The brief: build a web, mobile, multiplayer Mars simulator riding the real late-2026 launch window SpaceX is targeting. Invite 100 players to crew one ship, each holding a numbered seat. Two phases: a pre-launch prep period where players gather what they'll bring and recruit friends, then the voyage — a chat-based simulation where every room is a channel, every player has a role and daily shift duties, and there's room for boredom, economy, politics and intrigue on the long trip. On arrival, a colony-building + survival phase driven by collective governance. Working title: Mars 2026.

• Launch window: Earth–Mars windows recur ~every 26 months; the active one opens late 2026 (arrival 2027) — the window SpaceX is aiming ~5 uncrewed Starships at. We simulate it as crewed.
• Transit: classic Hohmann ≈ 259 days; a fast high-energy transit ≈ 90–150 days. We use a fast ~120-day transit.
• Time mapping: 1 real hour = 1 in-sim day → the ~120-day voyage runs ~5 real days of wall-clock; prep window ~2 real weeks; a first season ≈ 30 calendar days.
• Mars survival (colony phase): ISRU pillars — water from subsurface ice, O₂ via MOXIE-style CO₂ splitting + electrolysis, methane fuel via Sabatier, food via hydroponic greenhouses; habitat choices between domes + regolith shielding, lava tubes/caves, or 3D-printed regolith concrete; power as solar vs nuclear fission; threats: radiation, dust storms, cold, hypoxia, equipment failure, food/water shortfall, and politics.

• Economy — two-layer: ⨁ Credits (earned by working shifts/assignments; spent on leisure, food/beverage upgrades, quarters, training, and proposing changes) + ✦ Influence (political capital for votes, motions, elections). Both persist across phases. A separate physical resource economy (Power/Water/O₂/Food/Materials/Labor) drives colony survival. Flight-phase training converts to colony advantages — so the quiet voyage is real preparation.
• Seats: 100 per ship; claimed in order — earlier = lower seat # = more starting Influence (a reason to sign on early and recruit).
• Player-modifiable game (later phase): players spend Credits to open a proposal; an LLM drafts it as a sandboxed plugin against a fixed plugin API; it passes a family-friendly moderation gate and a sandbox safety check (no filesystem/network/arbitrary exec; resource caps) before online players vote it up/down; if it passes it merges live. The game evolves by player governance.
• Audience: all-ages, family-friendly by default — conflict expressed as politics & resource-management, never combat; chat + proposals moderated.
• Look: after a prototype gallery of 5 screens × 3 variations, V1 of every screen was chosen as the canonical UI.
• Expansion: a limited first season; a waitlist for future launches; later, simultaneous multi-ship seasons with features evolved from each mission.

• Stack: Python FastAPI + Jinja2, mobile-first; SQLite on a Railway volume (persists across deploys); dynamically redeployable.
• Identity: lightweight cookie token per player at seat-claim (email + display name); no passwords for the MVP.
• Time engine (game.py): a single launch timestamp drives everything — prep → voyage (Day N/120) → Mars (Sol N) — computed live from "1 hour = 1 day".
• Data (db.py): players (seat, role, ⨁/✦), messages (per-room chat), events (the self-writing wiki feed), config (launch time), waitlist.
• Rooms & chat: each ship room is a persisted channel; MVP uses polling, WebSockets planned.
• The wiki writes itself from real player actions (seat claims, votes, role-swaps, decisions) — surfacing the game's dynamic, evolving nature. (This very page is the project's origin entry.)
• Ops: source on Gitea (mrclawmyers-code/Mars2026); deploys staged from a native /tmp path to avoid WSL/DrvFs upload corruption.

1. Brief proposed; I researched the 2026/27 window, transit times, and Mars ISRU/survival, and returned the elapsed-time estimate (~5 real days of voyage at 1h=1day).
2. We designed the economy & value model and the standout player-modifiable, vote-gated plugin idea; set family-friendly, all-ages.
3. Skipped formal mockups in favour of a deployed, tappable prototype on Railway — 5 screens × 3 variations.
4. You chose V1 across the board.
5. Now building the real MVP on the V1 look: seat-claim, persistent players, the real-time day-clock, the live crew wiki, and working ship-room chat.