Rabbit Holes

A visualization of browsing journeys—how a single spark becomes momentum, how curiosity tightens into flow, and the tender moment when we surface.

↓ scroll to explore

The Anatomy of Curiosity

Over months of browsing, 3,560 distinct chains emerged—each one a path carved by attention.

Some are brief sparks: a single click, a fleeting question answered. Others are deep dives, spiraling down for hours, accumulating 283 pages in a single thread.

The longest journey lasted 4.8 days. The most intense consumed 9 hours of pure focus.

The PyCon Rabbit Hole

September 13, 2025, 4:01 AM
It begins at in.pycon.org/2025/program/schedule—scanning the conference schedule.

One talk catches the eye: "Data Morph: A Cautionary Tale of Summary Statistics". Click. 7.5 minutes reading the abstract. Curiosity piqued.

The path branches: back to schedule → "DuckDB is the New Pandas" (5 min) → "Memory is the Agent: Architecting Stateful Reasoning" (18 minutes of deep focus).

Then the momentum builds: back to the full schedule, scrolling, comparing, planning. A single spark— one interesting talk title—becomes 6.3 hours of sustained exploration across 9 pages and 2 domains.

Chain #247892: Conference Deep Dive

9
Pages
6.3h
Foreground Time
2
Domains

Branching Wide: The Omakub Exploration

September 24, 2025, 1:46 PM
Landing page: omakub.org—"An Omakase Developer Setup for Ubuntu by DHH". One click to GitHub.

Then the branching begins. Not a linear descent, but an explosion of tabs:

  • /applications directory → opened 5 times, examining different files
  • install.sh → read 4 times, understanding the setup
  • /install/terminal → dove into language selection, storage, apps
  • Individual scripts: zellij, lazygit, lazydocker, github-cli, btop...
  • Desktop apps: Alacritty, Typora, LocalSend...

48 pages visited, but only 2 clicks deep. Branching factor: 24×. Each page opened directly from the root, tabs fanning outward like a peacock's tail. Not a rabbit hole descending—a survey, an inventory, a systematic cataloging of possibilities.

Chain #255363: The Wide Survey

2
Max Depth
48
Total Pages
24×
Branching
13.5m
Duration

This is texture: wide, not deep. Reconnaissance, not excavation.

The Firebase Configuration Marathon

August 17, 2025, 2:34 AM
Entry point: firebase.google.com → Firebase console.

What follows is not curiosity, but work. A methodical tour through configuration panels:

  • Project overview → Settings → Web app configuration
  • IAM permissions → Service accounts
  • Authentication setup → Sign-in methods
  • Cloud Firestore → Rules → Data explorer
  • Navigating to /notes/OymphbNDCi8V16CSeGO8—a specific document

37 pages across 3 domains (firebase.google.com, console.firebase.com, Firebase APIs). Only 6 clicks deep, but 2 hours of sustained focus. Branching factor: 6.2×.

This is the texture of configuration: returning repeatedly to index pages, opening panels in new tabs, cross-referencing settings. A hub-and-spoke pattern, not a linear thread.

Chain #200853: Configuration Work

6
Max Depth
37
Pages
6.2×
Branching
2.1h
Focus Time

Patterns of Exploration

Three distinct textures emerge from the data:

Linear Spirals

Root
└─ A
   └─ B
     └─ C
       └─ D
         └─ E

One page → next → next. Deep focus, single thread. Branching ≈ 1.0

Hub & Spoke

Root
├─ A
├─ B
│  ├─ B1
│  └─ B2
├─ C
└─ D

Return to index, open new tab. Configuration work. Branching 2-6×

Wide Survey

Root
├─ A
├─ B
├─ C
├─ D
├─ E
├─ ...

Many tabs from root. Cataloging, surveying. Branching 10-24×

The visualization above shows these patterns as flowing streams. Each particle is a page view, gathering speed as chains deepen. Color represents time intensity (blue → purple → pink). Size scales with foreground duration. Flow captures momentum—how chains accelerate, branch, and eventually dissipate.

What the Data Reveals

Across 3,560 browsing chains, patterns of attention emerge:

  • Deep linear dives are rare. Most chains with depth > 100 are calendar refreshes, authentication flows, or automated patterns. True content exploration tends to branch.
  • Branching correlates with learning. The Firebase configuration (6× branching) and Omakub survey (24× branching) show systematic knowledge gathering—opening many tabs to compare, cross-reference, and understand.
  • Time doesn't equal depth. The PyCon chain: only 9 pages but 6.3 hours. The Omakub chain: 48 pages but 13 minutes. Intensity of engagement varies independently of navigation breadth.
  • Work patterns vs curiosity patterns have distinct signatures. Configuration work (Firebase) shows hub-and-spoke. Pure curiosity (PyCon schedule) shows depth-first with strategic backtracking.

This is not productivity theater. There's no shame in the 48-tab survey that lasts 13 minutes. There's no prize for the 6-hour deep dive. These are simply the textures of thought—the way attention moves when following genuine interest.

The Tenderness of Surfacing

Every rabbit hole eventually ends. Not with shame, but with completion. A question answered, a need met, a configuration saved.

The PyCon schedule browsing ends when the session plan is complete. The Omakub exploration ends when the mental inventory is cataloged. The Firebase configuration ends when the document is found, the settings are saved.

These visualizations celebrate the deep dive—the willingness to open 48 tabs if that's what understanding requires, to spend 6 hours on 9 pages if that's what the question needs, to spiral down 247 clicks deep if that's where the answer lives.

This is the texture of digital thought: spiraling, branching, gathering momentum, and finally, tenderly, letting go.

Legend

Brief visits
Moderate engagement
Deep focus (hours)
Size = time spent
Flow = chain momentum