decision-log.md

Decision Log

Status markers:

• locked: should guide implementation unless a deliberate reversal is documented
• provisional: preferred default, but still open to change after paper or prototype testing
• open: intentionally unresolved

Current Decisions

Date	Topic	Status	Decision	Why
2026-03-11	Simulation priority	locked	The game is simulation-first and not avatar-first.	This is the core product identity.
2026-03-11	World model	locked	The authoritative model is a true 3D discrete grid.	Fake 3D would weaken the main differentiator.
2026-03-11	Presentation	locked	The default presentation is constrained 3D with slicing and overlays.	Readability matters more than immersion.
2026-03-11	Engine choice	provisional	Build the runtime in Bevy.	It fits the code-first workflow and ECS-heavy view layer.
2026-03-11	Architecture split	locked	Keep the simulation core separate from Bevy-facing entity state.	This is required for tests, saves, and replay.
2026-03-11	Theme direction	provisional	Use an arcane-industrial core with a bio-industrial secondary layer.	It preserves the factory identity without collapsing into generic fantasy.
2026-03-11	Enemy pressure	locked	Exclude enemies from the first playable slice.	The first milestone is about building and inspection clarity.
2026-03-11	Machine scale	provisional	Most slice machines are 1 x 1 x 1, with limited multi-cell infrastructure.	This keeps placement readable while still proving 3D occupancy.
2026-03-11	Transport subset	locked	Use belts, lift wheels, and wires for the first playable slice.	This is enough to prove horizontal routing, vertical routing, and power topology.
2026-03-11	Resource model	locked	Use discrete solid items plus an explicit power network and contamination field for the first playable slice.	It gives mixed systemic behavior without overbuilding chemistry.
2026-03-11	Contamination model	provisional	Model contamination as a volumetric field plus occasional material impurity.	It creates spatial pressure without infecting every recipe with bookkeeping.
2026-03-11	Time controls	locked	Support pause, 1x, 4x, and single-step in the first playable slice.	Simulation debugging and inspection need all four.
2026-03-11	Inspection tooling	locked	Slice controls, placement ghosting, blocked feedback, and at least connectivity/throughput overlays are required in the first playable slice.	3D density is unreadable without them.
2026-03-11	Prototype exclusions	locked	Exclude enemies, avatar mode, procedural maps, fluids, long-range logistics, moving contraptions, and multiplayer from the first playable slice.	The slice must stay narrow enough to validate the core idea.
2026-03-12	Build inventory model	provisional	Keep a shared global build inventory for the first playable slice.	It keeps fabrication, placement, and UI simpler while the core loop is still being validated.
2026-03-12	UI screen model	provisional	Prefer one continuous in-world HUD plus contextual panels over multiple modal inventory/configuration screens.	This keeps the player in the build/inspect loop and avoids UI sprawl.
2026-03-12	UI tone	locked	Keep navigation and diagnostics GUI-first rather than diegetic in the first playable slice.	The game is analytical and slice-driven; diegetic UI would obscure state instead of clarifying it.
2026-03-12	Asset sequencing	locked	Finish a final pre-asset UI pass before the first serious Blender blockout pass.	Machine silhouettes should answer a stable interaction contract, not chase moving HUD/inspect/slice decisions.

Questions Still Open

• Should the long-term game stay mostly industrial or push harder into bio-industrial and arcane systems?
• Does contamination deserve active mitigation machines after the first paper design pass?
• Should the first scenario use one handcrafted objective or a short chain of milestone objectives?
• At what point do multi-cell production machines become necessary instead of just multi-cell infrastructure?
• How much progression should exist before the project has proven the base building loop?