ACCEPTEDRoadmap: V3.12026-09-06 — updated 2026-09-06
ADR-005: Semantic layer depth partitioning for complex graphs
Partition the 3D knowledge brain into 3 progressive semantic layers: Core, Practice, and Foundations.
Launch stateThese entries are placeholders while the real case studies and experiment write-ups are prepared for publication.
Context
- As dozens of technologies, logs, ADRs, and projects were added, a flat 3D graph suffered from visual clutter and cognitive overload.
- Different audiences need different levels of depth: recruiters and peers start with shipped systems (Core), while deep-divers explore decisions (Foundations).
Decision
- Establish three concentric depth layers: Core (Work, Architecture, About), Practice (Labs, Logs), and Foundations (ADRs, Stack).
- Only the active layer is fully interactive; deeper layers render as non-interactive luminous background echoes (InstancedMesh).
- Camera automatically reframes (centroid + bounding radius) when toggling layers or following a guided Jarvis tour.
Alternatives considered
- Keyword filter pills without 3D depth. Rejected because hiding nodes abruptly breaks spatial continuity.
- Single monolithic cluster with opacity sliders. Rejected because dense clusters still occlude each other in 3D perspective.
Consequences
- Clean visual hierarchy with zero spatial confusion.
- Lower GPU overhead by rendering non-active layers as batched instanced meshes with minimal draw calls.
- Requires explicit mapping in brain-layout.ts for every newly introduced content kind.
Roadmap & Marcos de Implementação
V3.1Concentric Spatial Hierarchy
Progresso da Execução Arquitetural100% CONCLUÍDO
- Tri-Tier Concentric Stratification(Phase 1: Semantic Layer Taxonomy)Concluído
Grouped all site content into 3 clear layers: Core (Work, Architecture, About), Practice (Labs, Logs), and Foundations (ADRs, Stack).
- Non-Interactive Background Echoes(Phase 2: Luminous Instanced Echoes)Concluído
Rendered inactive depth layers as lightweight instanced meshes (InstancedMesh) maintaining spatial depth without interaction overhead.
- Automated Camera Flight & Zoom(Phase 3: Kinetic Camera Centroid Reframing)Concluído
Added automatic centroid and bounding radius calculation to smoothly glide the camera when layer filters change or Jarvis tours begin.
- Synchronized UI Controls(Phase 4: Adaptive Visual Legend & Grouped List)Concluído
Dynamic legend and categorized fallback views reflecting only items relevant to the currently inspected depth layer.