Inspiration
Textiles and looms
Many independent threads can cross without losing their individual identity. Structure emerges from their relationships, not from flattening them into one list.
Local-first relational knowledge environment
Weave the threads.
Explore the continuum.
Lūm keeps time, place, relationship, evidence, provenance, and narrative connected as one canonical body of knowledge. Timeline, world, map, evidence, analysis, and story are different readings of the same continuum.
- Data model
- Entities, relationships, occurrences, places, evidence, stories
- Architecture
- Canonical domain → deterministic projections → renderers
- Operating model
- Browser-first, local-first, backend-free core
Sources
Entities
Occurrences
Places
Evidence
Stories
Continuumone canonical state
Timeline
World
Map
Analysis
Project history
From chronology to a woven knowledge environment
Lūm began with a timeline problem and expanded as the project repeatedly encountered the same limitation: chronology alone could not faithfully express who acted, where something happened, why a claim was believed, or how one fact belonged to multiple stories.
-
01
Timeline first
The initial focus was temporal navigation: point events, ranges, zoom, retained context, and a mobile-first chronology that could remain readable while moving through large bodies of material.
-
02
Relationships become situated facts
The graph model was tightened around durable entities and directed actions. Time and place moved onto occurrences instead of being treated as decorative metadata or graph nodes.
-
03
One continuum, many projections
Timeline, relational world, geography, evidence, and stories converged around shared canonical identity. Renderer state, force positions, camera coordinates, and clustering became explicitly disposable.
-
04
Lūm
The loom metaphor gave the project its mature identity: loose threads are reconciled into a coherent fabric, while each projection exposes a different structure in that same fabric. The metaphor explains the product; precise domain vocabulary still governs persistence.
-
05
Investigation and assisted authoring
Evidence, provenance, candidate claims, import review, structured occurrence composition, local inference, and agent-assisted proposals extended Lūm beyond visualization into a controlled knowledge-work environment.
Inspiration
Investigative chronology
A useful chronology must preserve uncertainty, provenance, competing explanations, and the ability to move from a claim to its source.
Inspiration
Maps and graph thinking
Geographic position and relational topology are complementary. Lūm treats both as projections over canonical records rather than competing databases.
Inspiration
Local-first tools
The project assumes the user should be able to keep core knowledge work in the browser without requiring a hosted backend to own the data.
Ambition
A general environment for reasoning across time, place, relation, and evidence
Preserve identity across every view. Selecting an occurrence in a timeline, map, graph, evidence dossier, or story should always resolve to the same canonical fact.
Support inquiry before certainty. Questions, assumptions, candidate claims, conflicting evidence, and falsification paths should coexist without prematurely becoming accepted fact.
Make scale navigable. Large chronologies and dense relational worlds should remain usable through deterministic projection, semantic zoom, filtering, and retained context.
Remain modular. Timeline, world, composer, evidence, and diagnostic capabilities should be reusable without creating parallel state authority.
Onboarding
Understand Lūm in four moves
Start with durable nouns
Create entities for people, organizations, objects, concepts, or other things that need independent identity.
Entity: Ada Lovelace
State what happened
Relationships use directed actions. When a relationship is situated in time, it becomes an occurrence.
Ada → wrote → Notes
Attach context, not duplicates
Reuse canonical places and evidence. Time and place belong to the occurrence; provenance points back to the supporting source.
1843 · London · source page 12
Change the projection
Inspect the same records as chronology, relationship topology, geography, evidence, analysis, or authored story.
continuum → projection
Applications
Conventional and unconventional uses
The model is intentionally general. The examples below differ in subject matter, but they all use the same entity → relationship → occurrence → place/evidence → projection structure.
Conventional
Historical research
Track actors, dated events, places, sources, and competing narratives across a long chronology.
Investigations
Connect claims to evidence, preserve provenance, expose contradictions, and distinguish accepted facts from unresolved candidates.
Project and incident review
Reconstruct decisions, dependencies, handoffs, failures, and remediation across teams and systems.
Genealogy and biography
Follow people across places, relationships, documents, life periods, and narrative threads without duplicating identity.
Unconventional
Scientific hypothesis history
Represent experiments, observations, publications, rebuttals, and theory changes as evidence-linked occurrences.
Fiction continuity
Model characters, locations, artifacts, hidden knowledge, parallel timelines, and canonical story order while separating narrative order from chronology.
Personal decision archaeology
Trace major choices back to information available at the time, assumptions, alternatives, and later outcomes.
Infrastructure dependency archaeology
Relate services, vendors, hardware, outages, migrations, and operational decisions to see how today’s system accumulated.
Embedded examples
The same domain model, different questions
Select a scenario. Each miniature projection below is generated from example canonical data embedded in this page.
Timeline
Relations + place
Evidence
Continue
Move from the presentation into the continuum.
The live workspace is the authoritative product surface. This page explains the model; the application lets you manipulate it.
Open Lūm workspace