entry_id: RD-2026-006
project_id: PRJ-2026-003
core_activity: CA-01
title: Can six hundred source files become a usable three-dimensional city?
timezone: Australia/Perth
period:
  started: 2026-07-12
  ended: 2026-07-15
research_effort:
  hours: 30
  basis: estimated
  note: Reconcile this public estimate with private contemporaneous company time records.
source_record:
  codex_session_id: 019f4fcf-b843-7222-b270-c23e8ec3a77a
  repository: mobily-enterprises/vibe64
  worktree_branch: system-world-v1
  initial_commit: 218ae6305c380746f5f8da363acb17fac946b20e
  final_experimental_commit: df7fe04a66a92fdf5e02f2ccc82c035f0dca7fc4
  final_experimental_commit_url: https://github.com/mobily-enterprises/vibe64/commit/df7fe04a66a92fdf5e02f2ccc82c035f0dca7fc4
  note: >-
    The complete live-evaluation transcript, renderer revisions, screenshots and
    intermediate worktree state are retained privately. The entry records the
    observed sequence, including rejected changes and unfinished work.
hypothesis:
  recorded_at: "2026-07-12 08:35 +0800"
  statement: >-
    If the filesystem remains the stable spatial ground truth, with directories
    as nested precincts, files as buildings sized by physical lines of code and
    architecture added as a selective overlay, then a repository of roughly six
    hundred source files can remain responsive and reveal both physical and
    logical structure without collapsing into a dependency web.
primary_success_condition: >-
  Render the complete Vibe64 source set at an interactively usable speed; keep
  directory depth, file identity and large-file mass visible; allow selection
  to expose exact local relationships; and add subsystem relationships without
  drawing unsupported file-to-file claims or an unreadable global web.
controls:
  - The same Vibe64 working tree and generated 600-file source model were used throughout the trial.
  - File size represented physical line count rather than an inferred complexity score.
  - Canonical file and directory positions remained deterministic.
  - Relationship claims retained their evidence kind and exact endpoints where known.
  - Visual changes were evaluated in one live browser by the product inventor.
measured_outcomes:
  subject_source_files: 600
  subject_packages: 17
  revision_035_connections:
    imports: 2392
    injections: 71
    declarations: 83
  revision_035_verification:
    build: passed
    lint: passed
    client_tests_passed: 538
    focused_server_tests_passed: 11
    wider_server_tests_passed: 1248
    wider_server_tests_failed_unrelated: 8
  final_focused_verification:
    client_layout_tests_passed: 9
    system_route_and_service_tests_passed: 6
    build_check: passed
  retained_feature_history:
    focused_commits: 17
    package_files: 37
    focused_test_files: 6
    total_files: 43
    lines_added: 16793
  architectural_levels:
    implemented: true
    usability_evaluation_complete: false
result: inconclusive
reason: >-
  The prototype became subjectively responsive and the evidence model supported
  exact imports, declared-only dependencies and subsystem bundles. However, no
  frame-time target or comparative comprehension protocol had been fixed in
  advance, and the implemented architectural layers had not completed live
  usability evaluation within the recorded period.
next_test: >-
  Freeze a representative repository snapshot and define a repeatable trial:
  measured frame-time and interaction latency at several file counts, followed
  by timed architecture questions answered in File City and in a conventional
  file tree by participants who did not build the system.
