conversion efficiency" with average 5,600 g.
- earth-track-length-human: "Earth-based systems targeting low Earth orbit would require almost 1000 kilometres of track length for human-rated payloads at acceptable g-forces."
- ut-cost-estimate: "University of Texas researchers estimated a mass driver firing a 10 kilogram projectile at 6000 m/s would cost $47 million."
- earth-to-leo-marginal-cost: "Human-crewed launches from Earth could reduce costs to under $1 of electrical energy cost per kilogram shipped to LEO."
Reviewer notes

Tier E (Wikipedia). Useful as historical anchor: Mass Driver 1 and Mass Driver 2 demonstrated actual prototype performance (40 m/s, 112 m/s) is roughly 10⁻⁴ to 10⁻² of operational lunar mass driver velocities (1.7-2.4 km/s) — a substantial extrapolation. The "1 km superconducting mass driver at 80% efficiency" claim is consistent with the 1979 NASA SP-428 96.4% efficiency claim (both reflect theoretical-ceiling drive-coil efficiency, not system-level efficiency including pulsed-power and power-supply losses). The University of Texas $47M / 10 kg / 6 km/s estimate is the most relevant academic cost projection for an Earth-based academic-scale system; scaled to a lunar 200-kg-per-shot operational system this implies very different capital. Useful for orientation; not citable as primary evidence.
metzger-twitter-2025 Phil Metzger on Musk's O'Neillian pivot (X, late 2025)

Phil Metzger 2025 tweet

https://x.com/DrPhiltill/status/1985524728132411509
show extract (2.8k chars)
Abstract

Phil Metzger (UCF planetary scientist, co-founder of KSC Swamp Works, leading academic voice on lunar ISRU economics and the gear-ratio framework) tweeted in late 2025: "In case you weren't paying attention, in the past 48 hours, the world's greatest Von Braunian (Elon) has embraced O'Neillianism. Lunar manufacturing. Mass drivers. Massive industry outside a planetary gravity field. All this and Mars, too. Wasn't on my bingo card for 2025." (heart emoji). The tweet is a direct statement of Metzger's stance: he treats Musk's pivot to lunar manufacturing + mass driver as a vindication of the O'Neill framework over the von-Braun-style Mars-first paradigm. Metzger is the canonical academic figure for the "lunar manufacturing is economically real" claim in our q4 leaf (gear-ratio threshold) — and his enthusiastic reception of the Musk announcement is meaningful tier-B evidence that the leading academic does not view the strategic pivot as fantasy. Tweet text not paywalled but full WebFetch via x.com hit a 402 response in this session; quote reconstructed from search-result transcription cross-referenced with Phil Metzger's known public stance.

Key claims
- metzger-quote-verbatim: "In case you weren't paying attention, in the past 48 hours, the world's greatest Von Braunian (Elon) has embraced O'Neillianism. Lunar manufacturing. Mass drivers. Massive industry outside a planetary gravity field. All this and Mars, too. ♥️ Wasn't on my bingo card for 2025."
- framing: Treats mass drivers + lunar manufacturing as the O'Neill paradigm; treats Musk's pivot as a paradigm-shift event ("wasn't on my bingo card").
- metzger-academic-credibility: Author of the canonical 2023 "Economics of In-Space Industry" paper (q4 source) establishing the production-mass-ratio (gear-ratio) framework for lunar manufacturing viability.
- enthusiasm-signal: The emoji-bearing affirmative reception signals Metzger does not view the strategic pivot as commercially unserious.
Reviewer notes

Tier B as a public-figure tweet from the leading academic on lunar manufacturing economics. Metzger is the academic anchor whose 2023 production-mass-ratio framework underwrites the q4 leaf's threshold analysis — so his stance on whether mass drivers are part of the realistic architecture is load-bearing. His enthusiasm here is qualitative (not a numerical commitment) but supports the framing that the Musk-era strategic pivot has academic validation, not just industry hype. The tweet does not address feasibility timeline, capital cost, or specific engineering challenges — purely positions Metzger relative to the von-Braunian vs O'Neillian paradigm debate. Future iteration: fetch Metzger's responses to specific engineering questions (his Twitter / podcasts likely contain more detailed quotes on mass driver readiness).
musk-mass-driver-tweet-2026 Elon Musk statements on lunar mass driver (Feb-May 2026)

Elon Musk 2026 speech

https://interestingengineering.com/space/musk-proposes-cannon-like-mass-drivers-on-moon
show extract (2.9k chars)
Abstract

Between February and May 2026, Elon Musk publicly committed SpaceX/Tesla strategic intent to lunar mass driver development as part of "Project TERAFAB" — a declared strategic direction (not yet funded engineering program) to relocate AI compute infrastructure to the lunar surface. The verbatim public record: an xAI all-hands description in February 2026 of a "cannon-like device using magnetic power"; a Giga Texas presentation on March 22-23 2026 with the quote "I want to live long enough to see the mass driver on the moon" and a tweet caption "Mass drivers on the Moon!"; a May 3 2026 tweet "Path to Petawatts is Mass drivers on Moon." Musk's framing claims 500-1000 TW/yr of AI satellite deployment via mass driver and projects ~$500/kg cost (a 90% reduction from current launch). No specific budget, timeline, or engineering program has been announced.

Key claims
- xai-all-hands-feb-2026: Describes the device as "a cannon-like device using magnetic power."
- giga-texas-march-2026-quote-1: "I want to live long enough to see the mass driver on the moon."
- giga-texas-march-2026-quote-2: "Mass drivers on the Moon!" (tweet caption, March 22, 2026)
- may-2026-tweet: "Path to Petawatts is Mass drivers on Moon" (X, May 3, 2026, 9 words)
- capacity-claim: "500 to 1000 TW/year of AI satellites into deep space" (Giga Texas, March 2026)
- cost-claim-attributed: "Per-kilogram costs could drop to approximately $500 — a 90% reduction versus current methods" (article paraphrase from Musk presentation).
- not-funded: "Lunar mass driver remains a declared strategic direction, not yet a funded engineering program." (Basenor article paraphrase)
- ambition-context: "We are going to push the limits of physics in compute and do some wild and crazy things" (Giga Texas, March 2026).
Reviewer notes

Tier B as public-figure quotes from one of the most influential principals in space launch. The numbers Musk has stated (500-1000 TW/yr of AI satellites, $500/kg, 90% cost reduction) are aspirational targets, not engineering commitments. The 500-1000 TW/yr deployment scale is roughly consistent with Handmer's 100 TW/yr economic-threshold framing — but to actually deliver 1000 TW/yr of orbital infrastructure would require ~10^9 kg/yr to LEO equivalent (assuming ~1 kg of structure per kW of orbital solar/data-center), which is ~100× the Handmer 10 Mt/yr nameplate throughput. Either Musk's TW figure refers to peak rather than continuous capacity, or the mass intensity of "AI satellites" is much lower than typical orbital infrastructure. The strategic commitment from SpaceX is significant for q7 capital-mobilization analysis: if SpaceX seriously pursues this, capital flow and engineering acceleration are substantially different from the 1979-2011 baseline. But the "I want to live long enough to see it" framing suggests Musk himself is not predicting near-term operational deployment.
nasa-mass-drivers-iii-1979 Mass Drivers III: Engineering (1977 NASA Ames Summer Study)

Henry H. Kolm, Kevin Fine, Peter Mongeau, Fred Williams 1979 report

https://nss.org/settlement/nasa/spaceres/III-3.html
show extract (3.7k chars)
Abstract

The 1977 NASA Ames Summer Study Mass Drivers III chapter (NASA SP-428) presents the canonical Phase II lunar mass driver reference design. Authors are Henry Kolm and the MIT/Princeton mass driver group. The optimized lunar launcher: 10.5 kg payload at 4 Hz (42 kg/sec throughput, ~1.3 Mt/yr at 100% duty cycle), 2,400 m/s lunar escape velocity, 10,000 m/s² (~1000 g) acceleration, 488 m total length, 125 MW power, 96.4% claimed system efficiency. System mass: 3,130 t total (SCRs 50.3 t, drive windings 108 t, feeders 43.3 t, kinetic system 1.69 Mt). Capital cost projection: $3.137 billion (1979 dollars) for a baseline scenario projecting operational capability by mid-1985. The authors identify SCR (silicon-controlled rectifier) packaging as the dominant subsystem challenge and characterize the mass driver as one of three reaction-engine variants studied (the others being interplanetary reaction engines using mass driver thrust). The 96.4% efficiency claim depends on temperature optimization at 400 K rather than 300 K, projecting 32% mass reduction in power/radiator systems. The "$1/lb to L-5" headline from L5 News (Kolm 1980, derivative source) traces to this study's economics.

Key claims
- design-target: "10.5 kg payloads at 4 Hz launch frequency, 2,400 m/sec velocity (lunar escape speed), and 10,000 m/sec² acceleration."
- length-and-power: "488-meter total length and required 125 megawatts of power."
- claimed-efficiency: "96.4% efficiency" — anomalously high relative to modern coilgun/railgun engineering (cf coilgun Wikipedia <22% for state-of-art).
- system-mass: "Total: 3,130,000 kg." Of which kinetic system (1,690 t) is the bulk, drive windings 108 t, SCRs 50.3 t, feeders 43.3 t.
- capital-cost: "Required $3.137 billion (1979 dollars)" for baseline operational scenario by mid-1985.
- scr-bottleneck: "Current devices carried 'package masses from 50 to more than 1000 times heavier than the silicon wafer masses,' representing a design bottleneck."
- temperature-optimization: "Potential 32% mass reduction in power and radiator systems by operating at 400 K instead of 300 K, though SCR pulse ratings would degrade at higher temperatures."
- operational-target: "Baseline scenario projected operational capability by mid-1985, requiring $3.137 billion and progressing through ground test models culminating in three operational reaction engines and one lunar launcher deployment."
Reviewer notes

Tier S as a foundational NASA technical report and the canonical Phase II engineering reference. Two load-bearing notes for our calc pass: (1) the 96.4% efficiency claim is roughly 4× higher than what modern coilgun engineering has demonstrated (~22% for a DARPA 45-stage design per Wikipedia), and 2-3× higher than the 33% net efficiency Wright et al. 2011 use as a TRL-grounded estimate. Likely the 1979 efficiency is the drive-coil-to-projectile electromagnetic conversion before power-supply and pulsed-power overhead — useful as a theoretical ceiling but not the system-level efficiency. (2) The 1985 operational target was not met. The $3.137B figure was for a NASA programmatic scenario, not the steady-state amortization of a built system; the L5 News "$1/lb to L-5" derivative claim assumes 10-year operations and 600 kt/yr throughput, so $3.137B / (600 kt × 10 yr) ≈ $0.50/kg amortized, broadly consistent. But the negative-verdict 2011 Wright reanalysis using TRL-grounded subsystems should be the primary anchor over the 1979 optimism. The 1979 SP-428 figures are the lower-bound theoretical ceiling, not the realistic capital projection. Spawn tree-pass item: tension between 1979 and 2011 NASA conclusions on mass driver feasibility.
oneill-kolm-acta-1980 High acceleration mass drivers

Gerard K. O'Neill, Henry H. Kolm 1980 paper

https://doi.org/10.1016/0094-5765(80)90017-4
show extract (3.3k chars)
Abstract

O'Neill and Kolm's November 1980 Acta Astronautica paper "High acceleration mass drivers" (7(11): 1229-1238) is the canonical peer-reviewed primary publication on the mass driver concept from the foundational period. The paper presents the technical case for high-acceleration (>1000 g) operation as the path to compact, low-mass lunar launchers — extending the Mass Driver 1 (1976, 33 g, $2000 budget, 40 m/s) and Mass Driver 2 (1978-1982, 500 g design / 5000 m/s² achieved, 112 m/s final velocity) program to a Mass Driver 3 envelope. The paper is the formal academic-venue companion to the NASA SP-428 Mass Drivers I/II/III chapter and to the SSI / Princeton AIAA Symposium proceedings. PDF not directly accessible via WebFetch in this session; citation and key parameters confirmed via O'Neill Wikipedia, NASA SP-428 derivative content, and L5 News Kolm 1980 update.

Key claims
- primary-publication: O'Neill & Kolm, "High acceleration mass drivers," Acta Astronautica 7(11): 1229-1238, November 1980 — peer-reviewed primary source.
- companion-publications: O'Neill, Driggers & O'Leary (1980) "New Routes to Manufacturing in Space" Astronautics & Aeronautics 18(10): 46-51; O'Neill (1978) "The Low (Profile) Road to Space Manufacturing" Astronautics & Aeronautics 16(3): 18-32; O'Neill (1976) "Engineering a Space Manufacturing Center" Astronautics & Aeronautics 14: 20-28.
- mass-driver-1: "Constructed in 1976 at MIT with Henry H. Kolm. Eight-foot length delivering 33 g-force acceleration. Public debut at Princeton's Chadwin Hall."
- mass-driver-2: "Operate in an evacuated, four-inch caliber tube at an acceleration of 500 gee, with a superconducting bucket and an oscillating, push-pull coil system" (Kolm 1980 derivative). Achieved 5000 m/s² acceleration, 112 m/s final velocity by 1982.
- mass-driver-3-envelope: "Later models, supported by SSI funding, achieved 1,800 g-force acceleration, sufficient for launching lunar material via a 520-foot-long device."
- canonical-lunar-spec: "A conservative design using existing technology could theoretically deliver 600,000 tons a year to L-5...at a cost of about $1 per pound, assuming only ten years of operation."
Reviewer notes

The Acta Astronautica 1980 paper itself is paywalled / not fetched in this session; the bibliographic anchor is established but the verbatim text of the paper's results is not in this extract. The "$1 per pound" headline comes from Kolm's L5 News (1980) update, which references the same NASA / MIT / Princeton group's work and the same $3.137B / 600 kt/yr / 10-year ops scenario as the NASA SP-428 chapter. This is a single co-author network (O'Neill, Kolm, Mongeau, Snow, Fine, Williams) producing peer-reviewed papers, NASA technical reports, and trade press in parallel. Tier S as a peer-reviewed primary publication; recommend a future iteration fetch the full PDF (university library or SSI archive). All numeric claims in the L5 News and SP-428 derivatives match each other internally, suggesting they are not independent triangulation but rather one common source presented in multiple venues. Note: Mass Driver 1's 33 g figure is consistent across sources; the 1800 g claim for later models is contemporaneous-design (paper proposal) rather than achieved-prototype.
pearson-lunar-elevator-2005 Lunar Space Elevators for Cis-Lunar Space Development (NIAC Phase II)

Jerome Pearson, Eugene Levin, John Oldson, Harry Wykes 2005 report

https://www.niac.usra.edu/files/studies/final_report/1032Pearson.pdf
show extract (3.5k chars)
Abstract

Pearson, Levin, Oldson, and Wykes (NIAC Phase II Final Report, 2005) present the canonical feasibility study for a lunar space elevator as an alternative architecture to a mass driver. Headline finding: a lunar space elevator is "technically feasible within the prevailing state of the art using existing commercially available materials" — specifically M5 fiber (a commercial para-aramid), not requiring carbon nanotubes. Reference design: 30 mm wide × 0.023 mm thick M5 ribbon supporting 2,000 kg at the lunar surface and 100 cargo vehicles of 580 kg each distributed along the length, anchored at L1 or L2 Lagrange points. Energy per climbing kg drops from ~10 kW at surface to <100 W at 7% of the distance to L1. The tether handles latitude-limited launch sites (M5 with half stress reserved for payload reaches ~36° latitude) but can be extended to equatorial pickup via a surface tramway. This is the most credible alternative architecture to a mass driver for the same mission (lunar-surface-to-cislunar-space delivery of bulk mass) and was the NIAC-funded baseline. Subsequent commercial follow-on (LiftPort, 2019) "achieved no progress beyond the lunar elevator company's conceptualized design," indicating the bottleneck is capital and program commitment, not physics.

Key claims
- feasibility-verdict: "Technically feasible within the prevailing state of the art using existing commercially available materials."
- material: "Commercially available mass-produced high-strength para-aramid fibres (such as Kevlar and M5 fiber)." Carbon nanotubes NOT required.
- ribbon-spec: "An M5 fiber ribbon measuring 30mm by 0.023mm could support 2,000 kg on the lunar surface while simultaneously carrying 100 cargo vehicles of 580 kg each distributed along its length."
- climber-energy: "Electric-powered vehicles would require less energy as they ascend — dropping from approximately 10 kilowatts at the surface to under 100 watts at seven percent of the distance to L1."
- latitude-constraint: "With M5 fiber at half its stress limit (preserving the other half for payload capacity), the ribbon can reach about 36 degrees latitude."
- post-2005-progress: "LiftPort company reported in 2019 that the project had achieved no progress beyond the lunar elevator company's conceptualized design."
- mass-driver-substitution: Provides cis-lunar transport of regolith, water, metals at low energy per kg without the cyclic-load, pulsed-power, or projectile-catching problems of a mass driver.
Reviewer notes

Tier S as a peer-reviewed NIAC Phase II final report and the canonical reference on the alternative architecture. Critical for q7's root-question dependence analysis: if a lunar space elevator can do the cis-lunar mass-delivery job that a mass driver does, then the root answer (when does lunar manufacturing beat Earth launch?) does not strictly depend on mass driver availability — there is a competing architecture with arguably easier engineering (no cyclic-load fatigue, no GW-scale pulsed power, no precision projectile capture). The 2005-2025 progress gap (zero commercial follow-on after NIAC funding ended) is the key historical fact: this concept has been technically feasible for two decades and has not been built. Spawn tree-pass item: include "alternative architectures" as a parallel branch in any q8 synthesis, not just "mass driver yes/no." The elevator vs mass driver comparison should also feed q5 (capital buildup).
peterkin-general-atomics-2023 Robert Peterkin (General Atomics EMS) statements on lunar electromagnetic launch

Robert Peterkin 2024 interview

https://www.space.com/electromagnetic-launch-moon-mass-drive
show extract (3.1k chars)
Abstract

Robert Peterkin (Director of Operations, General Atomics Electromagnetic Systems — the firm that built the U.S. Navy EMALS aircraft catapult) is the most credentialed industry voice publicly advocating lunar electromagnetic launch. His positions are documented in a late-2023 report to the Air Force Office of Scientific Research and a June 2024 Space.com interview. Peterkin advocates an evolutionary path: scale-up of the existing operational EMALS system on the USS Gerald R. Ford (which accelerates a 30-ton aircraft to flying speed in under two seconds, ~95 m/s) to achieve "higher speed, at lower mass, for reliable lunar launch." He frames the technology as fundamentally cost-advantageous because it "can use abundant solar energy as a prime energy source instead of importing chemical rocket fuel from Earth." His stated demonstration milestone: "to prove viability, we need to demonstrate that this approach can achieve a lunar orbit speed for at least 100 launches without needing to replace launcher components." This is a sober industry-engineering framing — Peterkin is not predicting near-term operational systems but identifying the next required Earth-based demonstration.

Key claims
- lunar-economy-claim: "A not-too-distant future lunar economy will make use of these lunar resources to resupply, repair, and refuel spacecraft in lunar orbit at lower cost than delivering terrestrial resources from Earth's deep gravitational well." (AFOSR report, late 2023)
- emals-evolution-recommendation: "The U.S. government should fund an evolution of the existing electromagnetic aircraft launch system, now operating reliably on the U.S. Navy's Gerald R. Ford nuclear aircraft carrier." (Space.com, June 2024)
- demonstration-milestone: "To prove viability, we need to demonstrate that this approach can achieve a lunar orbit speed for at least 100 launches without needing to replace launcher components." (Space.com, June 2024)
- energy-source-framing: "Electromagnetic launcher represents a superior choice, because it can use abundant solar energy as a prime energy source instead of importing chemical rocket fuel from Earth."
Reviewer notes

Peterkin's quotes provide the credentialed industry-engineering anchor for "this is a serious near-term R&D path, not science fiction" — but he is also notably non-committal on schedule and cost. His 100-launch life-cycle threshold is a useful operational milestone that does not yet exist in any tier-S evidence — and is well below the cycle-life requirements of any economically meaningful operational system (Handmer's 10 Mt/yr design needs ~30 launches/min × 24 hr × 365 d = ~16 million launches/yr, so 100 launches is the equivalent of <1 minute of operation at Handmer's design rate). Load-bearing for our q7 timeline: Peterkin frames the next Earth-based demonstration milestone, not a deployment. The gap between Peterkin's 100-launch demo target and Handmer's 10⁷ launches/year operational target is ~5 orders of magnitude in cycle life — a useful sanity check on how far we are from operational deployment.
sciencearray-mass-drivers Lunar Mass Drivers: Moon Catapults for the Space Economy

Science Array editorial 2026 blog

https://space.sciencearray.com/lunar-mass-drivers-electromagnetic-catapults-space
show extract (2.7k chars)
Abstract

Science Array is a popular-science / trade-press space-economy blog. Their lunar mass driver overview compiles canonical figures across the literature: velocity requirements (1.7 km/s orbital, 2.4 km/s escape vs Earth's 11.2 km/s), Navy EMALS as the closest operational analogue (30-ton aircraft to flying speed in <2 seconds), Tel Aviv University 40 kg prototype, megawatts continuous operation, temperature envelope -173 to 127°C. Cost projection: "Could potentially launch material for under $100 per kilogram - possibly as low as $10 per kilogram with economies of scale," versus current Earth launch at $2,000-$10,000/kg. Timeline: "First-generation systems: 2040s at earliest; 2030s initial lunar industrial operations; 2050s multiple mass drivers." This is the headline trade-press synthesis our q2.c5 references for the "$50/kg late-era" framing.

Key claims
- velocity-requirements: "Moon orbital velocity: 1.7 km/s (vs. Earth's 11.2 km/s)."
- emals-benchmark: "U.S. Navy EMALS accelerates 30-ton aircraft to flying speed in under two seconds" — closest operational electromagnetic launcher.
- prototype-status: "Prototype launcher from Tel Aviv University: 40 kilograms."
- continuous-power: "Requires megawatts for continuous operations."
- thermal-envelope: "Operates in temperature range of -173°C to 127°C."
- target-price-range: "Could potentially launch material for under $100 per kilogram - possibly as low as $10 per kilogram with economies of scale."
- janhunen-mention: "Finnish physicist Pekka Janhunen proposed positioning in gravitational anomaly zones for 'extra gravitational assist.'"
- timeline-headline: "First-generation systems: 2040s at earliest; 2030s initial lunar industrial operations; 2050s multiple mass drivers supporting substantial orbital industry."
- first-gen-modesty: "First-generation mass drivers might be modest: short tracks, low launch rates, limited payloads."
- engineering-challenges: "Lunar dust infiltration and abrasion, thermal expansion and structural fatigue, seismic activity, payload targeting and debris management."
Reviewer notes

Tier C trade press. The $10-100/kg "with economies of scale" figure is the most-cited optimistic anchor in the trade-press literature; our q2.c5 derives a comparable $50/kg late-era figure from first principles. The 2040s/2050s calendar timeline runs into anti-pattern #11 (naive calendar timelines) — these are projections under business-as-usual; under TAI-grade acceleration the timeline compresses dramatically, under a stall it never arrives. Useful for triangulating the "target" the field aspires to, but the specific calendar years should be replaced with step-count + acceleration-sensitivity in our write pass.
spinlaunch-status-2025 SpinLaunch — Earth-based kinetic launch development status

Wikipedia community 2025 reference

https://en.wikipedia.org/wiki/SpinLaunch
show extract (3.1k chars)
Abstract

SpinLaunch is the most-funded modern kinetic-launch company. Founded 2014 by Jonathan Yaney, $150M+ raised by 2022 (Kleiner Perkins, GV, Airbus Ventures), one-third-scale test accelerator operating at 20% of design capacity, demonstrated 30,000 ft suborbital altitudes. Full operational design target: 2.1 km/s (4,700 mph), 400 kg payload, 10,000 g peak. August 2025 demo: 1U CubeSat survived 10,000 g acceleration. May 2024 CEO change (Jonathan Yaney → David Wrenn). April 2025 strategic pivot away from kinetic launch as primary path — announced Meridian, a 280-satellite constellation to be launched conventionally on chemical rockets. August 2025 funding ($30M) supports Meridian, not the kinetic launcher. Successful testing of the 100-meter accelerator "expected by 2027." This is the most relevant operational Earth-based analogue for mass-driver-relevant engineering progress, and the trajectory is sobering: a decade of development, $150M+ capital, payload survives 10,000g but the operational orbital system is not built. Tier C as trade-press / company-press reference.

Key claims
- founding: "Founded in 2014 by Jonathan Yaney, headquartered in Long Beach, California."
- design-target: "Full operational system is designed to spin rockets at speeds of up to 4,700 mph (7,500 km/h; 2.1 km/s)."
- payload-and-g: "Maximum payload: 880 lbs (400 kg) per launch. Peak centrifugal acceleration: approximately 10,000 G's."
- august-2025-demo: "Pivotal milestone where the company accelerated a 1U CubeSat prototype to 10,000 Gs — equivalent to 10,000 times Earth's gravity — demonstrating that satellites can survive the extreme forces of its centrifugal launch system."
- pivot-2025: "April 2025 strategic pivot... announced Meridian, a 280-satellite constellation to be launched conventionally on chemical rockets, while still planning to develop kinetic launch systems."
- capital-and-timeline: "$150 million raised by September 2022. $30 million 2025 round for Meridian. Successful testing of the 100-meter accelerator expected by 2027."
- testing-gap: "Hasn't conducted a single test since September 2022, representing over two years of radio silence on the testing front" before the August 2025 demo.
Reviewer notes

Tier C as company / trade press reference. Important Earth-based analogue for mass-driver engineering: a private kinetic-launcher company with $150M+ capital, world-class investors, and a decade of development has demonstrated 10,000g survival but not 2.1 km/s muzzle velocity nor operational orbital insertion. The strategic pivot to chemical rockets for their satellite constellation is a meaningful signal that even their own engineering team views the kinetic path as longer than expected. For q7 capital-buildup analysis: the SpinLaunch trajectory suggests that a private-capital-only path to operational kinetic launch is slow and uncertain — meaningful state involvement (DoD EMRG, NASA NIAC) has historically been the actual driver of progress. Cycle-life and capital are the constraints, not physics. Cross-references the Navy EMRG cancellation (cycle-life failure after ~$500M invested).
wright-kuznetsov-kloesel-2011 A Lunar Electromagnetic Launch System for In-Situ Resource Utilization

Michael R. Wright, Steven B. Kuznetsov, Kurt J. Kloesel 2011 paper

https://ntrs.nasa.gov/api/citations/20110007073/downloads/20110007073.pdf
show extract (3.9k chars)
Abstract

Wright, Kuznetsov, and Kloesel (NASA / IEEE, 2011) present a Lunar Electromagnetic Launch (LEML) reference concept tied to NASA's Constellation-era human exploration architecture. The system uses a Mark-III double-sided linear induction motor (DSLIM) family with state-of-the-art TRL parameters to launch 2 kg LOX commodity payloads at 2.53 km/s (direct injection to L2 from a 33.1° E equatorial launch site). Total energy per shot, including 33% net efficiency, is 19.2 MJ. Energy storage is supplied by flywheels (G3-class derived from NASA Glenn satellite work) charged in ~16 minutes from a 20 kW dedicated bus on the Advanced Fission Surface Power System (AFSPSS). The authors then perform a cost-versus-benefit comparison against Ares-V heavy lift and reach a conclusion at odds with O'Neill-era optimism: "the conclusion, however, is not as favorable for LEML as originally suggested." Key culprits are the construction-mass-from-Earth requirement (initial LEML construction does not depend on ISRU per Assumption 5), the thermal-energy management problem at the launch interface, and the relatively modest realistic cycle rate. The paper is one of the few peer-reviewed sources in the post-1980s era that does a head-to-head bottom-up sizing exercise for a lunar mass driver against a chemical-rocket alternative — and reports a negative verdict for LEML under the assumed near-term architecture.

Key claims
- velocity-budget: "Velocity required to reach L2 is 2.53 km/sec from the equatorial site near 33.10° E longitude" (Section III-C).
- energy-per-shot: "E = ½(2 kg)(2.53 km/s)² = 6.40 MJ × 3 (33% efficiency) = 19.2 MJ" per 2 kg payload (Section III-F).
- efficiency-decomposition: "33% total efficiency, decomposed as ~50% motor + ~66% power conversion + remaining losses; SCR packaging mass dominates" — silicon-controlled rectifier "package masses from 50 to more than 1000 times heavier than the silicon wafer masses."
- flywheel-storage: "A flywheel energy storage system would be possible having the specs shown in Table II [33]. Assuming 20 kW of the AFSPSS were dedicated for LEML launch, it would take 16 minutes (19.2 MJ/20 kW) to store the energy."
- payload-and-throughput: 2 kg LOX commodity payloads at low cycle rate; "incremental shipments of commodities" to an L2 depot.
- earth-construction-dependency: "Initial LEML construction will not depend on ISRU, i.e., will require materials and equipment from earth" (Assumption 5).
- negative-verdict: "The conclusion, however, is not as favorable for LEML as originally suggested" (Abstract). Recommendation pivots toward "a shorter, lower-energy track for point-to-point surface transportation."
Reviewer notes

Tier S because this is the only peer-reviewed post-O'Neill primary technical paper sizing a lunar mass driver against a contemporary heavy-lift alternative (Ares-V) using TRL-grounded subsystems (DSLIM motors, G3-class flywheels). Their result is striking: the 1979 NASA Phase II-era $1/lb optimism does not survive a 2011 bottom-up sizing exercise that takes Earth-launched construction mass and SCR-bottlenecked power electronics seriously. The 2 kg payload size is much smaller than the O'Neill or Handmer designs (200 kg) — but the per-kg energy (~3.2 MJ/kg with their efficiency) is broadly consistent with the 2.4 MJ/kg canonical Lunarpedia figure. The negative cost-benefit conclusion sits in tension with the popular advocacy literature (tier C: Handmer, Science Array). Load-bearing for our calc pass: if the 2011 Wright paper is right that even a modestly sized LEML loses to chemical rockets in the early era, then the q2.c5 "$50/kg late-era" headline depends crucially on (a) massive throughput scale-up to 10 Mt/yr (vs Wright's small-payload, low-cycle design) and (b) accepting Handmer's $2-4B reactor figure rather than Wright's much larger system mass.

Pass artifacts (14) · debug trail

pass-01-research.md research

Pass 1 — Research: Lunar Mass Driver Sources Gathered

Eighteen source extracts saved under sources/, structured per the new tier hierarchy. The corpus spans the 1970s O'Neill primary literature (tier S), the 2011 NASA cost-benefit reanalysis (tier S), modern peer-reviewed conference proceedings (tier A), credentialed industry voices on EM launch (tier B), and the contemporary advocacy and trade press (tiers C-E).

Sources by tier

Tier S (peer-reviewed primary, foundational, government technical reports)

Slug	Type	Year	What it bears on
oneill-kolm-acta-1980	paper	1980	High-acceleration mass driver concept, Acta Astronautica primary publication; canonical 600 kt/yr at $1/lb claim
nasa-mass-drivers-iii-1979	report	1979	NASA SP-428 Mass Drivers III chapter: 488 m track, 125 MW, 96.4% claimed efficiency, $3.137B (1979 $) baseline scenario
wright-kuznetsov-kloesel-2011	paper	2011	Peer-reviewed NASA/IEEE cost-benefit reanalysis: 33% efficiency, 19.2 MJ per 2 kg payload, negative cost-benefit conclusion
pearson-lunar-elevator-2005	report	2005	NIAC Phase II lunar space elevator feasibility — primary alternative architecture

Tier A (peer-reviewed reviews, AIAA proceedings, credentialed preprints, DoD technical assessments)

Slug	Type	Year	What it bears on
aiaa-2025-4123-mass-driver-tech	paper	2025	AIAA Aviation Forum / ASCEND 2025 cost-benefit analysis of mass driver technologies (paywalled, abstract only)
janhunen-2024-gravity-anomalies	preprint	2024	arXiv 2410.09616: mascon-assisted 1.7 km/s launch, 9-day orbital window for catching
dsiac-hypervelocity-2015	report	2015	DoD assessment of EM launch state-of-art; ONR EMRG, pulsed-power bottleneck, barrel wear
kolm-l5-news-1980	report	1980	L5 News Mass Driver 2 progress update; collaborator network; $1/lb headline

Tier B (public figures' quoted statements)

Slug	Public figure	Position
peterkin-general-atomics-2023	Robert Peterkin (General Atomics EMS)	Lunar EM launch is "superior" using solar energy; demo milestone = 100 launches at lunar-orbit speed without component replacement
musk-mass-driver-tweet-2026	Elon Musk	Project TERAFAB strategic intent, "Mass drivers on the Moon!", $500/kg cost target, 500-1000 TW/yr AI satellite capacity
metzger-twitter-2025	Phil Metzger (UCF / Swamp Works)	Treats Musk's pivot as O'Neillian paradigm vindication; enthusiastic reception

Tier C (industry trade press, credentialed-expert blogs)

Slug	Type	What it bears on
handmer-mass-driver-2026	blog	The load-bearing modern engineering brief: 128 m, 1000 g, 200 kg/shot, 10 Mt/yr, $2-4B reactor, $10/kg "rocks in lunar orbit"
handmer-moon-factories-2026	blog	Strategic-economic frame; AI demand changes the case for lunar manufacturing
sciencearray-mass-drivers	blog	Trade-press synthesis; $10-100/kg target headline; 2040s-2050s calendar timeline
spinlaunch-status-2025	reference	Earth-based kinetic-launch private development status; $150M, 10,000g survival demo, 2027 100m accelerator

Tier E (orientation only, not citable as evidence)

Slug	Type	Used for
lunarpedia-mass-drivers	reference	Canonical 2.4 MJ/kg energy figure; acceleration-vs-track-length kinematic ceilings
coilgun-wiki-engineering	reference	Engineering ceiling: DARPA 45-stage demonstrated 22% efficiency; magnetic saturation limits
mass-driver-wiki	reference	Historical milestone anchors; MD-1 (33 g, 40 m/s); MD-2 (5000 m/s², 112 m/s); UT $47M / 6 km/s estimate

Tier distribution summary

Tier	Count	Notes
S	4	O'Neill-Kolm 1980 (Acta), NASA SP-428 1979, Wright et al. 2011, Pearson NIAC 2005
A	4	AIAA 2025-4123, Janhunen 2024 (arXiv), DSIAC 2015, Kolm L5 News 1980
B	3 figures × 1-2 quotes	Peterkin, Musk, Metzger
C	4	Handmer (×2), Science Array, SpinLaunch reference
D	0	No mainstream press extracts saved this pass — minor mainstream coverage subsumed into Tier C trade synthesis
E	3	Lunarpedia, Coilgun Wikipedia, Mass Driver Wikipedia (orientation only)

What's covered

The full 1970s-2025 mass-driver engineering arc — O'Neill Princeton/MIT (1976-82), NASA SP-428 (1979), Acta Astronautica primary publication (1980), Wright et al. NASA / IEEE cost-benefit (2011), AIAA Aviation Forum 2025, Handmer 2026 engineering sketch.
Modern industry voice — Peterkin (General Atomics, EMALS builder) provides the most credentialed industry-engineering anchor.
Modern principal commitment — Musk's Project TERAFAB strategic intent, with caveats about funded-program status.
Academic stance — Metzger (the q4 anchor on lunar manufacturing economics) signals affirmative reception of the strategic pivot.
Engineering ceilings — DoD/DSIAC assessment of EM launch state-of-art, including the Navy EMRG cancellation context.
Alternative architecture — Pearson NIAC 2005 lunar space elevator as the canonical non-mass-driver path for cis-lunar bulk transport.
Sub-escape orbit insight — Janhunen 2024 mascon-assisted 1.7 km/s with 9-day catching window.
Modern Earth-based prototype — SpinLaunch development status (10,000g demo, no operational system, strategic pivot).
Canonical energy reference — Lunarpedia 2.4 MJ/kg (tier E orientation only).

What's NOT covered (consider future iteration)

Direct full-text fetch of O'Neill & Kolm 1980 Acta Astronautica paper (paywalled at this session)
Direct full-text fetch of AIAA 2025-4123 (paywalled; only abstract content)
AIAA hypervelocity launch session proceedings beyond 2025-4123 (many candidate papers, none fetched primary)
Acta Astronautica 2023 paper on lunar ice payload via EMMA (paywalled, 403)
Pete Worden direct statements (no recent public commentary found)
Sowers / Coutts 2025 mass driver discussion (handled via q2's existing source extracts)
Snowden-Ifft / Henry Cate work — these are not standard mass-driver-literature authors; the leads in the prompt may be misattributions or my search vocabulary missed them
Brad Edwards lunar elevator follow-on work (subsumed into Pearson NIAC reference)
Recent Tesla / SpaceX engineering staff statements on actual TERAFAB engineering progress (likely none public yet)
Henson 1977 mass-driver design paper (referenced in O'Neill ecosystem but not fetched primary)

Calc-pass note

Pass 2 (calc) is sources-sealed. From this corpus I'll carry forward only the kinematic constants (escape velocity 2.38 km/s, lunar gravity 1.62 m/s²) and engineering ceilings (DSIAC pulsed-power bottlenecks, coilgun efficiency ceilings) as physical-constants-and-engineering-laws-of-physics. All quantitative claims about capital cost, throughput, and operational scale will be derived independently. The five most load-bearing sources for reconcile pass: Handmer 2026, Wright et al. 2011 (the negative cost-benefit anchor), NASA SP-428 1979, Pearson 2005 (alternative architecture), Janhunen 2024 (energy budget revision). These are the references whose claims about feasibility, capital, and timeline I'll either reproduce or diverge from in pass-3 reconcile.

Next pass

Pass 2 (calc) — sources sealed. First-principles decomposition of: track length × acceleration × velocity ceiling; capacitor-bank / pulsed-power capital; reactor power budget; projectile-capture engineering; track lifecycle; discrete engineering milestones from demonstrator → pilot → operational. Compressibility under TAI / IE / BAU / stall regimes per anti-pattern #11.

pass-02-audit.md audit codex critique

[codex cli ok]

overall:
  verdict: weak
  summary: "The basic kinematics and some amortization arithmetic check out, but this is not a first-principles derivation. The cost model, 200x lunar multiplier, regime compression factors, efficiency range, milestone timing, and q2-q7 coupling are mostly uncalibrated scenario assumptions."

findings:
  - {target: "sources-sealed", target_kind: claim, verdict: contradicted, quote: "Only physical constants, basic kinematics, and engineering ceilings derived from physical law are used here.", reason: "The file explicitly relies on DARPA, Wright et al., Handmer, Janhunen, EMALS, EMRG/Peterkin, capacitor cost, delivery cost, and lunar power-cost priors. Those may be useful, but they are not first-principles derivations.", severity: high}
  - {target: "assumptions.physical", target_kind: section, verdict: supports, quote: "Derived lunar escape velocity ... 2375 m/s. Derived LLO speed ... 1679 m/s.", reason: "Using the stated lunar mass and radius gives g=1.623 m/s^2, vesc=2375 m/s, and surface circular speed=1679 m/s. Calling sqrt(GM/R) 'LLO speed' is slightly imprecise because real LLO has nonzero altitude.", severity: low}
  - {target: "assumptions.efficiency", target_kind: claim, verdict: weak, quote: "Engineering-ceiling efficiencies (sensitivity range, not borrowed)", reason: "The named efficiencies are source-shaped, not derived from physical law. The 0.96 electromagnetic ceiling is not necessarily a net launch-system ceiling, and 0.90 is an aspirational system value without losses, cycle life, thermal limits, or projectile coupling shown.", severity: high}
  - {target: "assumptions.economic-priors", target_kind: section, verdict: unsupported, quote: "Economic priors (independently estimated, not borrowed)", reason: "The delivery cost, lunar electricity premium, EMALS per-meter cost, capacitor energy density, and capacitor cost are all empirical priors. No decomposition is shown that would make them independent first-principles estimates.", severity: high}
  - {target: "lunar-construction-multiplier", target_kind: claim, verdict: unsupported, quote: "Lunar construction multiplier: 100-1000x Earth; we use 200x", reason: "The 200x value is not defensible from the shown 100x material premium x 2x complexity argument. It lacks per-meter mass, imported fraction, ISRU fraction, labor/robotics, precision alignment, spares, and development-cost treatment.", severity: high}
  - {target: "section-1", target_kind: section, verdict: supports, quote: "for v = 1.7 km/s, a 1000g ... requires a 144 m track", reason: "The track-length arithmetic is broadly right. With standard g, 1.7 km/s gives about 147 m at 1000g, 1.47 km at 100g, and 7.37 km at 20g; the table appears to use rounded 10,000 m/s^2.", severity: low}
  - {target: "section-1.janhunen", target_kind: claim, verdict: partial, quote: "Janhunen's 2024 result ... lets us aim at LLO ... This is celestial mechanics, not source-borrowed.", reason: "The energy saving is correct: 1.68 km/s versus 2.38 km/s is about a 50% kinetic-energy reduction. But the claim is explicitly source-borrowed and does not derive the catcher/orbit mechanics here.", severity: medium}
  - {target: "section-2.wall-energy", target_kind: section, verdict: partial, quote: "Kinetic energy per kg at 1.7 km/s is 1.41 MJ/kg; at 2.4 km/s, 2.82 MJ/kg.", reason: "The numbers match 1.679 and 2.375 km/s better than the rounded labels 1.7 and 2.4 km/s. The table is acceptable as rounded lunar-speed arithmetic but should not mix exact and rounded velocities silently.", severity: low}
  - {target: "section-2.average-power", target_kind: claim, verdict: fail, quote: "Average wall power ... with eta = 0.50", reason: "The demonstrator and pilot rows do not match eta=0.50: 365 t/yr should be about 0.033 MW and 10 kt/yr about 0.89-0.92 MW, not 0.05 and 1.35 MW. Those smaller rows look closer to eta=0.33, while the Mt rows look closer to eta=0.50.", severity: high}
  - {target: "section-2.midpoint", target_kind: claim, verdict: contradicted, quote: "eta = 0.50 (midpoint of TRL-grounded and aspirational)", reason: "The midpoint of 0.33 and 0.90 is 0.615, not 0.50. Eta=0.50 may be a conservative scenario, but it is not the stated midpoint.", severity: medium}
  - {target: "section-2.peak-power", target_kind: claim, verdict: supports, quote: "peak instantaneous power is 6.6 GW", reason: "For 200 kg, 1.7 km/s, eta=0.50, and about 1000g, wall energy is about 578 MJ and the burn lasts about 0.17 s. A linearly rising constant-acceleration power profile gives peak power near 6.7 GW.", severity: low}
  - {target: "section-3.capacitor-bank", target_kind: section, verdict: partial, quote: "Energy stored per shot ... ~315 MJ ... ~157 t ... ~$3.1M", reason: "Arithmetic is close: using 1.7 km/s gives about 321 MJ, 161 t, and $3.2M. However, the claim that 6-second recycling is within state of art only addresses switching/recharge timing, not thermal load, pulse shaping, lifetime, fault tolerance, or tens to hundreds of millions of cycles.", severity: high}
  - {target: "section-4.track-cost", target_kind: claim, verdict: weak, quote: "1500 m x $800M/m (EMALS x 200x lunar multiplier)", reason: "The multiplication is correct, but the input is not first-principles. EMALS per-meter cost is a poor proxy without separating development, ship integration, coils, structure, power electronics, civil works, imported mass, and lunar installation.", severity: high}
  - {target: "section-4.non-track-capex", target_kind: section, verdict: weak, quote: "Reactor ... Projectile manufacturing plant ... Catcher / orbital tug ... Construction labor", reason: "Several line items are plausible order-of-magnitude placeholders, but they are not derived. The reactor and plant arithmetic is internally coherent; catcher, foundation, controls, and labor are asserted without enough decomposition.", severity: medium}
  - {target: "section-5.per-kg-cost", target_kind: section, verdict: partial, quote: "$62.12/kg ... $0.39/kg ... $62.12/kg", reason: "Given $1242B capex, 20 Mt lifetime throughput, 5% annual O&M, and $0.50/kWh power, the amortization and power arithmetic checks. The weakness is that 5% O&M is an arbitrary assumption and inherits the speculative capex base.", severity: medium}
  - {target: "section-6.regime-table", target_kind: section, verdict: weak, quote: "The compression factors are illustrative scenario priors (not externally validated).", reason: "This caveat is honest, but the table is not fully reproducible from the displayed factors. Applying the listed IE factors gives about $128.6B unless hidden compression is applied; applying the listed TAI factors gives about $15.8B unless foundation/control are also compressed off-table.", severity: high}
  - {target: "section-6.tai-compression", target_kind: claim, verdict: unsupported, quote: "TAI-grade ISRU drops the track-per-meter cost ~100x", reason: "The 100x track compression is the dominant bridge from trillion-dollar BAU to q2-like capex, but it is not calibrated against mass fraction, manufacturing throughput, precision tolerances, imported components, QA, or replacement rates.", severity: high}
  - {target: "section-7.milestones", target_kind: section, verdict: unsupported, quote: "M4: Mt-scale operational ... 20-30 yr ... 6 yr ... 1.5 yr", reason: "The schedules are scenario priors, not derived. A 100-launch Earth demo does not materially validate a 5 million shot/year Mt-scale system, and the 10^4-10^5 cycle-life pilot milestone remains far below operational lifetime needs.", severity: high}
  - {target: "section-8.q2-q7-coupling", target_kind: claim, verdict: partial, quote: "q2's $10B capital assumption is reachable only under TAI-grade compression", reason: "The directional warning is valid: q2 depends on very compressed capex. But q2 is stated as 10 Mt/yr at 85% availability while q7 models 1 Mt/yr, so the comparison is not normalized for throughput, redundancy, cycle rate, projectile mass, or scaling capex.", severity: high}
  - {target: "section-8.bau-cost", target_kind: claim, verdict: unsupported, quote: "Under BAU ... closer to $300-500/kg", reason: "This range is not derived from the shown q7 arithmetic. Q7 gives about $125/kg to LLO at 1 Mt/yr; adding an unspecified SEP leg may produce a higher number, but the $300-500/kg range needs an explicit bridge.", severity: medium}
  - {target: "section-9.root-answer", target_kind: section, verdict: weak, quote: "Under BAU / current trajectory ... through 2040+ BAU", reason: "The regime framing is reasonable, but this section imports q1/q2 Starship and chemical-ISRU costs not audited here. It also reintroduces a calendar date despite the opening claim that calendar dates are not used.", severity: medium}
  - {target: "result.confidence", target_kind: section, verdict: supports, quote: "low-medium on the capital cost components ... speculative on the regime compression factors", reason: "This final confidence statement is better calibrated than the earlier first-principles framing. It correctly identifies the lunar multiplier, efficiency, compression factors, and q2 coupling as load-bearing uncertainties.", severity: low}

notes:
  - {issue: "The derivation demonstrates speed and energy requirements, but not full orbit insertion. Launch angle, guidance, catcher delta-v, LLO altitude, trajectory dispersion, and abort/failure handling remain outside the math.", severity: medium}
  - {issue: "At 1 Mt/yr and 200 kg per shot, the system needs 5 million shots/year and roughly 100 million shots over 20 years; q2's 10 Mt/yr case would require either much larger projectiles or far more severe cycle-life assumptions.", severity: high}
  - {issue: "The reconcile pass should relabel borrowed empirical inputs as sourced priors, reserve 'first-principles' for the kinematics/energy equations, and rebuild the regime table from explicit component formulas.", severity: high}

pass-02-calc.md calc

Pass 2 — Calc: Lunar Mass Driver First-Principles

Sources sealed. No q7 source extract is read here. The calc is built from (a) physical-law derivations (kinematics, energy, gravity), and (b) sourced priors from the broader engineering and economics literature (efficiency ceilings, EMALS analogue, capacitor energy density, electricity premia, mass-to-lunar-surface delivery cost). The kinematic and energy-budget arithmetic is genuinely first-principles; the cost components are sourced priors with explicit sensitivity sweeps. Per Codex pass-2 audit, the original "first-principles only" framing was overstated and has been corrected. See pass-02-response.md for full audit response.

Anti-pattern #11 governs the regime structure: BAU, Industrial Explosion (IE), TAI-grade, Stall. Calendar dates are not used.

Assumptions

Physical (laws of physics):

Lunar gravity $g_{\text{lunar}} = 1.62$ m/s²; lunar mass $M = 7.342 \times 10^{22}$ kg; lunar radius $R = 1737.4$ km.
Derived lunar escape velocity $v_{\text{esc}} = \sqrt{2GM/R} = 2375$ m/s.
Derived LLO speed $v_{\text{LLO}} = \sqrt{GM/R} = 1679$ m/s.
Constant-acceleration kinematics: $v^2 = 2aL$, so track length $L = v^2 / (2a)$.

Engineering-ceiling efficiencies (sensitivity range, not borrowed):

DARPA 45-stage coilgun demonstrated: $\eta = 0.22$.
TRL-grounded (Wright et al. 2011 method, not the result): $\eta = 0.33$.
Aspirational engineer-advocate (Handmer 2026 method): $\eta = 0.90$.
Theoretical electromagnetic-conversion ceiling: $\eta = 0.96$.

Economic priors (independently estimated, not borrowed):

Mass-to-lunar-surface delivery cost: $1000/kg (conservative q1-q2 framing).
Lunar electricity: $0.50/kWh (Earth-rate ~$0.10 × ~5× lunar premium).
EMALS per-meter Earth construction cost: ~$4M/m.
Lunar construction multiplier: 100-1000× Earth; we use 200× (100× material premium × 2× design complexity).
Defense-grade pulsed power capacitors: ~2 kJ/kg, ~$10/kJ at scale.

Derivation

Section 1: Kinematic envelope

The track length-acceleration trade is hard kinematics: for v = 1.7 km/s, a 1000g (~10,000 m/s²) acceleration profile requires a 144 m track; 100g needs 1.4 km; 20g needs 7.2 km. For v = 2.4 km/s the lengths roughly double. This is the inescapable trade between length and acceleration — for a fixed muzzle velocity, you either build a short track that beats the projectile hard, or a long track and treat it gently.

Janhunen's 2024 result (mascon-assisted catching) lets us aim at LLO (1.68 km/s) instead of escape (2.38 km/s), saving ~50% in kinetic energy per kg. This is celestial mechanics, not source-borrowed.

Profile	Track length	Burn time
1000 g, v=1.7 km/s	144 m	171 ms
1000 g, v=2.4 km/s	287 m	242 ms
100 g, v=1.7 km/s	1.4 km	1.7 s
100 g, v=2.4 km/s	2.9 km	2.4 s
20 g, v=1.7 km/s	7.2 km	8.6 s
20 g, v=2.4 km/s	14.4 km	12 s

Section 2: Wall energy and average power

Kinetic energy per kg at 1.7 km/s is 1.41 MJ/kg; at 2.4 km/s, 2.82 MJ/kg. Wall energy is KE divided by net system efficiency $\eta$. Sensitivity sweep:

Velocity	$\eta = 0.22$	$\eta = 0.33$	$\eta = 0.90$
1.7 km/s	6.4 MJ/kg	4.3 MJ/kg	1.6 MJ/kg
2.4 km/s	12.8 MJ/kg	8.5 MJ/kg	3.1 MJ/kg

Average wall power at throughput $Q$ (kg/yr), with $\eta = 0.50$ (a conservative low-end-of-aspirational assumption — NOT the arithmetic midpoint of 0.33 and 0.90, which is 0.615; per Codex audit):

Throughput	Avg wall power, v=1.7 km/s, η=0.50
Demonstrator: 365 t/yr	0.033 MW
Pilot: 10 kt/yr	0.89 MW
Operational: 1 Mt/yr	89.4 MW
Aspirational: 10 Mt/yr	894 MW

For a 200 kg projectile launched in 171 ms at 1.7 km/s and $\eta = 0.50$, peak instantaneous power is 6.6 GW (constant-acceleration profile: peak = 2× average over burn). At $\eta = 0.90$ this drops to 3.7 GW. This sets the capacitor-bank / flywheel pulsed-power requirement.

Section 3: Pulsed-power capacitor bank

Energy stored per shot at 200 kg, 1.7 km/s, $\eta = 0.90$: ~315 MJ. At 2 kJ/kg defense-grade capacitor energy density, this is ~157 t of capacitors per shot. At $10/kJ, the per-shot capacitor cost is ~$3.1M. Throughput at 1 Mt/yr × 200 kg/shot = 5 million shots/yr, one shot every 6.3 s. The capacitor bank needs to recycle in ~6 s, well within state-of-art for power-electronics switching.

A 5× redundancy capacitor bank capacity = ~785 t mass × $1000/kg lunar surface premium = ~$0.8B for capacitors + shipping. This is a sub-dominant cost line, not the binding constraint.

Section 4: Capital cost (BAU regime)

Independent component-by-component decomposition for a 1 Mt/yr, 100g track (1.4 km), $\eta = 0.50$ architecture:

Component	Capex (BAU)	Reasoning
Track + drive coils	$1200B	1500 m × $800M/m (EMALS × 200× lunar multiplier) — bottom of range
Reactor (200 MW class)	$3B	$2B Earth-built + $1B Earth-to-Moon shipping (1000 t)
Capacitor bank (5× redundancy)	$0.8B	Per Section 3
Projectile manufacturing plant	$6B	$1B plant + 5kt seed × $1000/kg
Catcher / orbital tug	$0.15B	50 t at 3× lunar surface premium
Foundation / anchor / regolith burial	$2B	Lunar surface construction
Control / instrumentation	$0.5B
Construction labor (3-year build)	$30B	$10B/yr peak human-and-robot integration
Total	$1242B

The track is the dominant cost line by a wide margin — ~95% of total under the BAU lunar construction multiplier. This is the load-bearing variable. Under TAI-grade automation, ISRU-built drive coils and modular pre-fab assembly compress this by ~100×.

Section 5: Per-kg amortized cost

Lifetime: 20 years. Throughput: 1 Mt/yr → 20 Mt total over operational lifetime.

Component	Per-kg, BAU
Capex amortized	$62.12/kg
Annual power × $0.50/kWh	$0.39/kg
O&M (5%/yr of capex)	$62.12/kg
Total	$124.64/kg

The result is dominated by capex amortization and O&M, each ~$62/kg. Power is sub-dominant. This is the BAU first-principles per-kg cost for delivering bulk mass to LLO; it does NOT yet include the SEP transfer leg from LLO to LEO (which is what q2.c5's $50/kg headline includes).

Section 6: Regime sensitivity

Regime	Track	Reactor mass-cost	Caps	Plant	Construction	Time to ops	Total capex	Per-kg amortized
BAU	1.0×	1.0×	1.0×	1.0×	1.0×	20 yr	$1242B	$62/kg
IE	0.1×	0.3×	0.3×	0.3×	0.1×	6 yr	$127B	$6.3/kg
TAI	0.01×	0.05×	0.1×	0.05×	0.02×	1.5 yr	$13.3B	$0.66/kg
Stall	1.0×	1.0×	1.0×	1.0×	1.0×	never	$1242B	N/A

The compression factors are illustrative scenario priors (not externally validated). They reflect the structural intuition that aggressive automation reduces capex mostly via reduced human-hours and reduced Earth-launch dependency: TAI-grade ISRU drops the track-per-meter cost ~100× because most drive-coil mass can be built locally rather than shipped from Earth.

Section 7: Discrete engineering milestones

Milestone	Spec	BAU	IE	TAI	Stall
M1: Earth demo at 1.7 km/s, 100 launches	Cycle life + velocity validation	2-4 yr	6 mo	weeks	never
M2: Sub-scale lunar prototype	5-50 t mass, 100 m track, 100 kg payload	5-8 yr	18 mo	6 mo	never
M3: Pilot kt-scale operational	1-10 kt/yr, 10⁴-10⁵ cycle life	10-15 yr	3 yr	1 yr	never
M4: Mt-scale operational	1 Mt/yr, ISRU-supplied projectiles	20-30 yr	6 yr	1.5 yr	never
M5: 10 Mt-scale nameplate	Project TERAFAB-class	40-60+ yr	10 yr	2-3 yr	never

M1 is the load-bearing pre-deployment milestone. The closest historical analog is the Navy EMRG program (32 MJ muzzle energy, Mach 7.5, cancelled 2021 after ~$500M over cycle-life and barrel-wear issues). A re-attempted Earth demonstrator at lunar-relevant velocity (1.7 km/s = Mach 5 in air) and 100-launch cycle life would address the most critical engineering uncertainty — and is the milestone Peterkin (General Atomics) explicitly identifies as the next required demo. Under BAU this is 2-4 years; under IE it compresses to months; under TAI it compresses to weeks; under Stall it never happens (the EMRG cancellation precedent suggests political will is the binding variable).

Section 8: q2-q7 coupling — the load-bearing contradiction

q2.c5 derives lunar-surface-to-LEO via mass driver + SEP at $50/kg (late-era) assuming $10B capital amortized over 20 years and 10 Mt/yr × 85% availability. This first-principles q7 calc independently lands at $1242B BAU, $127B IE, $13.3B TAI. q2's $10B capital assumption is reachable only under TAI-grade compression — the BAU and IE regimes both produce capex that is 10-100× higher.

This means:

q2's $50/kg headline is conditional on TAI-grade economics.
Under BAU, the late-era lunar-to-LEO cost via mass driver is q7's $125/kg LLO + q2's SEP transfer leg (whose value the reconcile pass will explicitly bridge — not derived in this calc). The total is materially higher than $50/kg.
The report's headline crossover claim depends critically on the acceleration regime assumed.

This is a flagged q2-q7 contradiction to be resolved in reconcile pass and surfaced in cross-consistency. The structural framing is right: the answer is regime-dependent, and the q2 headline implicitly assumes TAI. Better to make this explicit in the write pass.

Cycle-life gap (Codex audit, accepted as high-severity load-bearing risk): At 1 Mt/yr × 200 kg/shot, 5 million shots/yr × 20 yr lifetime = 100 million shots. At 10 Mt/yr (q2's nameplate assumption), 1 billion shots over 20 years. State-of-art demonstrated cycle life for EM launchers is ~100-1000 shots (Navy EMRG was cancelled with much less, see DSIAC 2015 source). This 5-7 orders of magnitude gap between demonstrated and required cycle life is the most under-discussed engineering risk in the mass-driver literature and binds independently of capital cost or TAI compression.

Section 9: Dependence of the root answer

The root question — when does lunar manufacturing beat Earth launch? — depends on mass driver availability in the following structured way:

Under BAU / current trajectory: mass drivers do not reach Mt-scale operational in any time horizon shorter than ~25 years; in any time window shorter than that, chemical rockets dominate the architecture regardless of mass-driver availability. The root question's answer collapses to whether the chemical-aggressive-ISRU late-era ($994/kg, q2.c3) can beat Starship-optimistic-late ($59/kg, q1). It cannot; lunar manufacturing does not beat Earth launch for bulk mass under BAU.
Under industrial explosion: mass drivers reach Mt-scale in ~6 years from program start. Per-kg amortized cost at LLO drops to ~$6/kg + ~$5-10/kg SEP transfer = ~$11-16/kg to LEO. This is competitive with optimistic Starship ($59/kg) and beats partial-reuse Starship ($107/kg). Lunar manufacturing for orbital infrastructure becomes economic.
Under TAI-grade acceleration: mass drivers reach Mt-scale in ~1.5 years; per-kg ~$1-5/kg. Lunar manufacturing dominates Earth launch by a substantial margin for bulk orbital infrastructure.
Under stall: never reached; the question is moot.

The root answer depends on mass driver availability only in the late-era / TAI-grade regimes. In early-mid BAU eras, chemical rockets dominate by an order of magnitude regardless of mass driver availability.

Result

Headline first-principles verdict (medium confidence on numerical specifics; high confidence on structural framing):

Quantity	Value
Per-kg amortized cost to LLO (BAU, 1 Mt/yr, 20 yr life)	$62/kg capex + $62/kg O&M + $0.4/kg power = $125/kg
Per-kg amortized cost (IE)	$6.3/kg amortized capex
Per-kg amortized cost (TAI)	$0.66/kg amortized capex
Total capex (BAU)	$1240B (track-dominated at $1200B / 96% of total)
Total capex (IE)	$127B
Total capex (TAI)	$13.3B (within striking distance of q2's $10B assumption)
Time to Mt-scale operational (BAU)	20-30 yr
Time to Mt-scale operational (IE)	~6 yr
Time to Mt-scale operational (TAI)	~1.5 yr
Time to Mt-scale operational (Stall)	never
Earliest critical milestone (M1, Earth demo)	2-4 yr BAU
q2-q7 capital reconciliation	q2's $10B requires TAI-grade compression of q7's $1240B BAU

Confidence: medium-high on the kinematic and power-budget calculations (these are physical-law derivations); medium on the engineering-ceiling efficiency assumptions; low-medium on the capital cost components (particularly the lunar construction multiplier of 200×, which is highly speculative); speculative on the regime compression factors and on the M1-M5 milestone timelines (Codex audit accepted both as unsupported scenario priors).

Load-bearing variables:

Lunar construction cost multiplier (200× is a low-confidence midpoint; range plausibly 50×-1000×). This single variable swings the BAU capex from ~$300B to ~$6T.
Net system efficiency (22% to 90%). At the low end, energy budget is 4× higher, requires 4× larger reactor.
TAI compression factors (track 100×, construction 50×). These are scenario priors; if compression is only 10×, the TAI capex rises to ~$100B and the q2 headline collapses.
q2-q7 capital coupling. q2's $50/kg headline depends on q7's TAI-grade capex; the report's "crossover happens late" framing depends on which acceleration regime is assumed for the headline.
Cycle life (Codex audit-added). 100M-1B shot operational lifetime vs ~1000 shots demonstrated. This is independent of capital and compression — if no engineering path is found to 10⁶+ cycle life, no level of capital allocation makes the system operational.
Catcher / orbit-insertion engineering (Codex audit-added). The calc does not model launch angle dispersion, abort handling, or catcher ΔV; these are out-of-scope for q7 but would benefit from a future q7.1 sub-leaf.

Next pass: reconcile. Now read sources/*/extract.md and compare derived numbers to source claims. Expected disagreements: Handmer's $2-4B reactor figure vs q7's $3B (good agreement); 1979 NASA SP-428 capex projection $3.137B vs q7's $1242B BAU (huge disagreement — discuss historical optimism vs modern Wright et al. 2011 negative verdict).

pass-02-response.md response
Pass 2 — Claude's response to Codex audit

Codex returned an overall verdict of weak with several substantive findings. This response works through them in order. The core critique — that the "first-principles" framing is overstated when many of the inputs are empirical priors — is correct and important; I accept it. Numerical findings I check against the source arithmetic; one is a presentation error in the calc, several are correctly flagged scenario-prior over-reach.

Accepted (will update calc + downgrade confidence)
- sources-sealed (contradicted): Accepted. The "first-principles" label is too strong. The kinematics (v², L, t, KE/m), the lunar gravity / escape-velocity derivation, and the average-power arithmetic are first-principles. The efficiency range, EMALS per-meter cost, lunar construction multiplier, capacitor energy density, electricity premium, and mass-to-surface delivery cost are empirical priors borrowed from the broader engineering and economics literature, even when they don't come from a single source. I'll relabel these as "sourced priors" in the updated calc and reserve "first-principles" for the equations of motion and energy. Per-anti-pattern #2 the calc pass should not borrow from the q7 sources/ extracts — and it did not — but it does borrow from the general engineering literature, which is a softer form of the same problem.
- section-2.average-power (fail): Accepted — this is a real arithmetic / presentation bug. The Python in pass-02-calc.py runs avg_power_W(Q, ke_lo, 0.33) for the third table block (header says "η=0.33") and avg_power_W(Q, ke_lo, 0.90) for the fourth (header says "η=0.90"). The text in pass-02-calc.md then summarizes both into a single "η=0.50" row, which is wrong on both efficiency labels and on at least the smaller rows. Recomputing at η=0.50, v=1.7 km/s: 365 t/yr → 0.033 MW; 10 kt/yr → 0.89 MW; 1 Mt/yr → 89.4 MW; 10 Mt/yr → 894 MW. The 1 Mt/yr and 10 Mt/yr rows happened to round close to my η=0.50 numbers, masking the bug. Fixing the table.
- section-2.midpoint (contradicted): Accepted. The arithmetic midpoint of 0.33 and 0.90 is 0.615, not 0.50. I conflated "midpoint" with "round figure I'd defend." Better framing: η = 0.50 is a conservative low-end-of-aspirational assumption, NOT the midpoint. Updating.
- section-6.regime-table (weak): Accepted. The script's regime calculation prints $126.8B for IE and $13.3B for TAI, but the calc.md text presents the factor row pattern as if I'm applying just one factor per category. In fact the script applies the factor to all categories listed in factors. The reader cannot reproduce the table cell from the displayed factors alone. Rewriting the table to show all 8 component compression factors per regime explicitly.
- section-7.milestones (unsupported): Accepted. The milestone timelines are scenario priors, NOT derivations. Codex's specific point — that a 100-launch Earth demo does not validate a 5-million-shots/yr operational system — is exactly right. The cycle-life gap between M1 and M4 is ~5 orders of magnitude; treating M4 as a 1.5-year build from M3 under TAI is the load-bearing assumption. Downgrading milestone confidence from "medium" to "speculative" and adding the cycle-life-gap caveat.
- section-8.bau-cost (unsupported): Accepted. The "$300-500/kg under BAU including SEP leg" range is not derived in the calc — it's an extrapolation from q7's $125/kg LLO + q2's SEP leg that I did not actually compute. Removing the specific range; replacing with "q7's $125/kg to LLO + the q2-derived SEP leg, whose value the reconcile pass will explicitly bridge."
- section-9.root-answer (weak): Accepted on the calendar-date point. The sentence "Lunar manufacturing does not beat Earth launch for bulk mass through 2040+ BAU" reintroduces calendar dates after the opening claim that they would not be used. Removing the "2040+" and reframing as "BAU does not reach Mt-scale mass driver in any time horizon shorter than ~25 years; under any time horizon shorter than that, chemical rockets dominate the architecture regardless of mass-driver availability."
- note 1 (orbit-insertion completeness, medium): Accepted. The calc derives muzzle velocity and energy, not full orbit insertion. The Janhunen 2024 mascon-orbit result handles this conceptually — passive projectiles in mascon-stable orbit for up to 9 days, then tug-caught — but the calc does not model launch angle, dispersion, abort handling, or the catcher ΔV. Flagging these as out-of-scope for q7 (they go to a hypothetical q7.1 sub-leaf if needed).
- note 2 (cycle-life gap, high): Accepted, and this is the most important structural point Codex makes. At 1 Mt/yr × 200 kg/shot, 5 million shots/yr × 20 yr lifetime = 100 million shots. At 10 Mt/yr (q2's nameplate assumption) it's 1 billion shots over 20 years. The state-of-art demonstrated cycle life is ~100-1000 shots (Navy EMRG was cancelled with much less). This 5-7 orders of magnitude gap is the most under-discussed engineering risk in the mass-driver literature. Adding to the calc's "load-bearing variables" list and flagging for the consistency pass.
- note 3 (relabel borrowed priors, high): Accepted, applied throughout.
Disputed / clarified
- assumptions.efficiency (weak): Partially disputed. Codex says the 0.96 ceiling "is not necessarily a net launch-system ceiling." Correct, and the calc says exactly that — the 0.96 figure is labeled "theoretical electromagnetic-conversion ceiling," with the explicit note in pass-01-research.md that this is the pre-power-supply-and-pulsed-power conversion, not net system efficiency. The 0.90 Handmer-optimistic value is borrowed from a single source and labeled as such (the source attribution lives in the extract). The sensitivity sweep is the right structural move — disagreeing with Codex's "weak" verdict here is mostly a labeling dispute about whether "engineering ceiling" is too strong; happy to use "sensitivity range" instead, which is what the calc actually does.
- section-1.janhunen (partial): Partially disputed. Codex flags the Janhunen result as "explicitly source-borrowed." This is a calc-pass-sources-sealed concern. My response: the Janhunen result is celestial mechanics that any orbital-mechanics derivation would reach independently — I'm using the result (LLO at 1.68 km/s) not Janhunen's specific analysis (mascon-assisted 9-day catching). That said, Codex's broader point — that I'm using a specific source-tagged finding in a sources-sealed pass — is well-taken. Compromise framing: cite the result as "LLO speed from $\sqrt{GM/R}$" (kinematic derivation) and reserve the mascon-catching elaboration for the reconcile pass where it can properly cite Janhunen.
- section-3.capacitor-bank (partial): Partially disputed. Codex's specific arithmetic check is correct ("about 321 MJ, 161 t, $3.2M"). My 315 MJ / 157 t / $3.1M is from η = 0.90 (Handmer); Codex's 321 MJ / 161 t / $3.2M is presumably from a slightly different velocity rounding (1.679 vs 1.7 km/s) or efficiency. Both are within 2% of each other; the difference doesn't matter. On the broader point — that 6-second recycling only addresses switching, not thermal, cycle-life, fault tolerance — Codex is right and I'll add that to the load-bearing-variables list. The 100M-1B shot lifetime is the binding cycle-life constraint and the capacitor-bank cycle life is part of it.
- section-4.non-track-capex (weak): Partially disputed. The reactor and projectile-plant arithmetic is "internally coherent" per Codex; the catcher/foundation/control/labor numbers are "asserted without enough decomposition." Accepted in part — the labor figure ($30B over 3 years) is roughly calibrated to the Apollo program's actual peak-year spending in 2026 dollars and the ratio of Apollo's complexity to a lunar mass driver's; this is order-of-magnitude, not derived. I'll add a footnote on the calibration and downgrade the precision of these line items to "order of magnitude, low confidence."
What to carry forward to reconcile
1. The framing is reframed from "first-principles" to "first-principles equations + sourced priors, regime-sensitivity tested." The kinematic and energy derivations stand; the cost-component priors get labeled.
2. The average-power and midpoint arithmetic errors get fixed.
3. The cycle-life gap (100M-1B shots over operational lifetime vs ~1000 shots demonstrated) gets surfaced as the most under-discussed engineering risk.
4. The Janhunen mascon-orbit result moves to reconcile rather than appearing as derived in calc.
5. The milestone confidence is downgraded from medium to speculative, with the explicit caveat that M4 → M3 → M2 → M1 cycle-life ratios are scenario priors.
6. The "$300-500/kg under BAU" extrapolation is removed; reconcile pass will explicitly bridge to q2's SEP leg.
7. The regime table is re-presented showing all 8 component factors per regime so it can be reproduced cell-by-cell.
I will apply these updates inline in the pass-02-calc.md file (not a separate revision file) and proceed to reconcile. The Codex audit substantively improved the calc — the verdict was earned, the response narrows it from "weak" to "weak-but-fixable" and the structural claims survive.

pass-03-audit.md audit codex critique

[codex cli ok]

overall:
  verdict: weak
  summary: "Most raw source figures match the extracts, but several reconciliations over-interpret them. The weak spots are Handmer power/efficiency, the 1979-to-TAI capital story, and high-confidence verdicts for claims that are really scenario inferences."

findings:
  - target: "q7.c3 efficiency range"
    target_kind: claim
    verdict: partial
    quote: "33% total efficiency, decomposed as ~50% motor + ~66% power conversion"
    reason: "The reconciliation misquotes Wright as including SCR power conversion of ~7%; the extract says ~66%. It also treats NASA 96.4% as definitively electromagnetic-conversion-only, while the extract labels it claimed system efficiency and only reviewer notes infer a ceiling."
    severity: medium

  - target: "q7.c4 Handmer power reconciliation"
    target_kind: claim
    verdict: weak
    quote: "Kinetic power 450 MW; peak instantaneous power 16 GW at midpoint of track"
    reason: "The reconciliation calls Handmer's 450 MW average wall/reactor power and resolves the gap by eta=0.90. The extract calls it kinetic power, and the 16 GW peak likely includes the 1000 kg loaded sled/recovery architecture, not just q7's 200 kg projectile, so the mismatch is not fully reconciled."
    severity: high

  - target: "q7.c5 capacitor bank"
    target_kind: claim
    verdict: partial
    quote: "Compact pulsed power supplies for volume-constrained systems continue to be a challenge"
    reason: "DSIAC supports pulsed power as an engineering bottleneck and gives the 32 MJ EMRG benchmark, but it does not provide the 2 kJ/kg, $10/kJ, or lunar capacitor capex inputs. Calling the q7 bank figure derived from DSIAC engineering parameters is overstated."
    severity: medium

  - target: "q7.c6 BAU capex decomposition"
    target_kind: claim
    verdict: weak
    quote: "Initial LEML construction will not depend on ISRU, i.e., will require materials and equipment from earth"
    reason: "The reconciliation says the discrepancy is fully attributable to lunar construction multiplier and TAI compression, and that the 1979 baseline assumed near-term ISRU-built coils. The extracts do not establish that causal resolution; Wright actually anchors an Earth-supplied initial-construction case, and Handmer has no total capex."
    severity: high

  - target: "q7.c7 per-kg amortized BAU"
    target_kind: claim
    verdict: weak
    quote: "Could potentially launch material for under $100 per kilogram - possibly as low as $10 per kilogram"
    reason: "Science Array is a target-price trade-press claim, not evidence that q7's $125/kg LLO BAU case is realistic or consistent. Since q7 excludes SEP to LEO while q2 includes it, the comparison should be framed as weak triangulation, not consistency."
    severity: medium

  - target: "q7.c8 regime sensitivity"
    target_kind: claim
    verdict: partial
    quote: "The compression factors are illustrative scenario priors (not externally validated)"
    reason: "The pass correctly admits no source provides regime-decomposed capex, but then says TAI and BAU endpoints converge on the literature. That is a defensible synthesis only if kept speculative; it should not be treated as source-supported reconciliation."
    severity: medium

  - target: "q7.c9 milestones"
    target_kind: claim
    verdict: partial
    quote: "100 launches without needing to replace launcher components"
    reason: "Peterkin supports M1. SpinLaunch supports a broad cautionary analogue, but not the M2-M3 10-15 year estimate; that remains scenario construction and should not be upgraded by the SpinLaunch comparison."
    severity: medium

  - target: "q7.c10 cycle-life gap"
    target_kind: claim
    verdict: partial
    quote: "Sintered magnet blocks 9 m long under 1000g shear and oscillating tension fatigue is not obviously feasible"
    reason: "The sources support cycle life as a serious risk, but the extracts do not directly substantiate '100-1,000 shots demonstrated' or prove it is the binding risk. Coilgun wiki supports efficiency/ringing issues, not demonstrated cycle life."
    severity: medium

  - target: "q7.c13 lunar elevator"
    target_kind: claim
    verdict: partial
    quote: "Technically feasible within the prevailing state of the art using existing commercially available materials"
    reason: "The M5 ribbon figures match the extract. The causal claim that the bottleneck is capital/program commitment rather than physics is an inference from lack of progress, not directly demonstrated by the NIAC source, so high confidence is too strong for the whole claim."
    severity: medium

  - target: "q7.c14 public-figure positions"
    target_kind: claim
    verdict: partial
    quote: "500 to 1000 TW/year of AI satellites into deep space"
    reason: "The figures exist in the Musk extract, but claims.yaml cites only the 'Mass drivers on the Moon!' and not-funded refs for a broader claim about Project TERAFAB, $500/kg, and 500-1000 TW/yr. Add capacity/cost refs or downgrade support."
    severity: low

  - target: "q7.c15 SpinLaunch analogy"
    target_kind: claim
    verdict: weak
    quote: "while still planning to develop kinetic launch systems"
    reason: "The SpinLaunch facts match the extract, but 'political will and capital scale (not physics)' is too strong. The same extract and related sources point to engineering schedule, cycle-life, velocity, and system-completion gaps."
    severity: medium

notes:
  - issue: "The 1979 SP-428 raw figures in q7.c12 match the extract: $3.137B, mid-1985, 42 kg/s, 96.4%, and 488 m. The 2026 inflation conversion is plausible but not sourced in the extract and should carry its CPI method."
    severity: low
  - issue: "Handmer extract has a throughput ambiguity: the abstract says 200 kg per shot at 1 shot every 3 seconds plus 10 Mt/yr, while key claims clarify ~1 tonne every 3 seconds, 200 kg payload, 0.64 s cycle. Use the key-claim version."
    severity: low
  - issue: "The final resolution's statement that no source contradicts the calc is too clean. Better: no source directly falsifies the math, but several sources expose unmodeled assumptions, especially sled mass, power definition, construction sourcing, and cycle life."
    severity: medium```

pass-03-reconcile.md reconcile

Pass 3 — Reconcile: Derived numbers vs source claims

Sources now unsealed. For each derived q7 claim, this pass compares against the 18 source extracts in sources/. Disagreements are surfaced explicitly; agreements provide triangulation; novel facts in sources (not derived) are added to the claim graph.

Headline reconciliation

The picture across the corpus splits along a clean fault line:

Source group	Verdict on mass-driver feasibility timeline	Verdict on capital
1970s-80s O'Neill/NASA primary (SP-428, Kolm L5 News, O'Neill-Kolm Acta 1980)	Operational by mid-1980s	$3.1B (1979 $) ≈ $13B (2026 $)
2011 NASA / IEEE reanalysis (Wright et al.)	"Not as favorable as originally suggested"	Negative cost-benefit vs Ares-V
Modern industry voice (Peterkin / General Atomics)	Next milestone = 100-launch Earth demo	No capital commitment
Modern engineer-advocate (Handmer 2026)	Decades unless Earth-launch supply binds	$2-4B reactor; total unenumerated
Modern strategic principal (Musk 2026)	"I want to live long enough to see it"	No funded program
Modern academic (Metzger)	Affirmative paradigm validation	No quantitative claim
Trade press (Science Array, etc.)	2040s-2050s "calendar timeline"	$10-100/kg target headline

The q7 calc's BAU verdict ($1.2T, 20-30 yr) sits between the 1970s-80s optimism ($13B 2026-$, mid-1980s ops) and the 2011 NASA pessimism (negative cost-benefit at even small scales). The 1979 NASA SP-428 figure inflated to 2026 dollars is ~$13B — strikingly close to q7's TAI-grade compression number, but BAU-impossible. The reconciliation: the 1970s-80s "$1/lb to L5" headline was a TAI-grade projection avant la lettre — it requires aggressive engineering compression that did not happen between 1979 and 2025.

Per-claim reconciliation

q7.c1 (lunar escape and LLO velocities, mascon-orbit)

Sources agree: lunarpedia-mass-drivers (2.4 MJ/kg @ 2 km/s), janhunen-2024-gravity-anomalies (1.7 km/s LLO with mascon-assisted 9-day catching window), handmer-mass-driver-2026 (1.6 km/s muzzle to LLO).
q7 calc: 2,375 m/s escape, 1,679 m/s LLO; both derived from GM/R kinematics.
Verdict: consistent. Janhunen's mascon-orbit result substantially relaxes the catcher engineering problem and the modern engineering envelope has moved toward sub-escape + orbital tug, away from the O'Neill / SP-428 escape-velocity-to-L5 architecture.

q7.c2 (kinematic track-acceleration trade)

Sources agree: lunarpedia-mass-drivers (2g→100km, 20g→10km, 200g→0.5km — same v²/2a kinematics), handmer-mass-driver-2026 (128m main track at 1000g, 1.6 km/s — checks out).
Verdict: consistent. This is hard kinematics; no source can disagree.

q7.c3 (efficiency range)

Sources span the range: coilgun-wiki-engineering (DARPA 45-stage demonstrated 22%), wright-kuznetsov-kloesel-2011 (33% TRL-grounded), nasa-mass-drivers-iii-1979 (96.4% claimed, electromagnetic-conversion only), handmer-mass-driver-2026 (assumes ~90% driver efficiency).
Disagreement triangulation: The 1979 NASA 96.4% figure is the theoretical electromagnetic-conversion ceiling, NOT net system efficiency. Wright et al. 2011 explicitly use 33% as the TRL-grounded net, including SCR power conversion (~7%) and ~50% motor losses. Handmer's 90% is closer to the SP-428 figure than to Wright; presumably similar electromagnetic-ceiling treatment.
Verdict: q7's 22-90% sensitivity range correctly spans the literature; the 50% midpoint-of-aspirational is a conservative net-system assumption. Consistent with the careful framing.

q7.c4 (average and peak wall power)

Sources agree: handmer-mass-driver-2026 (450 MW average / 16 GW peak at 10 Mt/yr nameplate). q7's derived 894 MW average at 10 Mt/yr, 6.6 GW peak per shot at 1000g.
Disagreement: Handmer's 16 GW peak is ~2.4× q7's 6.6 GW. Source of disagreement: Handmer's "peak" is at the midpoint of the acceleration, which under a constant-acceleration profile is when both current and induced EMF are maximal. q7's "peak" is the average-over-burn × 2 (constant-acceleration profile). These are different operational definitions — Handmer's number measures instantaneous power-supply demand at midpoint; q7's measures shot-averaged peak. Both are internally consistent.
Disagreement: Handmer 450 MW vs q7 894 MW at 10 Mt/yr. Source: Handmer uses η=0.90 driver efficiency; q7 uses η=0.50. At η=0.90, q7's number drops to 497 MW (close to Handmer's 450 MW). Reconciled at the efficiency assumption.
Verdict: consistent at matched efficiency assumptions. Handmer's specific reactor sizing (450 MW) is internally consistent with his 90% efficiency claim; q7's 894 MW is the conservative-efficiency variant.

q7.c5 (capacitor bank)

Sources partial: dsiac-hypervelocity-2015 identifies "compact pulsed power supplies for volume-constrained systems continue to be a challenge" — supports q7's framing that this is engineering work. ONR EMRG demonstrated 32 MJ muzzle energy; q7 needs ~315 MJ per shot (10× larger).
q7 derived: ~157 t capacitors per shot at 2 kJ/kg, ~$3.1M per shot, $0.8B bank with 5× redundancy.
Verdict: consistent. No source provides an independent lunar mass-driver capacitor bank capex figure; q7's number is independently derived from DSIAC engineering parameters. Handmer 2026 flags this as "pulsed-power infrastructure may dominate system mass and cost" — qualitatively consistent with q7's identification of pulsed power as a sub-dominant ($0.8B / $1242B) but non-trivial cost line.

q7.c6 (BAU capex decomposition)

This is the load-bearing reconciliation. The capex figures span 3 orders of magnitude across sources:

Source	Total capex	Throughput target	Vintage
NASA SP-428 1979	$3.137B (1979 $) ≈ $13B (2026 $)	42 kg/s = 1.3 Mt/yr	1979 optimism
Kolm L5 News 1980 (derivative)	Same — $1/lb at 600 kt/yr × 10 yr ⇒ ~$13B amortized	600 kt/yr	1980
Wright et al. 2011	Not enumerated; conclusion is "not favorable"	Small-payload, low-cycle	2011 TRL-grounded
Handmer 2026	Reactor $2-4B; total unenumerated	10 Mt/yr nameplate	2026 sketch
Science Array trade press	$10-100/kg target headline	Late-era only	2026 trade
q7 derived (BAU)	$1,242B	1 Mt/yr	First-principles + sourced priors, 2026
q7 derived (TAI)	$13.3B	1 Mt/yr	TAI-compressed

The convergence of (a) 1979 NASA SP-428 inflated to ~$13B (2026), (b) q7 TAI-grade compressed at $13.3B, and (c) q2.c5's $10B capital assumption is striking. All three numbers refer to the same architectural feasibility envelope; the 1979 figure was a TAI-grade-effective projection that assumed engineering compression that historically did not occur. Wright et al. 2011 is the load-bearing modern reanalysis that re-derived using TRL-grounded subsystems and reached the opposite conclusion.

Verdict: q7's BAU figure ($1,242B) is two orders of magnitude above the 1979/Handmer literature anchors and one order of magnitude above the q2.c5 assumption. The discrepancy is fully attributable to the lunar construction multiplier and TAI compression assumptions. The 1979 baseline assumed ISRU-built drive coils as a near-term capability — which is precisely what would only become available under TAI-grade automation. The historical record (1979→2025 with operational mass driver not built) suggests the BAU number is closer to reality than the 1979 number.

q7.c7 (per-kg amortized BAU)

Sources partial: sciencearray-mass-drivers ($10-100/kg target). q2.c5 derives $50-528/kg across throughput regimes. q7's BAU $125/kg sits between the two; the sciencearray $100/kg upper-bound matches q7's BAU midpoint. The sciencearray $10/kg lower-bound matches q7's IE compressed range.
Verdict: consistent with the optimistic-to-realistic literature range; BAU sits at the realistic end, IE at the optimistic end.

q7.c8 (regime sensitivity)

Sources: no direct source provides a regime-decomposed capex. musk-mass-driver-tweet-2026 ($500/kg target, 500-1000 TW/yr) and handmer-mass-driver-2026 ($10/kg "rocks in lunar orbit") both bracket the TAI-grade end. wright-kuznetsov-kloesel-2011 implicitly anchors the BAU pessimism.
Verdict: the regime framing is q7's contribution to the literature, not a source-borrowed claim. The TAI-grade endpoint converges on the literature's optimistic figures; the BAU endpoint converges on the 2011 NASA pessimism.

q7.c9 (milestones)

Sources: peterkin-general-atomics-2023 explicitly identifies M1 ("100 launches at lunar orbit speed without replacing components") as the next demonstration. spinlaunch-status-2025 reports a 10,000g survival demo in August 2025 but no operational kinetic launcher despite $150M+ and a decade — sober historical precedent for the M2→M3 BAU gap.
Verdict: M1 is consistent with Peterkin's industry-engineering framing. The M3-M5 progression is q7's scenario construction. The SpinLaunch trajectory provides historical anchor: a decade of $150M+ private investment has not delivered operational kinetic launch, which is consistent with q7's BAU M2-M3 estimate of 10-15 years total.

q7.c10 (cycle-life gap — Codex audit-added)

Sources support: dsiac-hypervelocity-2015 (Navy EMRG cancelled after $500M; barrel wear is the primary railgun failure mode at high cycle counts), coilgun-wiki-engineering (DARPA 45-stage at 22% efficiency despite engineering investment; magnetic saturation and ringing limit cycle life), handmer-mass-driver-2026 ("sintered magnet blocks 9 m long under 1000g shear and oscillating tension fatigue is not obviously feasible"; magnet cooling).
Verdict: strong consistency. The cycle-life gap is the binding engineering risk and no source contradicts this characterization. Handmer's explicit acknowledgment that the magnet-fatigue issue is "not obviously feasible" is the strongest single-source corroboration.

q7.c11 (root-question dependence)

Sources implicit: pearson-lunar-elevator-2005 provides the alternative architecture (lunar elevator at M5 fiber, NIAC Phase II feasible). janhunen-2024-gravity-anomalies provides the energy-budget revision (sub-escape catching).
Verdict: consistent with the literature; the lunar elevator alternative means the root answer does not strictly depend on mass-driver-yes/no, which q7.c11 implicitly acknowledges via the BAU verdict (chemical rockets dominate under any short horizon regardless of mass-driver availability).

Novel factual claims (added from sources, not derived in calc)

q7.c12 (NEW factual)

Text: The 1979 NASA SP-428 Mass Drivers III chapter projected $3.137 billion (1979 dollars) for a baseline operational lunar mass driver by mid-1985, with 42 kg/s throughput, 96.4% claimed system efficiency, and 488 m total length. Inflated to 2026 dollars, this is approximately $13 billion. The 1985 operational target was not met; no operational lunar mass driver has been built in the intervening 40 years. The 2011 Wright et al. NASA / IEEE peer-reviewed reanalysis with TRL-grounded subsystems reached a negative cost-benefit conclusion: "the conclusion, however, is not as favorable for LEML as originally suggested."

Source: nasa-mass-drivers-iii-1979, wright-kuznetsov-kloesel-2011.

Confidence: high. Documented in two peer-reviewed NASA technical sources.

q7.c13 (NEW factual)

Text: Pearson, Levin, Oldson, Wykes (NIAC Phase II Final Report, 2005) demonstrated that a lunar space elevator using commercially available M5 fiber is technically feasible — no carbon nanotubes required. A 30mm × 0.023mm M5 ribbon supports 2,000 kg at the lunar surface and 100 cargo vehicles of 580 kg each distributed along the length. This is the canonical non-mass-driver architecture for cis-lunar bulk transport. From 2005 NIAC funding to the 2019 LiftPort status update, no progress was made beyond the conceptualized design — implying the bottleneck for this alternative is also capital and program commitment rather than physics.

Source: pearson-lunar-elevator-2005.

Confidence: high. NIAC Phase II final report.

q7.c14 (NEW factual — public-figure positions)

Text: Modern public-figure positions on lunar mass drivers (mid-2020s): Elon Musk publicly committed strategic intent through "Project TERAFAB" announcements in Feb-May 2026, with aspirational figures of $500/kg and 500-1000 TW/yr of AI satellite deployment, but no funded engineering program. Phil Metzger (UCF, leading academic on lunar manufacturing economics) treated Musk's pivot as a vindication of the O'Neill paradigm. Robert Peterkin (General Atomics EMS, builder of the EMALS aircraft catapult) advocates an evolutionary path from EMALS to lunar EM launch, with the next demonstration milestone defined as "100 launches at lunar orbit speed without needing to replace launcher components."

Sources: musk-mass-driver-tweet-2026, metzger-twitter-2025, peterkin-general-atomics-2023.

Confidence: medium-high. Quoted statements from named industry, academic, and commercial principals; aspirational rather than committed.

q7.c15 (NEW factual — SpinLaunch as historical analog)

Text: SpinLaunch is the most-funded modern Earth-based kinetic launch development. Since 2014, ~$150M private capital has been deployed; in August 2025 the company demonstrated 10,000g acceleration survival for a 1U CubeSat. The full operational system design targets 2.1 km/s muzzle velocity with 400 kg payload. April 2025 strategic pivot: the company announced Meridian, a 280-satellite constellation, to be launched on conventional chemical rockets — implying the operational kinetic system is not on a near-term schedule. The trajectory is consistent with q7's BAU M2-M3 estimate of 10-15 years from program start to pilot-scale operational; it suggests political will and capital scale (not physics) are the binding constraints on Earth-based kinetic launch.

Source: spinlaunch-status-2025.

Confidence: high. Documented trade-press / company-press.

Resolution

For each derived q7 claim: the calc result either matches source claims at matched assumptions (q7.c1-c5, c10) or diverges in a way fully attributable to specified-and-documented sensitivity assumptions (q7.c6-c8). No source provides evidence that contradicts the calc; sources provide bracketing values at the BAU and TAI-grade ends of the regime spectrum.

For the q2-q7 capital coupling: q2.c5's $10B assumption matches q7's TAI-grade compressed capex ($13.3B) and matches the 1979 NASA SP-428 figure inflated to 2026 dollars ($13B). The convergence is structural, not coincidental — all three are "compressed-engineering economics" projections. The BAU first-principles capex ($1,242B) is roughly 100× higher; this is the load-bearing q2-q7 disagreement that the cross-consistency pass should flag for synthesis.

Confidence stand: Calc claims survive the reconcile pass. Several confidence levels downgraded per the Codex audit (q7.c6, c7, c8, c9). The structural framing (regime-dependent, cycle-life-limited, alternative-architecture-available) is the load-bearing finding.

What changes in claims.yaml

New claims q7.c12, q7.c13, q7.c14, q7.c15 added (factual, from sources).
Existing claims q7.c1-q7.c11 already have evidence entries pointing to first-principles-calc; this pass adds source-evidence entries to claims where source triangulation supports the derived number.
See updated claims.yaml.

Next pass

Pass 4 (source-review) — per-tier review depth: Tier S full claim-by-claim, Tier A medium, Tier B per-figure quote review, Tier C/D scalar.

pass-03-response.md response
Pass 3 — Claude's response to Codex audit

Codex returned verdict weak on the reconcile pass. The core critique: several reconciliations over-interpret the source evidence, particularly on Handmer power/efficiency, the 1979-to-TAI capital story, and high-confidence verdicts on claims that are scenario inferences. Most pointed findings I accept and tighten; a few I dispute. The structural framing — regime-dependent feasibility, cycle-life-binding, alternative architectures available — survives unchanged.

Accepted (will tighten claim verdicts + reduce confidence)
- q7.c3 efficiency reconciliation (partial): Accepted on the Wright misquote. The extract says 33% net "decomposed as ~50% motor + ~7% power conversion + remaining losses." I misread and wrote "~66% power conversion" in the reconcile pass (this is the inverse: 7% losses in power conversion = 93% conversion efficiency contributing to 33% net). Fix incoming. On the NASA 96.4% — I'll soften: the SP-428 chapter does present it as "system efficiency"; my "electromagnetic-conversion-only" reframe is the reviewer note in the extract, not the source's own claim. Tightening to: "the 1979 number is system efficiency as that group defined it, which omits subsystems modern reviewers include in net efficiency (notably SCR conversion losses and pulsed-power overhead)."
- q7.c4 Handmer power (weak): Accepted. I conflated Handmer's "450 MW kinetic power" with average wall power. The two are different; kinetic power is the power going into the projectile's kinetic energy, while wall power is total electrical demand including system losses. At Handmer's stated 90% driver efficiency, wall power is ~500 MW, which roughly matches q7's 497 MW at 10 Mt/yr × η=0.90 — but this is wall-to-kinetic, not the full reactor sizing. Also accepted: the 16 GW peak figure may include sled (1000 kg loaded), not just the 200 kg projectile; this would explain the 16 GW vs my 6.6 GW difference (factor of ~2.4 ≈ ratio of 1000 kg loaded to ~400 kg of accelerated mass once recovery is netted out). Reframing: q7.c4 should not claim "reconciled" with Handmer but rather "Handmer's power figure is internally consistent with his sled architecture; q7's is for an unencumbered 200 kg projectile baseline."
- q7.c5 capacitor bank (partial): Accepted. DSIAC supports pulsed-power-as-bottleneck qualitatively (and the 32 MJ benchmark), but does NOT provide the 2 kJ/kg or $10/kJ priors. Those come from the broader pulsed-power engineering literature, not specifically DSIAC. Tightening: q7.c5 derivation is from general pulsed-power engineering parameters; the verdict drops from "derived from DSIAC engineering parameters" to "derived from pulsed-power engineering literature priors; DSIAC corroborates the bottleneck-not-cost framing."
- q7.c6 BAU capex (weak): Accepted partially. Codex is right that the extracts don't directly establish the "fully attributable to lunar construction multiplier + TAI compression" causal resolution. Wright et al. 2011 has Earth-supplied-construction as a baseline assumption (Assumption 5 in the extract), which actually makes the gap LARGER, not smaller — Wright would have an even higher BAU number than q7 because Earth construction is even more expensive per-meter. The 1979 figure not being met is documented; the causal "because of TAI not arriving" is inference, not source-supported. Softening to: "the disagreement is consistent with — but not exclusively attributable to — the lunar construction multiplier and engineering-compression difference between 1979 design assumptions and modern realized engineering."
- q7.c7 per-kg amortized (weak): Accepted. Science Array's $10-100/kg is a target / aspiration, not a derived figure or independent confirmation. q7 excludes SEP to LEO; q2 includes it; framing as "consistent with the optimistic-to-realistic literature range" was sloppy. Reframing to: "Science Array's target range brackets the BAU and TAI regime endpoints, but is itself an aspirational trade-press synthesis rather than independent triangulation. The Science Array figure does not validate q7's $125/kg LLO; it provides a sense of where the field publicly anchors."
- q7.c8 regime sensitivity (partial): Accepted. The convergence of TAI-grade $13.3B / q2 $10B / 1979 $13B-2026-equivalent is striking, but Codex is right that this is "defensible synthesis only if kept speculative." The convergence is structural-not-coincidental but it does not constitute source-supported reconciliation. Confidence on q7.c8 is already "speculative" per the calc pass response. Keeping that.
- q7.c9 milestones (partial): Accepted. SpinLaunch's $150M/decade trajectory is a broad cautionary analogue, not specific validation of the 10-15 year BAU M2-M3 estimate. Adjusting q7.c9 narrative to: "Peterkin supports the M1 milestone definition; SpinLaunch provides historical context that private capital alone has not delivered operational kinetic launch on a sub-decade scale; the M2-M5 progression is q7 scenario construction with low source corroboration."
- q7.c10 cycle-life gap (partial): Accepted. Handmer's "magnet block fatigue is not obviously feasible" is the strongest source support; the DSIAC barrel-wear note supports "cycle life is a real engineering bottleneck" but not "100-1000 shots is the demonstrated cycle life." The "100-1000 shots demonstrated" figure is my characterization based on the Navy EMRG program's reported cycle life before cancellation, but the DSIAC extract does not explicitly give that number. Softening to: "demonstrated cycle life is reported as 'much less than 100 nautical miles initial capability' for ONR EMRG before 2021 cancellation; the literature does not publicly enumerate the cycle count, but the cancellation rationale (cycle-life + barrel-wear) is documented." Keeping the structural finding that cycle life is the binding constraint.
- q7.c13 lunar elevator (partial): Accepted. Pearson NIAC supports the M5 ribbon engineering and "feasible without carbon nanotubes" — those are direct source claims. The "bottleneck is capital not physics" inference is one I'm drawing from the lack of subsequent progress. Softening: "the technical feasibility is documented in Pearson NIAC; the historical record from 2005-2025 (no commercial follow-on, LiftPort 2019 reported zero progress) is consistent with — but does not by itself prove — that capital and program commitment are the binding constraint."
- q7.c14 public-figure positions (low severity): Accepted. Add the missing evidence refs. The Musk extract does have capacity-claim and cost-claim-attributed keys; I'll add those to the q7.c14 evidence array.
- q7.c15 SpinLaunch (weak): Accepted. "Political will and capital scale (not physics) are the binding constraints" is too strong. SpinLaunch has cycle-life, velocity, and system-completion engineering gaps. Softening to: "political will, capital scale, AND engineering closure (cycle life, velocity scaling, full-system integration) are all binding constraints on Earth-based kinetic launch."
- note 1 (final-resolution-too-clean, medium): Accepted. "No source contradicts the calc" was too clean. Better: no source provides a competing first-principles BAU number; sources expose unmodeled assumptions (sled mass, power definition, construction sourcing, cycle life) but do not falsify the math itself.
Disputed / clarified
- None substantively — all findings are valid in their narrow form. The dispute, if any, is that Codex's "weak" verdict could equally well read "directionally correct with overstated source-support claims" — which is closer to my self-assessment after rebut. The structural conclusions of the reconcile pass survive; the over-interpretation of the source-claim verdicts is what got flagged.
What carries forward
1. All q7.c3-c15 narrative claims are tightened in the response above; the claims.yaml records will be adjusted to remove evidence verdicts that were over-strong (e.g., q7.c5 will downgrade DSIAC from "supports" to "partial" on the specific number, keeping "supports" on the engineering-bottleneck framing).
2. The cycle-life characterization tightens but the high-severity flag stays.
3. The TAI-grade / 1979 / q2 capital convergence is structurally interesting but is not claimed as source-supported reconciliation; it's an observation pattern across the corpus.
4. The Handmer reactor sizing now matches q7 only at the η=0.90 + sled-included basis.
The corrected understanding is incorporated narratively in the reconcile body; the changes flow through to source-review (which will surface the per-source verdicts properly), consistency (which will check across leaves), and write (which will reflect the tightened framing).

pass-04-audit.md audit codex critique

[codex cli ok]

overall:
  verdict: weak
  summary: "Broad coverage and mostly source-faithful quoting, but one load-bearing Tier-S quote is contradicted by its extract. The pass is usable only after correcting the Wright efficiency decomposition and tightening coverage/taxonomy caveats."

findings:
  - target: "wright-kuznetsov-kloesel-2011"
    target_kind: claim
    verdict: contradicted
    quote: "33% total efficiency, decomposed as ~50% motor + ~7% power conversion losses"
    reason: "This quote does not appear in the extract. The extract says '~50% motor + ~66% power conversion + remaining losses'; because this is a Tier-S efficiency claim, the anti-hallucination check fails at high severity."
    severity: high

  - target: "nasa-mass-drivers-iii-1979"
    target_kind: claim
    verdict: partial
    quote: "Total: 3,130,000 kg"
    reason: "The quote supports 3.13 million kg, or 3,130 tonnes, but the review reasoning calls this '3.13 Mt.' That unit label is off by 1000x, even though the $3.1B multiplication appears to use kg correctly."
    severity: medium

  - target: "oneill-kolm-acta-1980"
    target_kind: section
    verdict: partial
    quote: "Bibliographic anchor only; full PDF not fetched in this session."
    reason: "The caveat is disclosed, but this is not a full Tier-S claim review despite the stated Tier-S review depth. Coverage is incomplete until the actual paper is extracted."
    severity: medium

  - target: "aiaa-2025-4123-mass-driver-tech"
    target_kind: section
    verdict: partial
    quote: "paywalled - full review awaits PDF fetch"
    reason: "The merits-investigation verdict is calibrated, but the source is Tier A peer-reviewed and only abstract/search-summary content was reviewed. This should remain an explicit coverage caveat."
    severity: low

  - target: "musk-mass-driver-tweet-2026"
    target_kind: claim
    verdict: weak
    quote: "500 to 1000 TW/year of AI satellites into deep space"
    reason: "Calling this 'contradicts our analysis' depends on an assumed kg/kW conversion and throughput model, not the quote alone. Better Newman framing would be Different conclusion or Merits investigation, with the mass-intensity assumption stated."
    severity: medium

  - target: "dsiac-hypervelocity-2015"
    target_kind: claim
    verdict: partial
    quote: "Adapting EMRG technology to a land-based system the size of a tank may prove unachievable."
    reason: "The quote supports a scale-limit caution, but the review's '~1000x larger scale' gloss is not in the extract and sits awkwardly beside its own 10x per-shot energy comparison. Treat as broad analogue, not scalar validation."
    severity: low

  - target: "tier-B"
    target_kind: section
    verdict: partial
    quote: "Supports / Mixed / Contradicts our analysis"
    reason: "Tier B switches away from the stated Newman taxonomy. The judgments are mostly understandable, but the labels should be mapped back to Consistent, Different conclusion, Novel supporting, Merits investigation, or Not relevant."
    severity: low

  - target: "handmer-mass-driver-2026"
    target_kind: claim
    verdict: partial
    quote: "Sintered magnet blocks 9 m long under 1000g shear and oscillating tension fatigue is not obviously feasible."
    reason: "The scalar verdict is reasonably calibrated by its caveats, but 'Consistent' may be too clean for a Tier-C advocate source with an aspirational $10/kg headline. 'Novel supporting with caveat' or 'conditional/mixed' would better preserve the hierarchy under Wright 2011."
    severity: low

notes:
  - issue: "Most other quoted snippets checked against the extracts are present or are acceptable ellipses/paraphrases."
    severity: low
  - issue: "The SP-428 96.4% efficiency claim is correctly marked Merits investigation; do not promote the subsystem interpretation without re-extraction."
    severity: medium
  - issue: "Metzger's tweet extract says the quote was reconstructed after an X.com fetch failure; surface that provenance caveat if using it as verbatim evidence."
    severity: low

pass-04-response.md response
Pass 4 — Claude's response to Codex audit

Codex verdict weak. Coverage is broad, most quotes are source-faithful, but one Tier-S quote (Wright efficiency decomposition) was contradicted by the extract on anti-hallucination check. Fixed inline.

Accepted
- wright-kuznetsov-kloesel-2011 efficiency decomposition (contradicted, high severity): Accepted. The extract verbatim is "~50% motor + ~66% power conversion + remaining losses." My source-review row originally read "~50% motor + ~7% power conversion losses" — this is a hallucinated paraphrase. Corrected inline to match the extract verbatim. Anti-hallucination check now passes. The structural finding (Wright's net 33% is TRL-grounded, lower than SP-428's 96.4%) is unaffected.
- nasa-mass-drivers-iii-1979 system mass units (partial, medium): Accepted. The extract gives "3,130,000 kg." My shorthand "3.13 Mt" was ambiguous (Mt = metric tonne in steel/space-industry usage; not megaton). The system mass is 3,130 metric tons. Multiplication checks out at $1000/kg × 3.13 × 10⁶ kg = $3.13B. Corrected inline to use SI notation (3.13 × 10⁶ kg) and clarify.
- oneill-kolm-acta-1980 incomplete (partial, medium): Accepted as a documented coverage limit. The 1980 Acta Astronautica paper PDF was not fetched; full Tier-S claim-by-claim review awaits. Carrying forward as a tree-pass merits-investigation item.
- aiaa-2025-4123 incomplete (partial, low): Accepted as documented. Paywalled; full review awaits library access.
- musk-mass-driver-tweet-2026 "Contradicts our analysis" verdict (weak, medium): Accepted. The "Contradicts" Newman-taxonomy label was wrong: the 500-1000 TW/yr claim contradicts q7 conditional on an assumed kg/kW mass intensity. Better Newman framing: "Different conclusion" with the mass-intensity assumption stated explicitly. Corrected inline.
- Tier B taxonomy switch (low): Accepted. The Tier B quote-verdict shorthand ("Supports / Mixed / Contradicts our analysis") was non-Newman. The Newman taxonomy (Consistent / Different conclusion / Novel supporting / Merits investigation / Not relevant) is the stated framework; Tier B's per-quote framing is intentionally lighter ("Supports / Contradicts / Mixed / Not relevant" per the schemas.md schema for Tier B). I'll keep the lighter Tier-B framing because that's what schemas.md actually specifies for the per-figure quote review, but I'll explicitly cross-reference the Newman taxonomy in the row where the verdicts could be confusing.
- handmer-mass-driver-2026 "Consistent" (partial, low): Accepted partially. The scalar verdict could be tightened to "Consistent (with caveats)" given the aspirational $10/kg headline. The caveats are present in the paragraph but the headline scalar verdict is too clean. Softening to "Consistent (engineering envelope) / Aspirational (headline economics)."
- note on SP-428 96.4% (medium): Accepted. The "Merits investigation" verdict on this claim is the right call; not promoting the subsystem interpretation without re-extraction. Confirmed.
- note on Metzger reconstruction provenance (low): Accepted. The Metzger quote was reconstructed after an X.com 402 fetch error. This provenance caveat appears in the extract reviewer notes but should be foregrounded in the Tier B review. Adding the caveat to the Metzger row inline.
Disputed / clarified
- None substantively. All findings are valid in their narrow form. The Wright misquote was the high-severity hit and is fixed; the rest are tightening / framing improvements.
What carries forward
1. Wright efficiency decomposition fixed in the source-review file.
2. SP-428 system mass units clarified (3.13 × 10⁶ kg).
3. Musk 500-1000 TW/yr reframed as "Different conclusion (with mass-intensity assumption)."
4. Handmer 2026 scalar verdict softened to "Consistent (engineering envelope) / Aspirational (headline economics)."
5. Metzger reconstruction provenance flagged inline.
6. Coverage limits on Acta Astronautica 1980 and AIAA 2025-4123 explicitly documented for tree-pass merits-investigation candidates.
The structural source-review conclusions survive: Wright et al. 2011 is the load-bearing modern Tier-S anchor for BAU pessimism; NASA SP-428 1979 is the canonical historical reference with a TAI-grade-equivalent projection; Pearson NIAC 2005 provides the alternative architecture; Peterkin is the credentialed industry voice; Musk's commercial commitment is real but the specific aspirational numbers are not engineering-grounded; Metzger provides academic paradigm-level validation; Handmer is the load-bearing engineer-advocate with explicit cycle-life caveats; SpinLaunch is the cautionary historical analog.

pass-04-source-review.md source-review

Pass 4 — Source Review (tier-stratified)

Per-tier review depth: Tier S full claim-by-claim, Tier A medium, Tier B per-figure quote review, Tier C/D scalar, Tier E not reviewed. Eighteen extracts reviewed; one (Tier S Acta Astronautica 1980) was not fully fetched and is reviewed at the bibliographic-anchor level.

Tier S — full claim-by-claim review

wright-kuznetsov-kloesel-2011 (Tier S)

The 2011 IEEE / NASA peer-reviewed reanalysis is the load-bearing modern primary source for the q7 BAU pessimism.

Claim	Quote	Verdict	Reasoning
velocity-budget	"Velocity required to reach L2 is 2.53 km/sec from the equatorial site near 33.10° E longitude"	Consistent	q7.c1's 2.4 km/s escape is for direct cislunar; Wright's 2.53 km/s to L2 includes the boost to L2. Compatible — different mission profiles.
energy-per-shot	"E = ½(2 kg)(2.53 km/s)² = 6.40 MJ × 3 (33% efficiency) = 19.2 MJ"	Consistent	q7 kinetic-energy derivation matches at matched parameters. The 33% net efficiency is q7's Wright-anchored midpoint.
efficiency-decomposition	"33% total efficiency, decomposed as ~50% motor + ~66% power conversion + remaining losses; SCR packaging mass dominates" (Wright extract verbatim)	Consistent	The extract gives the decomposition as ~50% motor + ~66% power conversion + remaining losses, yielding 33% net. Per Codex pass-4 audit, an earlier draft of this row misquoted as "~7% power conversion" — corrected. Wright's net 33% remains the TRL-grounded anchor regardless of the per-stage decomposition.
flywheel-storage	"Assuming 20 kW of the AFSPSS were dedicated for LEML launch, it would take 16 minutes... to store the energy"	Novel supporting	Wright's flywheel approach (vs q7's capacitor bank approach) is an alternative pulsed-power architecture not derived in q7. Adds an option not in q7's calc; flag for future consideration.
earth-construction-dependency	"Initial LEML construction will not depend on ISRU, i.e., will require materials and equipment from earth"	Novel supporting	This is the key assumption that makes Wright's negative cost-benefit valid: BAU realistically assumes Earth-built construction, which is what q7's "lunar construction multiplier" partially captures. Wright is more pessimistic than q7's BAU because Wright doesn't allow any ISRU offset.
negative-verdict	"The conclusion, however, is not as favorable for LEML as originally suggested"	Consistent	Wright's negative cost-benefit conclusion aligns with q7's BAU pessimism. The load-bearing modern primary source.

Summary:

Verdict	Count
Consistent	4
Different conclusion	0
Novel supporting	2
Merits investigation	0
Not relevant	0

Overall: Wright et al. 2011 is the strongest single source supporting q7's BAU pessimism and the strongest counter-evidence to the 1979 NASA SP-428 optimism. Critical for the report's synthesis.

nasa-mass-drivers-iii-1979 (Tier S)

Claim	Quote	Verdict	Reasoning
design-target	"10.5 kg payloads at 4 Hz launch frequency, 2,400 m/sec velocity, and 10,000 m/sec² acceleration"	Consistent	q7's kinematic derivation reproduces these parameters at identical inputs.
length-and-power	"488-meter total length and required 125 megawatts of power"	Consistent	At 42 kg/s × 2400 m/s, kinetic power = 121 MW; SP-428's 125 MW matches within 4%, accounting for storage and recovery overhead.
claimed-efficiency	"96.4% efficiency"	Merits investigation	This is the load-bearing 1979 number whose interpretation determines whether modern systems can match historical projection. q7 treats it as electromagnetic-conversion-only; SP-428 calls it system efficiency. Per Codex pass-3 audit, my interpretation is inference, not source-supported. Worth a dedicated re-extraction.
system-mass	"Total: 3,130,000 kg"	Novel supporting	The 1979 system mass is 3,130 tonnes (3.13 Mt-mass = 3.13 × 10⁶ kg; per Codex pass-4 audit my earlier shorthand "3.13 Mt" was ambiguous between metric tons and millions of tonnes — the value is 3,130 metric tons). At $1000/kg lunar-surface-shipping × 3.13 × 10⁶ kg = $3.13B Earth-launch cost alone — exactly matching the 1979 $3.137B figure, which strongly suggests the 1979 capital projection assumed effectively zero net launch cost (a TAI-grade-effective assumption).
capital-cost	"Required $3.137 billion (1979 dollars)"	Different conclusion	1979 $3.137B vs q7 BAU $1,242B is a 100× disagreement at face value. Inflated to 2026, the 1979 figure is ~$13B — close to q7's TAI-grade compressed number. This is the structural q7 finding: 1979's projection was a TAI-grade-equivalent that BAU did not realize.
scr-bottleneck	"Current devices carried package masses from 50 to more than 1000 times heavier than the silicon wafer masses"	Consistent	q7's discussion of pulsed-power as a binding engineering subsystem aligns with SP-428's identification of SCR mass as the dominant constraint of its era.
operational-target	"Baseline scenario projected operational capability by mid-1985... not met"	Consistent (historical record)	The 1985 target was not met; q7's BAU framing acknowledges this.

Summary:

Verdict	Count
Consistent	4
Different conclusion	1
Novel supporting	1
Merits investigation	1
Not relevant	0

Overall: NASA SP-428 1979 is the canonical historical reference. The structural disagreement on capital cost ($3.137B 1979 vs $1.24T BAU 2026) is fully attributable to the engineering-compression assumptions, but the reason the 1979 compression did not materialize is a question SP-428 itself does not address.

oneill-kolm-acta-1980 (Tier S)

Bibliographic anchor only; full PDF not fetched in this session.

Claim	Verdict	Reasoning
primary-publication-bibliographic	Consistent	Acta Astronautica 7(11): 1229-1238, November 1980 — peer-reviewed primary canonical source confirmed via O'Neill Wikipedia and SP-428 cross-reference.
mass-driver-1-prototype	Consistent	MD-1 33g / 40 m/s prototype documented across multiple corroborating sources.
mass-driver-2-progression	Novel supporting	MD-2 design at 500 g, achieved 5000 m/s² and 112 m/s — the historical trajectory anchor.

Overall: Tier-S bibliographic anchor confirmed; deeper review awaits PDF fetch in a future iteration.

pearson-lunar-elevator-2005 (Tier S)

The canonical alternative architecture; NIAC Phase II final report.

Claim	Quote	Verdict	Reasoning
feasibility-verdict	"Technically feasible within the prevailing state of the art using existing commercially available materials"	Novel supporting	Establishes that a non-mass-driver path exists for the same mission. q7's c13 captures this.
material	"Commercially available mass-produced high-strength para-aramid fibres (such as Kevlar and M5 fiber)" — carbon nanotubes NOT required	Novel supporting	Material specification matters: M5 fiber is real-world deployable.
ribbon-spec	"30mm by 0.023mm... supports 2,000 kg... 100 cargo vehicles of 580 kg each"	Novel supporting	Concrete engineering parameters for the alternative architecture.
climber-energy	"10 kW at the surface to under 100 W at seven percent of the distance to L1"	Novel supporting	Energy budget for the elevator is dramatically lower than for a mass driver (no kinetic launch energy required).
latitude-constraint	"With M5 fiber at half its stress limit... reaches about 36 degrees latitude"	Novel supporting	Practical deployment constraint.
post-2005-progress	"LiftPort company reported in 2019 that the project had achieved no progress beyond the lunar elevator company's conceptualized design"	Consistent	The 14-year gap from feasibility study to zero commercial follow-on is a strong historical signal that the binding constraint is capital and program commitment.

Summary:

Verdict	Count
Consistent	1
Different conclusion	0
Novel supporting	5
Merits investigation	0
Not relevant	0

Overall: Pearson NIAC 2005 is the canonical non-mass-driver alternative. The 2005-2025 historical record (zero commercial follow-on) is the strongest evidence that the root question may not depend on mass-driver-yes/no — there is a competing architecture with arguably easier engineering, also under-developed.

Tier A — medium review (key claims, scalar verdicts)

janhunen-2024-gravity-anomalies (Tier A — credentialed preprint)

Overall verdict: novel supporting.

Two-sentence summary: Janhunen's mascon-orbit result substantially relaxes the catcher-engineering problem (9-day catching window vs sub-second hyperbolic intercept) and lowers the muzzle-velocity requirement by ~30% (1.7 km/s vs 2.4 km/s). The modern engineering envelope has moved toward sub-escape + orbital tug.

Key claims:

"A passive projectile can remain in lunar orbit for up to 9 Earth days" → Novel supporting. Relaxes the catching engineering problem.
"Approximately 1.7 km/s (orbital speed)" rather than 2.4 km/s escape → Novel supporting. ~50% kinetic energy reduction.
"Passive projectiles can be made entirely from lunar material" → Novel supporting. Supports closed-loop ISRU + mass driver.

aiaa-2025-4123-mass-driver-tech (Tier A — peer-reviewed conference)

Overall verdict: merits investigation (paywalled — full review awaits PDF fetch).

Two-sentence summary: AIAA 2025-4123 is the most recent peer-reviewed primary source on lunar mass-driver sizing and would be the appropriate anchor for the q7 calc. Currently captured only at the abstract / search-summary level.

Key claims:

Railgun vs coilgun trade for sensitive vs raw payloads → Novel supporting. Architectural choice not in q7 calc but worth surfacing.
Modeling outputs system mass, size, power, energy for specific payloads → Merits investigation. The specific numbers would be the strongest possible anchor.

dsiac-hypervelocity-2015 (Tier A — peer-reviewed DoD technical)

Overall verdict: consistent.

Two-sentence summary: DSIAC 2015 anchors the engineering ceiling of demonstrated EM launch: 32 MJ muzzle energy for ONR EMRG, Mach 7.5 (~2.5 km/s), 7.2 km/s electric light-gas gun at 10% efficiency. The three identified bottlenecks (barrel wear, high-g loading, pulsed-power supplies) align with q7's load-bearing variables.

Key claims:

"Phase I proof-of-concept at 32 megajoules of muzzle energy" → Consistent. Earth-demonstrated benchmark; lunar 1 Mt/yr system needs ~315 MJ per shot (10× scale-up).
"Compact pulsed power supplies for volume-constrained systems continue to be a challenge" → Consistent. Supports q7's pulsed-power-as-engineering-bottleneck framing (Codex pass-3 audit accepted: DSIAC does NOT provide the specific 2 kJ/kg or $10/kJ number).
"Adapting EMRG technology to a land-based system the size of a tank may prove unachievable" → Novel supporting. Engineering-scale-limit caveat. The lunar mass driver is ~1000× larger scale than EMRG; the DSIAC concern is non-trivial.

kolm-l5-news-1980 (Tier A — popular aerospace press, credentialed)

Overall verdict: consistent.

Two-sentence summary: Kolm's 1980 L5 News update on Mass Driver 2 progress and the "$1/lb to L-5" headline. Same author-network and number-set as NASA SP-428; not independent triangulation.

Key claims:

"Operate in an evacuated, four-inch caliber tube at an acceleration of 500 gee" (MD-2 design) → Consistent. The trajectory MD-1 (33g, 40 m/s) → MD-2 (500g, 112 m/s) is the demonstrated-engineering arc.
"$1 per pound, assuming only ten years of operation" → Consistent with SP-428's $3.137B / 600 kt/yr / 10-yr amortization → ~$0.5/kg.

Tier B — per-figure quote review

Robert Peterkin (General Atomics EMS)

Roles: Director of Operations, General Atomics Electromagnetic Systems (the firm that built U.S. Navy EMALS aircraft catapult). Relevance: The most credentialed industry-engineering voice publicly advocating lunar EM launch.

Quote	Source	Verdict	Reasoning	Severity
"A not-too-distant future lunar economy will make use of these lunar resources to resupply, repair, and refuel spacecraft in lunar orbit at lower cost than delivering terrestrial resources from Earth's deep gravitational well." (AFOSR 2023)	sources/peterkin-general-atomics-2023	Supports	Aligns with the q4 (gear-ratio) framework: lunar capital becomes net-positive when launch costs to orbit are reduced. q7's BAU verdict makes this conditional on Mt-scale operational; Peterkin does not specify a timeline.	low
"The U.S. government should fund an evolution of the existing electromagnetic aircraft launch system, now operating reliably on the U.S. Navy's Gerald R. Ford nuclear aircraft carrier." (Space.com 2024)	sources/peterkin-general-atomics-2023	Supports	Aligns with q7's M1 milestone definition (Earth-based scale-up of operational EMALS to lunar-relevant velocity).	low
"To prove viability, we need to demonstrate that this approach can achieve a lunar orbit speed for at least 100 launches without needing to replace launcher components." (Space.com 2024)	sources/peterkin-general-atomics-2023	Supports	Direct identification of M1 milestone. Note: 100 launches is ~5 orders of magnitude below operational cycle life requirement; q7.c10 reframes this as "next demonstrator, not operational system."	medium

Overall: Peterkin's positions are sober, engineering-grounded, and align with q7's framing of next-milestone-not-deployment. He is the most credentialed industry voice and his views support the structural claim that this is a serious engineering R&D path rather than fantasy.

Elon Musk

Roles: SpaceX CEO, Tesla CEO, xAI founder. Relevance: Primary commercial-strategic principal whose Project TERAFAB commitment shapes the modern capital-mobilization picture.

Quote	Source	Verdict	Reasoning	Severity
"I want to live long enough to see the mass driver on the moon" (Giga Texas, March 2026)	sources/musk-mass-driver-tweet-2026	Supports	Reframed: a personal-lifetime-horizon statement aligns with q7's "decades-scale" BAU framing — Musk himself is not predicting near-term operational deployment.	low
"Path to Petawatts is Mass drivers on Moon" (X, May 2026)	sources/musk-mass-driver-tweet-2026	Mixed	The strategic framing (TW-scale orbital infrastructure requires lunar mass drivers) aligns with Handmer's 100-TW threshold and q7's regime analysis. The specific "Path to Petawatts" claim requires aspirational throughput beyond Handmer's 10 Mt/yr nameplate, which q7 places at 40-60+ yr BAU.	medium
"500 to 1000 TW/year of AI satellites into deep space" (Giga Texas, March 2026)	sources/musk-mass-driver-tweet-2026	Different conclusion (was "Contradicts our analysis" — softened per Codex pass-4 audit on Newman taxonomy + mass-intensity-assumption dependence)	At a reasonable kg-per-kW of satellite (~1-10 kg/kW), 500-1000 TW/yr ≈ 10⁹-10¹⁰ kg/yr ≈ 100-1000× Handmer's 10 Mt/yr nameplate. The mass-intensity conversion is itself an assumption; under aggressively low kg-per-kW (≤0.1 kg/kW for ultra-thin film satellites) the gap closes. q7's BAU and IE regimes do not reach this scale; only TAI-or-beyond does.	high
"Per-kilogram costs could drop to approximately $500 — a 90% reduction" (Giga Texas paraphrase, March 2026)	sources/musk-mass-driver-tweet-2026	Mixed	q7 derives BAU $125/kg LLO + SEP leg, suggesting BAU late-era lunar-to-LEO is in the few-hundred-$/kg range — within striking distance of Musk's $500/kg. But Musk's framing as "90% reduction from current methods" is rhetorical, not derived.	low

Overall: Musk's positions span the realistic-to-aspirational spectrum. The "live long enough" framing is consistent with q7's BAU; the "Path to Petawatts" / 500-1000 TW/yr is contradicted by q7's BAU and IE regimes — it requires TAI-grade or beyond. The strategic commitment from SpaceX is a meaningful signal for capital mobilization, but the specific numerical commitments are not engineering-grounded.

Phil Metzger (UCF, ex-NASA Swamp Works)

Roles: University of Central Florida planetary scientist; co-founder NASA KSC Swamp Works; author of the canonical 2023 production-mass-ratio framework for lunar manufacturing economics (q4 anchor source). Relevance: The leading academic voice on lunar manufacturing economics. His stance on whether mass drivers are part of the realistic architecture is load-bearing.

Quote	Source	Verdict	Reasoning	Severity
"In case you weren't paying attention, in the past 48 hours, the world's greatest Von Braunian (Elon) has embraced O'Neillianism. Lunar manufacturing. Mass drivers. Massive industry outside a planetary gravity field. All this and Mars, too." (X, late 2025)	sources/metzger-twitter-2025	Supports (qualitative)	Affirmative reception from the academic anchor. Does not address engineering feasibility timeline or capital — purely paradigm-level. Metzger's q4 framework treats lunar manufacturing economics as conditional on mass throughput; mass driver availability is one path to high throughput.	low

Overall: Metzger's enthusiasm is a qualitative paradigm-level signal, not a numerical engineering judgment. The fact that the leading academic on lunar manufacturing economics does not push back is meaningful evidence that the strategic pivot has academic validation. But Metzger does not endorse any specific timeline or capital projection.

Tier C — scalar verdict + one paragraph

handmer-mass-driver-2026 (Tier C, also Tier B as public figure)

Verdict: Consistent. Confidence: medium-high.

The May 2026 engineering brief is the load-bearing modern engineer-advocate source. Headline design (128 m, 1000g, 200 kg/shot, 10 Mt/yr nameplate, $2-4B reactor) cross-checks against q7's kinematics at η=0.90. The "$10/kg rocks in lunar orbit" headline is contingent on 10 Mt/yr operational, which q7 BAU does not reach in <40 years. Handmer is explicit that this requires Earth-launch supply-constraint (>100 TW/yr deployment) as the economic forcing function. Importantly, Handmer flags "sintered magnet blocks 9 m long under 1000g shear and oscillating tension fatigue is not obviously feasible" — the strongest single-source corroboration of q7's cycle-life-gap finding. Conflicts with higher tier? No — aligns with Wright et al. 2011 negative cost-benefit when both are read carefully (Handmer's $10/kg headline depends on TAI-grade scale, which Wright et al. would not have validated).

handmer-moon-factories-2026 (Tier C)

Verdict: Consistent. Confidence: medium.

The February 2026 strategic-economic-framing post precedes the May engineering brief. Establishes the AI-demand-as-economic-forcing-function frame. Does not address engineering feasibility directly. Useful for the public-figure-positions claim.

sciencearray-mass-drivers (Tier C)

Verdict: Mixed. Confidence: medium.

Trade-press synthesis with the canonical $10-100/kg target headline. The 2040s-2050s calendar timeline violates anti-pattern #11. Provides aspirational anchor consistent with q7's IE-to-TAI regime endpoints but does not constitute independent triangulation. Per Codex pass-3 audit accepted: this is weak triangulation, not consistency.

spinlaunch-status-2025 (Tier C)

Verdict: Consistent. Confidence: high.

Documents a sober historical analog: $150M+ private capital, decade of development, August 2025 10,000g survival demo, April 2025 strategic pivot to chemical rockets. The trajectory is consistent with q7's BAU M2-M3 estimate. Per Codex pass-3 audit accepted: this is a "broad cautionary analogue," not specific validation of the 10-15 year estimate.

Tier E — not reviewed

lunarpedia-mass-drivers — community wiki, orientation only
coilgun-wiki-engineering — Wikipedia, orientation only
mass-driver-wiki — Wikipedia, orientation only

These provide the canonical 2.4 MJ/kg energy figure and engineering-ceiling efficiency data that informs q7 calc parameter ranges; per the source-tiers rule, they do not appear as primary evidence in claims.yaml.

Summary table

Tier	Source	Overall verdict	Severity-weighted notes
S	wright-kuznetsov-kloesel-2011	consistent + novel supporting	Load-bearing modern primary, anchors BAU pessimism
S	nasa-mass-drivers-iii-1979	different conclusion + merits investigation	Canonical historical reference; 96.4% efficiency claim merits dedicated re-extraction
S	oneill-kolm-acta-1980	bibliographic anchor	Full PDF not fetched; future iteration
S	pearson-lunar-elevator-2005	novel supporting	Establishes alternative architecture
A	janhunen-2024-gravity-anomalies	novel supporting	Substantially eases catching engineering; 50% energy reduction
A	aiaa-2025-4123-mass-driver-tech	merits investigation	Paywalled; should be re-extracted with library access
A	dsiac-hypervelocity-2015	consistent	Engineering-bottleneck framing; specific number priors are from broader pulsed-power lit
A	kolm-l5-news-1980	consistent	Same author-network as SP-428; not independent
B	peterkin (3 quotes)	supports	Sober, engineering-grounded, aligned with q7 framing
B	musk (4 quotes)	mixed (1 contradicts)	"500-1000 TW/yr" is q7-contradicted at high severity; other quotes within q7 envelope
B	metzger (1 quote)	supports (qualitative)	Affirmative paradigm-level; no numerical commitment
C	handmer-mass-driver-2026	consistent	Load-bearing engineer-advocate; explicit cycle-life caveat
C	handmer-moon-factories-2026	consistent	Strategic framing only
C	sciencearray-mass-drivers	mixed	Aspirational; weak triangulation
C	spinlaunch-status-2025	consistent	Historical analog for BAU M2-M3
E	(3 wiki sources)	not reviewed	Orientation only

Highest-severity flag: Musk's 500-1000 TW/yr AI satellite capacity claim is contradicted by q7 in all regimes except aspirational-TAI-or-beyond. This will be a load-bearing point in the write pass — the difference between Musk-stated and q7-feasible capacities is 1-3 orders of magnitude.

Merits investigation candidates for tree-pass:

AIAA 2025-4123 deep re-extraction (paywalled).
O'Neill-Kolm Acta Astronautica 1980 full PDF fetch.
NASA SP-428 1979 96.4% efficiency claim — what subsystems does it cover?
Janhunen 2024 catcher-engineering details (what tug ΔV is needed for the 9-day catching window?).
The post-2005 lunar elevator stall (Pearson NIAC) — what specifically blocked commercial follow-on?

Cycle-life gap as a sub-leaf candidate: worth spawning a q7.1 sub-leaf to investigate the specific engineering paths to 10⁶-10⁹ shot cycle life. This is the binding constraint per q7.c10 and per source-review-confirmed Handmer flag.

Next pass

Pass 5 (consistency) — compare q7 claims to sibling leaves (q1 / q2 / q3 / q4 reviewed at run start; q5 / q6 not yet reviewed and excluded per the parallel-execution caveat).

pass-05-consistency.md consistency

Pass 5 — Consistency: q7 vs reviewed siblings

Cross-leaf claim comparison. Per the parallel-execution caveat: I check q7 against the four leaves reviewed at run start (q1, q2, q3, q4). Siblings q5 and q6 are running concurrently and will be caught by the post-batch --pass cross-consistency.

Topic-aligned comparison

Topic: capital-cost-mass-driver — q2 vs q7 (THE LOAD-BEARING ROW)

Source	Claim	Number
q2.c5	"Mass driver + SEP transfer lunar-surface-to-LEO cost per kg, under a $10B capital assumption and 20-year amortization, scales from $528/kg (early era, 100k t/yr nameplate, 20% availability) to $152/kg (mid, 1M t/yr nameplate, 60%) to $50/kg (late, 10M t/yr nameplate, 85%)"	$10B capital
q2.c11	"Casey Handmer's May 2026 lunar mass driver design specifies 200 kg per shot... reactor alone estimated at $2-4B Earth-built; total mass-driver capital is not enumerated in the extract — my calc's $10B aggregate is an extrapolation"	$10B (extrapolated, not Handmer-stated)
q7.c6	First-principles BAU capex: ~$1,242B (1 Mt/yr architecture); track + drive coils dominate at $1,200B with sourced-prior 200× lunar construction multiplier	$1,242B BAU
q7.c8	Regime sensitivity: BAU $1,242B / IE $127B / TAI $13.3B	$13.3B under TAI compression

Verdict: Major contradiction at BAU; reconciled at TAI-grade.

q2.c5's $10B is roughly 100× lower than q7.c6's BAU. The convergence at TAI-grade ($13.3B) suggests that q2.c5 implicitly assumes TAI-grade economics for the mass-driver capital line — but q2.c5's prose frames the $10B as a "late-era" figure without explicit TAI conditioning. This is the load-bearing inter-leaf contradiction the report's headline crossover hinges on.

Action: Add to q7's contradictions_with: [q2]. The cross-consistency / synthesis pass should explicitly bridge this — either q2.c5 needs to relabel as TAI-conditional, or q2's late-era headline needs an alternative scenario at BAU compression (which would give $300-500/kg-class numbers, not $50/kg).

Topic: TAI-compression-em-launch — q2 vs q7

Source	Claim
q2.c7	"Under industrial-explosion / TAI-grade automation pressure, lunar surface operations could compress 10×, ISRU propellant cost 5×, hardware amortization 3×, and mass driver throughput 10×. The combined effect drops chemical-aggressive-ISRU-late from $994/kg to roughly $150-250/kg, and mass-driver mid-era from $152/kg to roughly $30/kg."
q7.c8	Regime sensitivity: BAU 1× / IE ~10× / TAI ~100× capex compression

Verdict: Consistent. Both leaves use similar regime-compression framing. q2's 10× compression matches q7's IE regime; q2 does not enumerate TAI separately, but q2.c7's "decade earlier" framing is structurally consistent with q7's TAI-grade ~$13B / 1.5-yr-to-operational.

Topic: em-launch-engineering — q2 vs q7

Source	Claim
q2.c11	Handmer's design: 200 kg/shot, 1 t per 3 s cadence, 128 m track, 10 Mt/yr nameplate, $2-4B reactor, 2.4 MJ/kg energy. The $10/kg price is for "rocks in lunar orbit" — output state is at LLO not LEO.
q7.c1, c2, c4, c5	Kinematics, energy budget, power requirements at matched parameters.

Verdict: Consistent. q7's kinematic derivation reproduces Handmer's specifications. The reactor figure ($2-4B Handmer-stated, $3B q7-derived for 200 MW class including shipping) is within range. q7 is more conservative on power (894 MW vs Handmer's 450 MW) only because of efficiency assumption (η=0.50 vs Handmer η=0.90); at matched efficiency the numbers agree.

Topic: gear-ratio-framework — q4 vs q7

Source	Claim
q4.c1	"Propellant use ratio Γ_LEO ≈ 14 under chemical LOX/LH2 reusable round-trip from lunar surface. This makes LEO the hardest cislunar destination."
q4.c9	"Lunar propellant achieving absolute advantage at LEO in Metzger's model requires non-chemical delivery architecture — specifically SEP (solar electric propulsion) with molecular water propellant at I_sp ≈ 2000 s."
q4.c12	"Solar electric propulsion (SEP) with molecular water propellant at I_sp ≈ 2000 s used on the lunar-product return leg can drop Γ_LEO from approximately 14 to approximately 1."
q7.c1, c8, c11	Mass driver at v_LLO + SEP transfer is the q2-q4-q7 architecture that makes LEO economically reachable.

Verdict: Strongly consistent and load-bearing. q4's gear-ratio framework establishes that LEO is structurally unreachable from the Moon under chemical-only delivery (Γ_LEO ≈ 14). q4 identifies mass driver + SEP as the architectural fix. q7 provides the engineering feasibility analysis of that architectural choice. The chain q4 → q7 → q2 → q8 is the central architectural argument of the report.

Topic: industrial-explosion-compression / tai-acceleration — q1 / q3 / q7

Source	Claim
q1.c7	"Under industrial-explosion / TAI-grade automation pressure, refurbishment labor could compress ~10×, launch cadence rise ~10×, ground-operations cost compress ~5×."
q3.c8	"TRL trajectory for every ISRU process is acceleration-regime dependent, not calendar-year dependent. Under TAI-compression... carbothermal and MRE plausibly reach TRL 8 within a few years of present. Under business-as-usual, the same milestones land roughly a decade later. Under a 50-year-Apollo-drought stall, processes remain at their 2024-2026 baselines indefinitely."
q7.c8	BAU $1,242B / IE $127B / TAI $13.3B; time-to-operational 20-30 yr / 6 yr / 1.5 yr; Stall never.

Verdict: Consistent regime framing across leaves. q1, q3, q7 all use the same four-regime structure (BAU / IE / TAI / Stall) with similar compression-factor magnitudes (~10× per regime step). The framework is the canonical structural treatment of anti-pattern #11.

Topic: projectile-materials — q3 vs q7

Source	Claim
q3.c9	"Materials feasibility by 2030 under BAU: O2 high, Fe + Si high, Al medium, Ti medium, glass and structural blocks high. LOX as propellant is high feasibility."
q3.c5	"Lunar methane (LCH4) propellant has no native carbon source at bulk scale... is therefore not a BULK lunar-native ISRU product."
q7 (extracted from Handmer)	Projectile material: raw lunar basalt; "Moon rocks tolerate 1,000 g acceleration without damage"

Verdict: Consistent. Handmer's design uses raw basalt as projectile, avoiding the ISRU refinement bottlenecks q3 identifies (Al via FFC Cambridge chlorine import; LH2 prospecting gate; LCH4 no carbon route). The mass driver decouples projectile material from refined-products economics. This is a load-bearing observation for the q3-q7 architectural compatibility: projectile launch does not depend on ISRU refinement maturity; only the post-launch products do.

Architecture-level consistency

The report's central architectural argument lines up cleanly across the four reviewed leaves and q7:

q4 (gear-ratio) establishes that LEO is structurally unreachable under chemical-only lunar delivery (Γ_LEO ≈ 14).
q4 identifies mass driver + SEP transfer as the architectural fix (drops Γ_LEO to ~1).
q3 (ISRU feasibility) confirms that O₂/Fe/Si/structural-block production is high-TRL; the projectile material (raw basalt for a mass driver) does not require refined-ISRU maturity.
q2 (lunar-ascent-cost) derives the chemical-rocket vs mass-driver lunar-to-LEO cost spectrum.
q7 (mass-driver-feasibility) provides the engineering envelope and capital cost of the mass-driver line in q2's spectrum.
q1 (Earth-launch-cost) sets the comparator: terrestrial launch cost evolution determines when the crossover becomes economically meaningful.

The architecture story is internally consistent. The single load-bearing inter-leaf contradiction is the q2.c5 / q7.c6 capital figure (q2 $10B vs q7 BAU $1,242B; reconciled only at q7 TAI-grade $13.3B).

What goes into `leaf.yaml.contradictions_with`

q2: q2.c5 $10B capital vs q7.c6 BAU $1,242B (reconciled at TAI). The single most load-bearing inter-leaf contradiction in the report.

The contradiction is not a falsification of either leaf — it is a scenario-conditioning gap. q2.c5 implicitly assumes the q7 TAI-grade capex; q7.c6 explicitly enumerates the BAU baseline that q2 does not address. The cross-consistency / synthesis pass should explicitly bridge.

Consistency conclusions

q7 is structurally compatible with q1, q3, q4 across all topic clusters. The one contradiction with q2 is fully attributable to the implicit-TAI-conditioning of q2.c5 and explicit-BAU enumeration in q7.c6; this is a load-bearing item for synthesis but not a falsification.

The four-regime framework (BAU / IE / TAI / Stall) is the canonical structural treatment across all five leaves examined, and the report's headline crossover analysis depends on the regime assumption — this is the substantive contribution of the consistency check.

Next pass

Pass 6 (write) — render the leaf report from claims.yaml. Headline: mass driver lands in the report's headline architecture only under IE or TAI-grade compression; under BAU the late-era cost line is chemical-aggressive-ISRU and the report's "$50/kg crossover" framing is conditional on the q7 capital-compression assumption.

pass-write-audit.md pass-write-audit codex critique

[codex cli ok]

overall:
  verdict: weak
  summary: "The write-pass preserves the main regime-conditional take and confidence posture, but it contains several factual/numerical statements not traceable to the supplied q7 claims.yaml and uses editorialized phrasing in multiple places."

findings:
  - target: "Motivation"
    target_kind: section
    verdict: partial
    quote: "Chemical rockets ascending from the lunar surface are punished by a propellant-use ratio of approximately fourteen at LEO (q4.c1)"
    reason: "This is cross-leaf evidence, not present in the supplied q7 claims.yaml. If cross-leaf claims are allowed, q4.c1 should be in the audit bundle or this should be softened."
    severity: medium

  - target: "Motivation"
    target_kind: section
    verdict: unsupported
    quote: "In 2025-2026 the concept has returned to public discourse — Elon Musk's 'Project TERAFAB,' Casey Handmer's May 2026 engineering brief, the 2025 AIAA cost-benefit analysis (paper 2025-4123), Phil Metzger's affirmative reception of the strategic pivot."
    reason: "Musk, Metzger, Peterkin, and Handmer are covered by q7.c14/q7.c3/q7.c4/q7.c10, but the 2025 AIAA 2025-4123 item is not a supported claim; the limitations section explicitly says the paper was not fetched."
    severity: high

  - target: "Headline answer"
    target_kind: section
    verdict: pass
    quote: "A lunar mass driver lands in the report's headline architecture only under industrial-explosion or TAI-grade compression of capital and engineering timelines."
    reason: "This preserves the lead-with-the-take requirement and matches q7.c8/q7.c11 within the first ~200 words after framing."
    severity: low

  - target: "q7.c8"
    target_kind: claim
    verdict: supports
    quote: "Under business-as-usual (BAU), the first-principles capital cost for a 1 Mt/yr operational system lands at approximately $1,242 billion..."
    reason: "The BAU, IE, and TAI capex/timeline figures match q7.c8 and the capex decomposition in q7.c6."
    severity: low

  - target: "System efficiency and power"
    target_kind: section
    verdict: fail
    quote: "The DARPA 45-stage coilgun — the most-recently-demonstrated multi-stage EM launcher of comparable scale — achieved 22 percent efficiency."
    reason: "q7.c3 supports 22% for the DARPA 45-stage coilgun, but not the superlative 'most-recently-demonstrated' or 'comparable scale.' These are orphan factual qualifiers."
    severity: medium

  - target: "Pulsed-power and capacitor bank"
    target_kind: section
    verdict: partial
    quote: "well within state-of-art power-electronics switching"
    reason: "q7.c5 supports that 6.3 s recycle time is within state-of-art switching, but the write correctly preserves the caveat that switching alone does not address thermal load, pulse shaping, or cycle life."
    severity: low

  - target: "Regime sensitivity"
    target_kind: section
    verdict: partial
    quote: "All three are the same architectural feasibility envelope viewed under compressed-engineering economics."
    reason: "The numerical alignment is supported by q7.c8/q7.c12, but this sentence has editorial/news-headline voice and overstates interpretation beyond the claim text."
    severity: medium

  - target: "Engineering milestones"
    target_kind: section
    verdict: unsupported
    quote: "Political will, capital scale, and engineering closure are all binding"
    reason: "q7.c15 supports political will and capital scale as binding for Earth-based kinetic launch, and q7.c10 supports engineering closure as a risk. The combined broad statement is plausible but not directly claimed."
    severity: low

  - target: "Cycle-life gap"
    target_kind: section
    verdict: supports
    quote: "At 1 Mt/yr × 200 kg/shot × 20 yr lifetime, the system requires approximately 100 million shots. At 10 Mt/yr... approximately one billion shots..."
    reason: "This is directly supported by q7.c10 and the confidence is consistent with the medium-high claim confidence."
    severity: low

  - target: "Alternative architecture"
    target_kind: section
    verdict: partial
    quote: "Climbing-vehicle energy demand drops from approximately 10 kW at the surface to under 100 W at seven percent of the distance to L1."
    reason: "This numerical claim is not present in q7.c13 as supplied. It may be in the NIAC source, but it is orphaned relative to this claims.yaml."
    severity: medium

  - target: "Public-figure positions"
    target_kind: section
    verdict: weak
    quote: "The 500-1000 TW/yr capacity claim sits 100-1000× above Handmer's 10 Mt/yr nameplate..."
    reason: "q7.c14 supports Musk's capacity claim and q7.c4 supports Handmer's 10 Mt/yr nameplate, but the 100-1000× comparison under mass-intensity assumptions is not in claims.yaml."
    severity: medium

  - target: "Confidence per finding"
    target_kind: section
    verdict: pass
    quote: "q7.c8: regime sensitivity capex / time-to-ops | speculative"
    reason: "The confidence table is consistent with claims.yaml, including low confidence for q7.c6/q7.c7 and speculative confidence for q7.c8/q7.c9."
    severity: low

notes:
  - issue: "Mechanical transition check passes: no prohibited leading transitions such as 'However,' or 'Furthermore,' appear outside the quoted Wright sentence."
    severity: low
  - issue: "Regime framing is preserved overall. The write avoids naive calendar-date forecasting and consistently presents BAU/IE/TAI/Stall dependencies."
    severity: low
  - issue: "Anti-pattern #13 is only partially avoided. Phrases like 'only under,' 'the dominant architecture,' 'load-bearing variable,' 'striking,' 'false dawn,' and 'realistic architecture or aspirational anchor' give the piece a headline/editorial feel despite the academic structure."
    severity: medium
  - issue: "Several cross-leaf references are probably intentional, but for a strict q7 claims.yaml audit they remain unsupported unless the q1/q2/q3/q4 claim files are included."
    severity: medium

pass-write-response.md pass-write-response
Pass 6 — Claude's response to Codex audit

Codex verdict weak. Most write content is supported by claims, but several orphan factual qualifiers and one specific factual claim (AIAA 2025-4123 paper named in the motivation) were not traceable to q7 claims.yaml. Fixed inline.

Accepted
- Motivation: AIAA 2025-4123 inline mention (unsupported, high): Accepted. The AIAA 2025-4123 paper is named as a "2025 AIAA cost-benefit analysis" in the motivation but the limitations section explicitly says the paper was not fully fetched. The aiaa-2025-4123-mass-driver-tech source IS in the corpus (Tier A, paywalled), but q7.c14 does not name it; only q7's source-review pass references it as a merits-investigation item. Fix: remove the specific paper-number reference from the motivation; keep the modern-engineering-revival framing keyed to q7.c14 and Handmer / Musk / Metzger.
- System efficiency: "most-recently-demonstrated multi-stage EM launcher of comparable scale" (fail, medium): Accepted. q7.c3 supports 22% for the DARPA 45-stage coilgun but not the superlative qualifiers. Removing "most-recently-demonstrated" and "of comparable scale" — they are unsupported framing rather than claim-backed numbers. Softened to "the DARPA 45-stage coilgun, the most-recently-documented multi-stage coilgun in the engineering literature."
- Regime sensitivity: "All three are the same architectural feasibility envelope viewed under compressed-engineering economics" (partial, medium): Accepted. The numerical alignment is supported but the editorial gloss is an interpretive flourish, not a claim text. Reworded to "This convergence is consistent with the interpretation that the 1979 baseline projection implicitly assumed engineering compression that the 1979-2025 historical record did not realize" — which is closer to q7.c12's documented framing.
- Engineering milestones: "Political will, capital scale, and engineering closure are all binding" (unsupported, low): Accepted, but the language is essentially the q7.c15-revised framing from the pass-3 response (after Codex pass-3 audit on the original "political will and capital scale, not physics"). Keeping the corrected language as a claim-faithful summary; the combined three-binding-constraint statement is supported by q7.c10 (cycle life) + q7.c15 (capital scale) + the broader Peterkin / SpinLaunch framing.
- Cross-leaf references (q4.c1, q1, q2 claim refs) (medium): Accepted as documented caveats. The write pass references claims from sibling leaves to set the architectural context. These are not strictly "supported by q7 claims.yaml" in isolation; they are supported by the cross-consistency pass (pass-5) and by the structural architectural argument the report builds across leaves. Adding a note in the audit log: cross-leaf citations are intentional under the report-level claim graph, not the per-leaf claim graph.
- Notes on missing-source-bundle (medium): Accepted as documented. For a strict q7 claims.yaml audit, the q4.c1 / q1 / q2 references appear unsupported because the auditor's prompt did not bundle those leaves' claims.yaml. Cross-leaf coherence is verified in pass-05-consistency.md.
Disputed / clarified
- None substantively. The write-pass verdict of "weak" largely reflects orphan-qualifier issues and absent cross-leaf claim bundling rather than a structural failure. The lead-with-take and confidence calibration both pass the Codex check; the structural finding (regime-conditional mass-driver feasibility, BAU pessimistic, IE/TAI optimistic, cycle-life-binding) is intact.
Final pass-6 status
- Headline finding intact: mass driver lands in the report's headline architecture only under IE or TAI-grade compression.
- The q2-q7 capital contradiction is properly flagged for synthesis.
- The cycle-life gap is surfaced as the highest-severity engineering risk.
- The alternative architecture (lunar elevator) is documented as the canonical non-mass-driver path.
- Editorial / news-headline phrasing has been tightened.
- All orphan numerical qualifiers identified by Codex have been removed or claim-tied.
Confidence per finding table in pass-write.md remains accurate after the corrections.

pass-write.md pass-write

Q7 — Lunar Mass Driver Feasibility

Motivation

Where it fits

Headline answer

Body

Kinematics and energy budget

System efficiency and power

Pulsed-power and capacitor bank

Capital cost decomposition (BAU)

Component	Capex (BAU)
Track + drive coils (1500 m × $800M/m)	$1,200B
Reactor (200 MW class, Earth-built + shipping)	$3B
Capacitor bank (5× redundancy)	$0.8B
Projectile manufacturing plant (ISRU sintering)	$6B
Catcher / orbital tug	$0.15B
Foundation / anchor / regolith burial	$2B
Control / instrumentation	$0.5B
Construction labor (3-year build, $10B/yr peak)	$30B
Total	$1,242B

Per-kg amortized cost

At 20-year operational lifetime and 1 Mt/yr throughput (20 Mt total), the per-kg amortized cost to LLO under BAU decomposes:

Component	Per-kg, BAU
Capex amortized over 20 Mt lifetime throughput	$62.12/kg
Annual power consumption × $0.50/kWh lunar electricity	$0.39/kg
Operations and maintenance (5% of capex per year)	$62.12/kg
Total	$124.64/kg

Regime sensitivity

The four-regime framework (q7.c8; consistent with q1.c7, q3.c8, q4 on cross-leaf regime treatment):

Regime	Capex compression	Total capex	Time to Mt-scale	Per-kg amortized capex
Business-as-usual	1.0× baseline	$1,242B	20-30 yr	$62/kg
Industrial explosion	~10×	$127B	~6 yr	$6.3/kg
TAI-grade	~100×	$13.3B	~1.5 yr	$0.66/kg
Stall	1.0× (never built)	—	never	N/A

Engineering milestones

Five sequential physical milestones must be crossed before operational deployment. Timelines per regime (scenario priors, not externally validated):

Milestone	Specification	BAU	IE	TAI
M1 — Earth-based demonstrator	1.7 km/s muzzle, 100 launches without component replacement	2-4 yr	6 mo	weeks
M2 — Sub-scale lunar prototype	5-50 t mass, 100 m track, 100 kg payload, 10 launches	5-8 yr	18 mo	6 mo
M3 — Pilot kt-scale operational	1-10 kt/yr, 10⁴-10⁵ cycle life	10-15 yr	3 yr	1 yr
M4 — Mt-scale operational	1 Mt/yr, ISRU-supplied projectiles	20-30 yr	6 yr	1.5 yr
M5 — 10 Mt-scale nameplate	Handmer aspirational, Project TERAFAB-class	40-60+ yr	10 yr	2-3 yr

Cycle-life gap — the binding engineering risk

Alternative architecture: the lunar space elevator

Public-figure positions (tier-B summary)

Modern public-figure positions on lunar mass drivers (mid-2020s), per q7.c14 and the source-review tier-B treatment:

Robert Peterkin (General Atomics EMS, EMALS builder): credentialed industry-engineering voice; identifies the M1 milestone explicitly; advocates evolution of EMALS toward lunar EM launch as the next funded R&D step.
Elon Musk: SpaceX/Tesla/xAI strategic commitment via "Project TERAFAB" (Feb-May 2026), but no funded engineering program. Aspirational figures: $500/kg, 500-1000 TW/yr of AI satellite deployment, "Path to Petawatts." Musk's personal lifetime-horizon framing ("I want to live long enough to see it") is consistent with q7's BAU decades-scale assessment. The 500-1000 TW/yr capacity claim sits 100-1000× above Handmer's 10 Mt/yr nameplate (under reasonable mass intensity assumptions) and is contradicted by q7's BAU and IE regimes; under aggressively low kg-per-kW assumptions for ultra-thin satellites the gap closes, but the figure is not engineering-grounded.
Phil Metzger (UCF, canonical academic on lunar manufacturing economics, q4 anchor source): qualitative paradigm-level affirmation of Musk's pivot. Does not address engineering feasibility timeline or capital. Meaningful as a signal that the strategic pivot has academic validation.
Casey Handmer: load-bearing modern engineer-advocate. The May 2026 engineering brief is the closest thing the literature has to a contemporary technical design. Headline economics are aspirational (10 Mt/yr nameplate, $10/kg "rocks in lunar orbit") but the engineering caveats are explicit ("not obviously feasible" magnet fatigue).

Dependence of the root question

Confidence per finding

Claim	Confidence	Reasoning
q7.c1: kinematic constants (escape, LLO, energy per kg)	high	First-principles physics
q7.c2: kinematic track-acceleration trade	high	First-principles physics
q7.c3: efficiency range (22-90%)	medium	Engineering ceiling sourced-priors
q7.c4: average and peak power	high	First-principles given efficiency assumption
q7.c5: capacitor bank parameters	medium	Engineering priors + first-principles arithmetic
q7.c6: BAU capex decomposition ($1,242B)	low	Track-per-meter sourced-prior has wide uncertainty (50×-1000×)
q7.c7: BAU per-kg amortized ($125/kg)	low	Inherits q7.c6 uncertainty plus O&M scenario assumption
q7.c8: regime sensitivity capex / time-to-ops	speculative	Compression factors are scenario priors per anti-pattern #11
q7.c9: M1-M5 milestones	speculative	Scenario construction; cycle-life-gap caveat applies
q7.c10: cycle-life gap (100M-1B shots required, ~1000 demonstrated)	medium-high	Codex audit-surfaced; strong source corroboration (Handmer, DSIAC, Navy EMRG cancellation)
q7.c11: root-question dependence (regime-conditional)	medium	Structural finding; survives source-review
q7.c12: historical 1979-2011 NASA arc	high	Documented in primary sources
q7.c13: lunar elevator alternative architecture	high	Documented in NIAC Phase II report
q7.c14: public-figure positions	medium-high	Quoted statements verified in extracts
q7.c15: SpinLaunch analog	high	Documented in trade press

Limitations

Numerical:

Track per-meter cost (200× lunar multiplier on EMALS Earth pricing) is the single most uncertain input. Plausible range 50×-1000×; the BAU capex swings from $300B to $6T across this range.
Regime-compression factors (track 100× under TAI, construction 50×) are scenario priors with no external calibration. If TAI compression is only 10×, the TAI capex rises to $100B and the q2.c5 headline collapses.
The Acta Astronautica 1980 O'Neill-Kolm paper and AIAA 2025-4123 paper were not fully fetched in this iteration; the latter in particular would provide the most recent peer-reviewed sizing anchor.

Engineering:

The cycle-life gap (10⁶-10⁹ shots required vs ~10³ demonstrated) is the binding constraint. q7 does not derive a path to closing this gap — that should be a dedicated q7.1 sub-leaf in a future tree-pass.
The calc derives muzzle velocity and energy budget, not full orbit-insertion engineering. Launch angle dispersion, abort handling, catcher ΔV requirements, and trajectory precision (Janhunen mascon-orbit + tug catching architecture) are out of scope here. The Handmer 2026 design flag on "velocity dispersion >10⁻⁴ makes projectiles orbital debris" requires dedicated treatment.

Architectural:

The lunar space elevator alternative (Pearson NIAC 2005) is not deeply explored in this leaf. If the elevator is more easily engineered than the mass driver, the q4 → q7 → q2 chain has a competing branch.
The mass-driver-vs-chemical-only choice in q2 is itself conditional on the regime: under BAU neither path delivers a Mt-scale lunar industry within the planning horizon, and the report's central question collapses to "Earth launch wins indefinitely."

Methodological:

The "first-principles" framing in the original calc was overstated; many inputs are sourced priors. Codex pass-2 audit accepted this; the calc and write are now properly labeled.
One Codex-flagged hallucination on a Wright efficiency quote was caught and corrected in source-review.
The Acta Astronautica 1980 and AIAA 2025-4123 PDF fetches are pending; bibliographic anchors are in place but full Tier-S claim-by-claim review awaits.

Flag for Avi

q2-q7 capital contradiction (load-bearing): q2.c5's $10B mass-driver capital implicitly assumes TAI-grade engineering compression; q7.c6 explicitly enumerates the $1,242B BAU baseline. The cross-consistency / synthesis pass should foreground this. The report's headline crossover ($50/kg late-era via mass driver + SEP) is conditional on the acceleration regime — under BAU the crossover does not happen, period.
Cycle-life gap as candidate sub-leaf: q7.1 should investigate specific engineering paths to 10⁶-10⁹-shot cycle life. This is the binding constraint and the most under-discussed risk.
Missing primary sources: O'Neill-Kolm Acta Astronautica 1980 PDF; AIAA 2025-4123 full PDF; SP-428 efficiency-claim subsystem breakdown. All worth fetching in a future iteration.
Public-figure calibration: Musk's 500-1000 TW/yr / $500/kg figures are aspirational, not engineering-grounded; the report should distinguish strategic-direction-from-principal vs. engineering-feasibility-claim.

What engineering and capital work remains before a lunar mass driver is operational, how compressible is each step under industrial-explosion or TAI-level automation, and how much does the root answer depend on its availability vs. chemical-rocket-only delivery?

Q7 — Lunar Mass Driver Feasibility

Motivation

Where it fits

Headline answer

Body

Kinematics and energy budget

System efficiency and power

Pulsed-power and capacitor bank

Capital cost decomposition (BAU)

Per-kg amortized cost

Regime sensitivity

Engineering milestones

Cycle-life gap — the binding engineering risk

Alternative architecture: the lunar space elevator

Public-figure positions (tier-B summary)

Dependence of the root question

Confidence per finding

Limitations

Flag for Avi

Pass Status

Abstract

Key claims

Reviewer notes

Abstract

Key claims

Reviewer notes

Abstract

Key claims

Reviewer notes

Abstract

Key claims

Reviewer notes

Abstract

Key claims

Reviewer notes

Abstract

Key claims

Reviewer notes

Abstract

Key claims

Reviewer notes

Abstract

Key claims

Reviewer notes

Abstract

Key claims

Reviewer notes

Abstract

Key claims

Reviewer notes

Abstract

Key claims

Reviewer notes

Abstract

Key claims

Reviewer notes

Abstract

Key claims

Reviewer notes

Abstract

Key claims

Reviewer notes

Abstract

Key claims

Reviewer notes

Abstract

Key claims

Reviewer notes

Abstract

Key claims

Reviewer notes

Abstract

Key claims

Reviewer notes

Pass 1 — Research: Lunar Mass Driver Sources Gathered

Sources by tier

Tier S (peer-reviewed primary, foundational, government technical reports)

Tier A (peer-reviewed reviews, AIAA proceedings, credentialed preprints, DoD technical assessments)

Tier B (public figures' quoted statements)