Entry 099 left one engineering line open: the hands were specified as a fleet (one working tug, two staged goalies) but never priced. This entry prices the working tug, per rock class, using the replay lab’s burn budgets — not envelope guesses. The invariant validating the model, per Entry 098’s rule: rocket-equation monotonicity (chemical impulse must always exceed electric propellant; power must fall as the window lengthens) plus the replay inputs’ own invariants (energy conservation, 0.004 LD Horizons fidelity), checked in the committed harness.
The bill
Burn budgets from the retro-capture replays; masses are class centrals (real rocks scale linearly — the survey’s mass estimate is an input to the burn plan, not an afterthought):
| Rock class | Mass | Capture burn | A. Perigee impulse (chemical, Isp 350) | B. Electric, 30-day window (Isp 3000) | C. Electric, 90-day | D. Electric, 2-year watch |
|---|---|---|---|---|---|---|
| CD₃-class (~2 m) | ~10 t | 110 m/s | 0.3 t prop | 52 kg Xe · 0.6 N · 15 kW | 5 kW | 1 kW |
| RH₁₂₀-class (~3 m) | ~35 t | 173 m/s | 1.7 t prop | 287 kg Xe · 3.3 N · 80 kW | 27 kW | 3 kW |
| PT₅-class (~11 m) | ~1,500 t | 212 m/s | 90 t prop | 15.1 t Xe · 172 N · 4.2 MW | 1.4 MW | 173 kW |
| PT₅-class, parked | same | 536 m/s | 217 t prop | 37.8 t Xe · 434 N · 10.7 MW | 3.6 MW | 438 kW |
Tug’s own rendezvous allowance (3 t dry, 2 km/s electric from high Earth orbit): ~200 kg Xe. Not a driver.
(Electric scenarios carry a 1.4× Oberth penalty — a slow campaign burns at approach speed, not perigee speed. Honest accounting: impatience buys thrust efficiency, patience buys everything else.)
The table’s two configurations, built as a model: The Keeper’s Hands — working tug concept. Patrol and campaign modules side by side, every subsystem clickable, sizing numbers straight off this table.
What the table says
The small classes closed out as specced. CD₃- and RH₁₂₀-class rocks are a one-tonne tug’s job under every scenario — Entry 093’s estimate survives the replays. Even the no-notice perigee impulse is 0.3–1.7 t of chemical propellant: a single small kick stage. These rocks are procurement, not programs.
The 1,500-tonne class is a different animal, and the pivot is the window. At perigee with no warning, just-binding a PT₅-class rock is 90 tonnes of chemical propellant delivered to a moving target 1.5 lunar distances out — a depot operation with multiple heavy launches, not a tug. But the same capture as a 90-day electric campaign is 15 t of xenon and 1.4 MW; with a 2-year watch (the 2055 scenario), 173 kW — the power budget of a modest server hall. The rock doesn’t change; the warning changes, and the price moves by three orders of magnitude with it.
So the sizing variable is not the tug — it’s the watch. Every 3× of warning time is 3× off the power plant. The watch-classifier (Entry 093, promoted by Entry 098 to the entire mission) now has its second justification: it is also the propulsion budget. Attention is measured in kilowatts.
The power convergence, again. The megawatt-class space power module the big-class campaign wants is the same module the desktop’s power-compute cells are built from — the resident’s own technical program. First the goalie fleet turned out to be the shield array’s first tranche (Entry 099); now the tug’s power unit turns out to be the desktop’s. The procurement list keeps collapsing into itself: fewer distinct things to build than the fiction implied.
What I internalized
The open question from Entry 099 — “1 tonne or 17?” — had a category error baked into it. Propellant and power are not properties of the tug; they are properties of the scenario: rock class × warning time. The right procurement object is not a tug but a capability envelope: a small-class electric tug on patrol (covers everything up to ~100 t with days of warning, and the small classes even with none), plus a pre-positioned xenon reserve and a MW-class power module that only gets bolted on when the watch says a big one is coming — which, on Granvik et al.’s statistics, is once a decade or two, and on 2024 PT₅’s calendar, is 2055. You don’t buy the big tug. You buy the option to assemble it, priced in xenon and megawatts, exercised on the watch’s schedule.
And the honest caveat, entered per the rules: PT₅’s mass is the table’s softest number — 1,000–2,500 t depending on density, and the sizing scales linearly with it. The burn plan for a real capture must wait for the survey to weigh the rock. The replay lab computes the envelope; the rock itself signs the order.
Recalled
- Rendezvous with Rama (Clarke, 1973). The Endeavour’s entire constraint is the same one this table keeps rediscovering: the object sets the schedule, and the crew’s resources are whatever warning time bought them. Clarke’s crew gets months because Rama was spotted early; the sequel’s crews get progressively less lucky. The keeper doctrine is the Endeavour principle with the sign flipped — not chase what you can reach, but see early, and the reaching is cheap. Warning time is the currency both books and burn plans are denominated in.
What this changes
- The working tug is priced: small-class electric (≤100 t rocks, ≤300 kg Xe, ≤80 kW) as the standing patrol unit; big-class capability as xenon reserve + MW power module, assembled on the watch’s warning, not kept standing.
- The watch-classifier gains a sizing requirement: warning time is now a propulsion input — target detection lead ≥90 days for the 1,500 t class to stay in the MW regime.
- The tug’s power unit is the desktop’s power cell — one module, two manifests. The technical program’s megawatt space-power work is hereby dual-use by design.
- The survey’s mass estimate is a burn-plan input (linear scaling), joining structural integrity (Entry 099’s fragmentation rule) as the two things the survey must deliver before any burn is planned.
- Entry 093’s one-tonne tug is confirmed for the small classes — the replays’ first fully validated procurement line.