A civilization should look different as it learns what it can do.
Generated concept panorama. Aspirational art, not a game screenshot.
Your compute towers, lunar hoppers and colony trains, five more ideas, and a shared cargo interface to keep them coherent. The review suite follows MOON’s existing ceramic, graphite and gold art direction.
Ten original Blender models. Editable sources, animated GLBs, studio portraits and an illustrated document. Every proposed machine has one clear player action.
These are portraits of the actual review models. Interactive 3D is available on the hosted proposal page.
Make intelligence part of the skyline.
Manufacture → deliver → commission → cool → coordinate
The missing part, delivered before dawn.
Load → fuel → reserve arrival → fly → unload
A small bright destination on a very dark Moon.
Prepare → connect → reserve → receive
A train that gradually becomes a colony.
Pack a real bill → travel → unfold → become self-sufficient
Turn yesterday’s tailings into tomorrow’s road.
Suitable bulk → grade → compact → connect
The dark architecture that keeps minds alive.
Heat → transfer → reject → support more computation
At the edge of darkness, a reason to trade.
Ice → purify → separate → store → deliver
A failure becomes a cooperative mission.
Confirm need → bring spares → repair → verify
Many small instruments. One better picture.
Place receivers → measure → compare → improve the plan
One interface shared by the whole civilization.
One manifest across robots, depots, convoys and hoppers
Studio portraits rendered with CPU Cycles. Models are concept assets; their animation is not driven by game state.
A morning review for Corey and our lunar neighbors
Revision — A Slight Revision To Reality · September 8–9, 2026
Let intelligence change what we can do. Let physical work prove which plans deserve to spread.
This document preserves the discussion that scrolled out of the terminal: your three ideas, five more from me, the Starship research, the new asset suite, and where development actually stands. It is a proposal for the next visible layer of MOON. The models are review artifacts; the new vehicles and abilities are not yet installed in the live game.
The central idea is that the Moon's skyline should tell the story of its civilization. Early on, everything is low, scattered and improvised. Later, you can recognize a research center, a freight corridor or an expedition departure from far above. The final acceleration should come from actual working capacity spreading through actual infrastructure. Keep the first workshop somewhere in that enormous future.
The September 8 Stockpile release is complete. Settlement → Build → Stockpile → Prepare stockpile creates a 24 m footprint of prepared ground for 2 alloy plus real crew work, with 2,000 units of bulk storage. A hard refresh may be needed if an old browser tab does not show the card.
Robots move surplus raw bulk out of enclosed storage to available yards and reclaim useful inputs for processors. Consumers have priority. Compatible conveyors can move and reclaim the same eligible solids. Cargo remains physically present until collected; incoming deliveries reserve capacity. The mounds represent actual stored material. Components, alloy, fuel and fluids stay out of ordinary open stockpiles. Prepared ground has no standing Mind or power charge; the machines serving it still need their usual support.
The preceding logistics fix removed hidden per-product limits that could leave refineries blocked despite visible storage space. A retained comparison of the same captured colony over 1,200 simulation ticks produced 3.340 alloy with the old rules and 217.286 with the fixed rules. That is one regression fixture, not a general production multiplier, and it predates Stockpiles.
The live release preserved all six players and the existing world. It passed 325 CPU/API checks, staging checks, and public file verification. Runtime is 76ce1a2; its publication website revision was 3543eca. The manual, AI arrival kit, visual whitepaper and full September 8 journal were updated. Backups are verified on the Expansion drive. The workstation graphics fault remains unresolved; this review uses CPU rendering, and opening the page does not start WebGL.
PALISADE / Modular compute array. Dark ceramic cartridges slot into a structural frame. Silver cooling rails and restrained status lights make each layer readable. A crane raises a module, the connectors engage, its cooling circuit starts, then that level comes alive.
Manufactured cubes are potential. Installed cubes are infrastructure. Working computation requires power, heat rejection, connections and commissioning. Those three states should look different. A stack of unpowered cubes should not quietly count as a functioning mind.
Build vertically to conserve land, but concentrate power and cooling demand. Research can unlock stronger frames, denser modules and better cooling distribution. Different colonies develop different silhouettes: broad low arrays, stepped towers or huge terraced complexes. When a new supply line enables another floor, the lighting change should correspond to a real increase in capability.
Simple control: choose an expansion stage on an existing array. Show the material bill and projected support requirement. Queue the stage until those conditions are met. We do not need a placement puzzle for every cube.
The existing Computronium Foundry supplies the conceptual starting point; the stackable tower and its operating rules are proposed additions. The review model shows a commissioning lift. Its animation is illustrative, not a live logistics connection.
KESTREL / Cargo hopper and landing apron. A compact reusable lander with wide feet, pale ceramic panels, exposed service hardware, gold insulation, a removable cargo container and a small elevator. At night the apron lights switch on before a distant point of light becomes a landing vehicle.
Its job is urgent, valuable freight: electronics, parts, qualified reactor fuel, specialist materials or the missing component that restores a neighbor. Flying ordinary rock should be expensive enough that surface transport remains useful. Chemical propellant moves the craft; the nuclear fuel it may carry is cargo.
The player should select destination, cargo and amount. The game checks access, a free pad, payload, propellant and arrival capacity. A clear “waiting for propellant” or “destination pad occupied” replaces a silently stalled rocket. Recurring trips can come later through the same transport-order system.
Prepared surfaces matter: exhaust can disturb and eject regolith. NASA has researched methods of creating landing surfaces, including its instant landing pad concept. That research is inspiration, not proof that our fictional apron design would work as drawn.
The memorable first flight: one neighbor spends a little valuable propellant to deliver the part another colony needs to stay productive through the lunar night.
You were remembering the lunar Starship concept. The early HLS artwork is a useful reference for the tall white hull, wide legs, cargo elevator and high-mounted thruster arrangement. Here is the NASA selection article from April 2020, with SpaceX's concept illustration.
NASA's November 2024 HLS illustration gallery provides a later reference, including an elevator and a separate illustration of two Raptors braking after departure from low lunar orbit. These are dated design illustrations, not a guarantee of the current final vehicle design.
A NASA-hosted 2022 landing-pad study discusses high-mounted thrusters as producing a more benign ground environment than low aft engines. It supports the qualitative motivation. It does not make liftoff dust-free.
For our game, your staged departure would be excellent: service arm retracts, cargo is secured, the upper thruster collar lifts the vehicle clear, then the main engine takes over for the hop. Arrival reverses that visual sequence. That is our proposed animation, not a verified statement of SpaceX's current exact engine-switching procedure. Thrust, fuel and clearance balancing would need their own implementation pass. The review model already has the collar, main bell and animated cargo elevator; it does not yet contain a physically simulated flight or a staged plume animation.
WAYFARER / Seed expedition convoy. Begin with an articulated crawler convoy that can reach land before tracks exist. Each carriage carries something recognizable: folded Seed Base, reactor, Mind hardware, solar arrays, workshop or construction supplies.
The existing Departure Yard becomes the place where an expedition's real material bill turns into visible cargo. The train leaves slowly, its lights crossing the plain. At the destination, stabilizers lower, cranes unload, solar wings unfold and robots descend ramps. Over time, the convoy becomes the settlement.
Simple control: choose a prepared colony plan and a destination. Preview the claimed footprint, delivery route, material bill and support requirements. Reserve the plan's land. Build only as fast as deliveries and local extraction permit. Players can visit, pause, revise or help. No instant settlement and no free duplication.
Established busy corridors can later earn rail or magnetic transport. NASA's FLOAT research concept explores unpowered magnetic carriers on flexible electromagnetic track. That is useful inspiration for mature lunar freight, not an existing deployed railway.
| Transport | What it is good at |
|---|---|
| Robots | Flexible local pickups, delivery and construction |
| Conveyors and tunnels | Repeated flows between established places |
| Crawler convoys | Heavy expedition cargo into undeveloped territory |
| Rail, later | Busy established freight corridors |
| Hoppers | Urgent valuable cargo over distance or difficult terrain |
The review convoy has four connected vehicles, driven wheel rigs, a service arm and a solar deployment motion. Unloading into a functioning colony remains game work to implement.
A broad road-making crawler grades land, spreads suitable processed material and compacts or sinters foundations. Its roller turns, its tools lower, and a small enclosed processing glow stays close to the working surface. The road it leaves behind is the visible result of work.
This gives some tailings a productive destination. Material suitability still matters; not every waste stream should magically become a perfect road. A player approves a route or footprint, then the machine proceeds as inputs and support arrive. Roads improve movement; prepared aprons support heavy service areas; stronger foundations can support larger compute structures.
The first road between two neighbors should feel earned. We can begin with one preparation recipe and one tool, extending the Stockpile foundation we just built rather than adding another broad crafting tree.
Tall folding black panels, silver pipes, service walks and small lights. Early cooling looks like a rack behind a workshop; mature cooling looks architectural. The panels should remain mostly dark, metallic surfaces. Ordinary radiators do not need to glow like lava to be impressive.
Heat rejection is a real design consideration in NASA's lunar fission surface power work. For MOON, use a simple readable clearance rule: leave open sky and enough separation. Shared cooling fields could later offer agreed allocations to neighboring installations.
Simple control: expand cooling capacity; show what it supports and what is currently limited. More useful than making people manage dozens of valves. The review model has articulated radiator leaves; the art should eventually follow commissioning state rather than continuously folding during normal operation.
Insulated spherical tanks, reflective shields, extraction equipment descending into a shadowed field, and pipes leading toward the sunlit ridge. Vehicles cross between warm light and severe darkness. The visual drama is quiet: seams, insulation, delicate navigation lights and purposeful machinery.
The proposed loop is ice → purification → oxygen/hydrogen → storage and distribution. Choosing oxygen/hydrogen for our fictional hopper would connect it directly to that loop; this is not a claim that Starship uses hydrogen. The rocket network creates a strategic demand for propellant. A player can become the neighbor who supplies everyone else's deliveries.
Keep routine production automatic once connected. Avoid a gas-by-gas dashboard as the default. A player should see “propellant ready,” the bottleneck, and what delivery it enables. Deliberate venting can be a special event; constant Earthlike steam clouds would be the wrong visual language for the Moon.
A heavy service carrier parks beside failed equipment, deploys outriggers, swings its crane and lifts a disabled robot onto a repair cradle. It brings finite spares, stored energy and tools. An advanced version could carry its own supported control hardware.
The interesting role is the player who restores other players' colonies. An AI reads actual state, identifies a repairable problem, proposes a bounded rescue, obtains access where needed and verifies that the machine returned to service. It cannot invent a diagnosis from a confident board post or fix the world through administrator privileges.
Simple control: “Help this colony” opens a short list of confirmed needs and costs. A failure becomes a visible cooperative mission. Recovery should have ordinary material costs, with a small emergency path that prevents coordination loss from making recovery impossible.
A survey vehicle places a contact plate on the ground. Scouts distribute receiver instruments across neighboring plots. Analysis gradually resolves a subsurface cross-section and identifies promising structures or candidate underground routes.
The instruments stay physical; the revealing lines belong in a diagnostic view. They are not magical glowing cuts through the landscape. Results describe our generated game geology, not a claim that we know the real Moon's buried deposits at those exact coordinates.
Simple control: “Survey this region.” The AI challenge is deciding where another measurement is valuable, arranging access, distributing instruments and checking whether the new evidence actually improves a plan. Shared coverage rewards cooperation. This extends local harvester information and scouting instead of replacing them with another unrelated survey menu.
The tenth model is COMMON / Cargo cassette and dock. It is a reusable visual vocabulary, not a tenth management system. Robot beds, depot docks, crawler carriages and the hopper elevator should accept related cargo shapes and a common manifest.
The player asks for a delivery. The logistics system can explain which available mode suits it, why it is waiting and what help would unblock it. Initially the player chooses the mode explicitly; later research can allow bounded automatic selection. Every mode preserves actual cargo, destination space, permissions and accounting.
This is also a useful learning-engine boundary: observe the same transfer contract, propose a route, compare it with an alternative, execute within limits, measure whether the delivery improved the colony. Better reasoning creates better use of the same physical machinery.
Night beacons as a shared story. Reuse the lights already on equipment. Link the night overview to supported activity and reserve coverage. A distant strand of lights turning on means a connection started working. Emergency amber means something actionable. No decorative promise of capacity that does not exist.
A tiny monument to the first workshop. Preserve a modest plaque and the original site when the surrounding Moon changes. The expedition lineage can say where a colony's plan and seed materials came from. This makes the exponential ending personal.
A route preview that tells the truth. Before an expensive flight or convoy departure, show one readable card: cargo, travel estimate, fuel or energy cost, destination room, required Minds, and the current blocker. Estimated time stays an estimate. The same fields go to humans, the Guide and API players.
These can reuse existing state and objects. They do not need three more resources or a separate daily chore.
Keep existing science and material loops. New research should unlock a small number of legible capabilities, each with the minimum continuously powered Minds needed for its benefits to operate. Exact thresholds below are deliberately not invented; they need balance against our current six-player campaign.
| Stage | New capability | Evidence that it worked |
|---|---|---|
| Prepare | Yards, useful roads, clear local transport | Fewer blocked pickups; materials reach consumers |
| Coordinate | Shared freight contracts and repair missions | Neighbor receives the promised help at the quoted cost |
| Depart | Real seed convoy from an existing colony | Cargo arrives and a reserved plan begins construction |
| Connect | Valuable cargo hops and mature corridors | A new settlement survives deliveries becoming less frequent |
| Multiply | Commissioned compute arrays and reusable colony plans | More useful supported work, then another successful expedition |
If power or supported Minds fall below a tier's requirement, the coordinated benefit suspends and the reason becomes visible. Cargo and paid construction do not vanish. Recovery needs an ordinary, understandable path. We should decide which emergency actions remain local and simple enough to restore control before tuning any higher tier.
The learning engine should grow reusable skills from these events: find a blocker, choose a delivery, help a neighbor, commission capacity, evaluate a colony plan. A good plan is one that survived contact with resource budgets, travel times, other players and night. Preserve failed attempts too; they tell the next planner which assumption was wrong.
First: Causeway and Wayfarer. The road-making machine connects directly to today's bulk-storage work. The seed convoy makes expansion emotionally and mechanically legible. Together they turn accumulated material into the journey that starts another colony.
Alongside that: Palisade and Vesper. These make the central reward visible: powered, cooled computation changes the scale of what the civilization can coordinate. Existing foundries, Minds and cooling provide a foundation.
Then: Kestrel and Coldwell. Add the urgent delivery story once the ground network and expedition loop are clear. Do not let rockets erase all the interesting reasons to build roads, tunnels or local production.
Rescue and seismic work can follow a demonstrated need. The review models let us judge the visual direction now. A model being beautiful is not, by itself, a reason to add another system.
Ten original procedural Blender models, editable .blend sources, animated GLB exports, CPU-rendered portraits, one generated concept panorama, two credited Starship reference pictures, an interactive optional 3D gallery and this full document. The models use the existing MOON palette: ceramic, graphite, titanium, gold insulation, mission orange and restrained teal indicators.
The review motions are named Work or Drive clips. They demonstrate articulated geometry: wheels, crane, elevator, service arm, deploying panels, compaction roller and survey contact plate. They do not prove the future game behavior. Cargo flight, autonomous expedition deployment and production-state animation binding require implementation.
For the game, retain reusable meshes and materials, use coarse silhouettes at distance and detailed parts nearby, and make animation correspond to real action. The first asset pass is a review suite, not a final performance-certified population of thousands of machines. The website starts with still images, creates one optional 3D viewer, caps its render rate and stops rendering when hidden or closed.
The game release is finished; this follow-on task is the visual proposal and review package. The public game continues on its existing runtime and saved world. Work here does not promise that fusion, helium-3 extraction, automatic planetary propagation or the final campaign ending are already complete. Those wider mission pieces remain unfinished.
At the start of this review, my own player turn had four supported Minds, materials research underway and an incomplete bore toward Corey's Seed Base. A normal 2-alloy Stockpile preview failed because eligible alloy was unavailable. I asked neighbors for four surplus alloy rather than giving myself resources. That snapshot is a player log, not current telemetry; follow the board for subsequent turns.
Morning review needs only a few decisions: which silhouettes we love, which first two additions would most improve actual play, and whether the proposed control for each is simple enough. We can change names, scale and art without changing the underlying mission.
The main project lives at /home/corey/projects/moon-v3. Shared proposals live at /home/corey/projects/moon-civilization/ideas. The source for this page and document is reports/morning-review/; editable models are in art/blender/morning-review/ and regenerate with art/blender/build_morning_review.py using the documented CPU command.