TMTravis Masingale

Summer 2026 / Case 02 of 05

C6 / Eye / Insight

Channel
Evidence index
Status
Atlas v1 working / WIP
Updated
05 July 2026
Span
1994–2026

Archive systems / creative record

Archive
Atlas

One source-traced index of a creative life; every generated view stays attached to its sources.

Work in progress

Atlas v1 is built and working. Preservation of vulnerable source media and fuller image hydration take priority over adding more generated views.

599,750
indexed items
32,678
constellation points
178
indexed entities
85.9%
indexed items hydrated

What becomes possible when every claim about a creative life traces back to a source?

The record spans art, photographs, design practice, teaching, exhibitions, applications, papers, and the storage systems that accumulated them. Archive Atlas does not relocate that material. It builds a read-only index over the sources already in place.

The outputs are views. The evidence remains the authority.

Scope

Six storage roots plus a teaching corpus feed one index with era taxonomy, privacy tiers, duplicate awareness, and master-source references.

Index once; generate many accountable views.

A source-linked entity can support a timeline, body-of-work view, thread, constellation, CV entry, exhibition application, site, or paper without being rewritten independently for each output. The compiler rejects entities that lack source evidence.

The semantic constellation adds pattern-finding without replacing the catalog. It maps embedded text and a sampled image set into a shared visual field, while every point retains a route back to the indexed item.

Relationship

Archive Atlas → Program Command. Both keep assumptions inspectable, but Atlas indexes a creative record while Program Command models institutional planning.

Archive Atlas → Writing. Evidence packets can support papers and applications without detaching claims from sources.

A static pipeline makes the evidence inspectable.

  • Index599,750 source items merged across storage zonesCompiled JSON
  • Entities178 works, courses, shows, people, documents, and related recordsSource-required
  • ObservatoryTimeline, Bodies of Work, Threads, and entity detailStatic site
  • Map32,678 embedded text and image points projected into two dimensionsCLIP + UMAP
  • VerificationDead-link and bit-rot checks plus open-question trackingOngoing

Evidence rule

The compiler aborts when an entity has no source reference. Explicit exclusions are applied before views build; privacy tiers remain attached to records for the still-unfinished public filter.

The pipeline protects source authority at every stage.

  1. Crawl read-only. Each source zone produces item records without moving or renaming originals.
  2. Adapt and compile. Teaching records join the same entity model; duplicates are reconciled and source requirements validated.
  3. Embed selectively. Text and a bounded image sample enter a joint semantic space; the content-keyed cache keeps the step resumable.
  4. Build static views. Compiled JSON becomes the Observatory without introducing a server or a second authority.

Decision

The archive stays in place. Atlas stores paths, facts, relationships, uncertainty, and privacy state—not another rearranged copy of the life.

Preservation comes before a denser map.

Some MFA-era masters live on vulnerable media, and much of the photographic corpus remains cloud-only. The next responsible move is to preserve and hydrate priority sources before increasing the image sample or generating more polished views.

After preservation, the next views are a verified CV and an exhibition-application assembler. Both must remain queries over the same compiled evidence.

Current edge

Preserve vulnerable masters, hydrate priority image folders, resolve the canonical-copy ambiguity, then densify the constellation.