Cargo WindowAdmin view. The team's notes for presenting Cargo Window. Users of the console never need these pages.
Beyond 2030 · Our idea
Miss the ship to Mars and you wait 26 months.
Cargo Window plans what flies to the first bases on the Moon and Mars. It measures every manifest in the days a base goes without, never lets a bid outrank survival, and works across a 44-minute signal delay.
1 · The problem
Deciding what flies when the next ship is two years away
Every rocket to an off-world base has a fixed mass and volume, and in 2035 every flight has more cargo waiting than room. Someone has to choose what goes. On Earth that choice is cheap to get wrong, because the next truck leaves tomorrow. To Mars, the next ship leaves in about 26 months, so the manifest decides who goes without for two years.
Today manifests for a handful of government missions are built by hand, in spreadsheets and meetings. That stops working once several bases and dozens of companies share each flight. Three things break at once:
- Fairness: there are no shared, published rules for who gets bumped, so money or politics decides.
- Foresight: nobody sees that a bumped load leaves a base short until the ship lands without it.
- Requests: crews ask over a delayed link, and every clarifying question to Mars costs up to 44 minutes.
2 · Who it affects
Five groups, one rocket
3 · Why it matters
A bumped load can be a health risk for two years
- The stakes are lives, not delays. A CO₂ scrubber cartridge that misses the Mars window leaves a crew short of air-cleaning spares until the next one.
- Every kilogram is expensive. Delivering cargo to the Moon costs hundreds of thousands of dollars per kilogram today. Wasted room is wasted money.
- Sharing needs trust. Agencies and companies will only share flights if the rules are fair, published and the same for everyone.
- Bases can't fix it alone. By the time a crew notices a part is missing, the window has closed.
4 · How it emerges
From today's lunar deliveries to full Mars windows
- 2018
FACTNASA starts buying cargo deliveries to the Moon from private companies.
- 2020s
FACTArtemis aims for a lasting human presence near the lunar south pole. Reusable heavy rockets push the cost per kilogram down.
- Early 2030s
ASSUMPTIONA permanent lunar base with flights every few weeks. Mining, research, media and tourism companies buy space next to agencies.
- 2035
ASSUMPTIONA first small crewed outpost on Mars, fully dependent on cargo from Earth and reachable only in windows about 26 months apart.
- Then
ASSUMPTIONEvery flight is oversubscribed, and the manifest becomes the most contested decision in the program.
Precedent
It already happens on Earth, on a small scale
The problem isn't invented. Places that are hard to reach already live by resupply windows, and they plan them by hand.
- Antarctica: the South Pole station gets no flights for about eight months of winter. What misses the last summer flight waits until spring.
- The space station: cargo flights are packed against priority lists, and when a flight is cut back, lower-priority items stay on the ground.
- Remote islands and research ships: supplies come with scheduled ships, and a missed sailing means weeks or months without.
Off-world bases take the same problem and make it bigger, slower to fix and less forgiving. The planning tools haven't caught up.
5 · What the solution does
Ask, decide, show the cost, agree
6 · Technologies
Each one earns its place
7 · Design and build
AI reads and writes. Rules decide. People approve.
- 01PEOPLECrew writes a requestIn plain words, under pressure, over a delayed link.
- 02AIClaude drafts the bookingLoad, class, mass, volume, need date. Missing facts become questions.
- 03PEOPLECargo officer approvesMissing numbers block it. Nothing is guessed.
- 04RULESPacker builds manifestsFlight by flight, in a fixed order, one reason per decision.
- 05RULESBoard shows the costArrival against need date, days a base goes short, fingerprint.
- 06AIClaude drafts the briefsThe officer edits and sends them to the base and each shipper.
for each base:
waiting = every booking for that base
for each flight to that base, in launch order:
sort waiting by class: life-critical, mission-critical, commercial
inside life and mission: earliest need date first
inside commercial: highest bid per kg first
for each load in that order:
if it fits the mass and volume left: put it on this flight
otherwise: it waits for the next flight
for each load:
arrives = launch + trip days
days short = arrives - need date (late if above zero)Originality
What we created that wasn't already there
A cargo planner that measures a manifest in the days people go without, where money can never outrank survival, that two planets can run separately and still agree on, and that gathers every missing fact in one message across a 44-minute delay.
The 2030+ test
Three questions every project must pass
Business and impact
Who pays and what changes
Next
After the hackathon
- Need dates from base data: forecast when scrubbers, filters and food run out from how fast the base uses them.
- A full copy at each base that syncs signed bookings when the link is up and compares fingerprints.
- Return cargo (samples back to Earth), hazardous goods rules and centre of gravity.
- Signed bookings and an audit log, so nobody can quietly change a priority.
- A priority policy the bases and operators sign in advance, so the classes themselves are agreed before any flight is full.
- Voice requests as recorded messages: crews in suits or mid-repair speak a request, it is transcribed at the base and sent as one message. Live voice agents don't fit a 44-minute round trip.
Example scenario. Bases, companies, people, prices and capacities are invented. Mars launch timing, trip length, signal delay and solar conjunction follow real orbital mechanics.