Production

Six milestones, one named MVP, and the gate each one has to pass.

This is the design-facing view of the same work docs/roadmap.md sequences from the engineering side. Where they disagree, the roadmap is the one that has to be right about seams and this one has to be right about what the player gets.

The shape

flowchart LR
    M1["<b>M1</b><br/>Vertical<br/>architectural proof"] --> M2["<b>M2</b><br/>The Believable<br/>World"]
    M2 --> M3["<b>M3</b><br/>The Pilot"]
    M3 --> M4["<b>M4</b><br/>The Explorer<br/>◆ MVP"]
    M4 --> M5["<b>M5</b><br/>On Foot"]
    M5 --> M6["<b>M6</b><br/>Conflict"]
    M4 --> M7["<b>M7</b><br/>Persistent<br/>Universe"]
    M6 --> M7

    style M1 fill:#065f46,stroke:#064e3b,color:#fff
    style M2 fill:#0369a1,stroke:#0c4a6e,color:#fff
    style M4 fill:#5b21b6,stroke:#3b0764,color:#fff

M1 is complete. The twelve foundation checks pass in Node and in recorded Chrome runs, including a warmed offline production build. Later milestone scope below is a design target, not a claim that every item is unfinished. The galaxy, terrain, camera, and surface character implement parts of M2, M4, and M5 without closing their gameplay acceptance gates.

The MVP: The Explorer

M2 + M3 + M4. The first releasable product, and it is a complete game rather than a slice of one.

You fly a ship through the real solar neighborhood. You can go to any star you can see. You can land on nearly anything solid and it looks like a place. You survey what you find, your name goes on what you found first, and the catalog gets better over time because real astronomy does.

Combat and multiplayer beyond discovery records remain outside the MVP. Basic on-foot traversal is implemented alongside exploration. Suit survival, object interaction, and continuous ship interiors remain M5 scope.

Why this is the right MVP

It is the only slice where the pillars are all fully expressed. Continuity, reality, orbital momentum, and an embodied player all matter to the survey loop. That loop does not require combat, suit survival, or ship interiors.

It is a whole game to a large audience. A substantial share of Elite Dangerous players spend nearly all of their time exploring, with no combat and no trading. That is not a niche within the audience; for this design it is the audience.

It is the shortest path to the thing nobody else has. Combat is where we compete with Elite's decade of tuning. Exploration on real, versioned astronomy is where nobody competes at all.

Interior interaction and networked play remain separate scope. Basic character locomotion does not settle continuous boarding, physical tools, pressure boundaries, or multiplayer reconciliation. Those systems retain their own acceptance gates.

M2 — The Believable World

The single hardest milestone, and the one that determines whether any of this looks like anything.

Implemented foundations: WebGPU/TSL, restricted morphing terrain, cross-face coverage, predictive terrain streaming, rock scatter, extended-range output, the sensor chain and its three camera modes, profiling tools, and the calibrated galaxy sky. Their measurements and limits are in technical, ADR-0037 and ADR-0038.

The M2 gate remains a target. Authored biome material sets, erosion, a walking player with rock interaction, and demonstrated 60 fps on the target laptop are not established by those implementations. The estimate above is the original scope estimate, not a fresh forecast for the remaining work.

🎮 Designer's Note: M2 has no new gameplay in it at all, which will make it feel unrewarding for months. It is still correct to do first. The roadmap reaches the same conclusion from a different direction: terrain is the visible ceiling on everything surface-related, it exercises streaming and LOD properly, and every later content system sits on top of it.

M3 — The Pilot

The ship becomes a machine you operate.

M4 — The Explorer ◆ MVP

The loop closes and the game becomes releasable.

Catalog ingest, version reporting, issue ordinals and predicate world search are implemented. Discovery credit, the Almanac reward loop, online sync and this milestone's player acceptance gate remain future work.

Ship it here. Not as early access with a promise of combat, but as a complete exploration game that may grow.

M5 — On Foot

Surface traversal already has a canonical character, local-gravity jumps, optional pointer lock, and first-person and third-person controls. Terrain and flat support disks supply contact. A trusted local session can permit flight. On foot distinguishes this implemented movement from the remaining suit, interaction, EVA, and interior systems. The full excursion gate below is still open.

M6 — Conflict

M7 — The Persistent Universe

Cadence and post-launch

After M4 ships, the project has a property almost nothing else has: it improves without anyone authoring content. A catalog revision is new content generated by the astronomy community, and the pipeline that delivers it is built once.

There is no season, no battle pass, no live-ops calendar and no event schedule. See charter.

Critical path and the things that will go wrong

Total to MVP: [Assumption: 11–18 months at one maintainer with coding agents, assuming sustained effort. Treat the lower bound as unlikely.]

🎮 Designer's Note: The plan's real risk is not any single item — it is M2's length. Four to seven months with no new gameplay, on the hardest technical work in the project, is exactly where solo projects stop. The mitigation is the 90-second acceptance test: a concrete, visual, binary gate that produces something worth watching the moment it passes, and that can be attempted in partial form long before it is met.

Related

• docs/roadmap.md — the engineering sequence
• technical — what each milestone requires of the platform
• risk — what could stop each one
• charter — why the plan is shaped like this