Accretion
A minimalist cosmic-colony sandbox builder
- When
- Jan 2026 – Present
- Where
- Los Angeles, US
- Team
- 7 people
- Platform
- PC
- Engine
- Unity
- Role
- Co-Director, Audio Director, Producer, Systems Designer, Programmer, QA
Team — My roommate and I as the full-stack core, plus 2 Berklee audio, 1 tech artist, 1 3D artist, and 1 design+programming hybrid.
Play as a core block that attracts, transports, places, and assigns other blocks to produce, upgrade, and build out entire planets — a sandbox hybrid of incremental, colony-sim, and survival genres.
Players collect resources → automate production → build logistics networks → craft higher-tier resources → unlock the next era (power, planet factories, warp gates) and repeat the optimization loop. The resource-flow graph was designed in Miro + Machinations; 1000+ sample Monte Carlo runs informed balance passes using playtest-telemetry-derived distributions. Multi-system coupling is kept tight so the probability space doesn't explode — the incremental loop needs carefully thresholded quantity-to-quality jumps to keep pacing right.
Design Highlights
- ▸No level design — instead, big-phase control typical of incremental/colony-sim/survival games (tech tree, milestones, milestone events).
- ▸Designed player pain points alongside their remedies (aspirin–headache no-mismatch principle) to keep positive-feedback frequency on target.
- ▸Sample variance is derived from playtest telemetry — real range/variance distributions fed back into Machinations as random flows, then grouped by player archetype for analysis.
- ▸2 Miro boards + 1 Machinations map. System count: 3Cs and basic verbs, tutorial, 5 production-side systems.
Production Highlights
- ▸Started in Notion; team then built an in-house production tool site for file management, backlogging, scheduling, burndown, and Gantt.
- ▸1-week sprints with a Sunday weekly meeting for sprint planning. Milestones: prepro → alpha → beta → rc.
- ▸Double scoping: met the class-project minimum (complete game + trailer) while simultaneously planning long-term commercial scope.
- ▸Once the class beta goals were essentially met, I led the team into long-term dev: 2-month re-prepro, with a hard rule to lock moodboard + art bible + lookbook + design doc before proceeding.
- ▸Everyone on the small team is otherwise busy, so I relaxed strict scrum priority ordering where pragmatic.
- ▸Acted as de-facto co-audio-director with the Berklee team — weighed in on genre and tempo decisions throughout.