READ4 min
WORDS834
SECTIONS3
TYPEGUIDE
CHECKED7 SEP 26
Every page on this site is a star. Every link between two pages is a line between two stars. The bright core is what the rest of the site keeps pointing back at; the dark rim is what nothing points at yet. Click the view and fly.
building the galaxy…
WASD flydrag lookshift boostE/Q up & downhover inspectclick pinenter open
How to read it
Distance from the core — how much of the site links to a page. The most-referenced pages sit in the bright middle; pages nothing points at yet drift to the rim.INBOUND
Which arm — its section. Five spiral arms, one per section, because pages within a section link to each other more than across.SECTION
Size — also inbound links, so the core reads bright at any distance.SIZE
Lines — a link inside the body of an article. Navigation, footers and read-next blocks are excluded, so what you see is editorial connection rather than site furniture.LINKS
Position — computed once from the link structure, then fixed. It is the same galaxy every visit, so a page's place in it means something.STABLE
WHAT THE SHAPE SHOWS
Two things are visible here that a list cannot show.
Which pages hold the site together. The corrections log is the most-linked page on the site — not an article, but the record of what this site got wrong. That is either the best thing about it or the most damning, and the map makes the position literal.
Where the sections actually touch. Craft and record are separate bodies of work, and the lines between them are the places where technique meets measurement.
What sits on the rim. The outermost stars have nothing pointing at them from inside an article — the list below marks each of them with a dash instead of a count. Most are section hubs, which are reached from the navigation bar rather than from prose — that is normal. The rest are genuinely unconnected: written, published, and never referred back to. Both kinds look identical out there, which is the uncomfortable part.
Two pages are deliberately excluded as link sources: this one and the index. Both list almost the whole site, and counting their links added 305 edges of pure furniture and gave every page on the site an inbound link — which would have erased the rim entirely.
Every page, as a list
Every page above, in plain HTML, grouped by section and ordered by how much of the site links to each one. The number beside a page is how many other pages point at it. This list is generated from the identical data the map draws, so the two cannot disagree — and a stale count here would be caught before it could ship. If the view above never loads, nothing is lost but the picture.
It also drives the map. Hovering a title here lights that star and every link into and out of it; clicking one flies you to it. Finding a specific page among 159 unlabelled points is much easier by name than by eye.
220 pages · 1224 links between them
Articles 101
- Why AI Makes Things Up — The Mechanism, Not the Mystery22 in
- Free vs Paid AI Image, Video and Music Tools: What the Free Plan Really Allows21 in
- HOW AN AI IS BUILT — How an AI Is Actually Built19 in
- How AI Is Tested — Benchmarks, Red Teams, and What Nobody Can Measure16 in
- Character Consistency in AI: Images, Video & Storyboards14 in
- Local or Cloud: Where Does AI Live Next? An Impartial Study12 in
- Open or Closed — Who Gets to Hold the Weights12 in
- What Makes a Prompt Good? The 6 Non-Negotiable Elements10 in
- Why AI Output Looks Fake10 in
- Every Google AI Tool in 2026, and What It Is Called Now9 in
- Free vs Paid AI Chat: What Each Plan Actually Gets You9 in
- Veo 3.1 vs Gemini Omni vs Grok Imagine Video9 in
- What Is Meta Muse? The Personal AI Agent From Meta, Explained9 in
- Critical Thinking Is the Skill of the Age8 in
- Dead Internet Theory Meaning: What It Claims and What the Evidence Supports8 in
- Gemini Temperature Setting Deprecated: What Changed8 in
- Gemini vs Grok API Prices, Side by Side8 in
- LLM API Prices: OpenAI, Anthropic, Google and xAI8 in
- The Work AI Cannot Take8 in
- AI Labs Outside the US and Their Open-Weights Licences7 in
- AI Prompting Terms Glossary7 in
- Grok 4.7 vs Grok 4.6: What Actually Changed7 in
- How Much Does a Veo 3.1 Video Cost?7 in
- Pages Written For Machines, Read By No One7 in
- The Iterative Workflow: How to Refine AI Output Instead of Starting Over7 in
- Which Claude Model to Use, and What Each Costs7 in
- Who Builds AI — The Labs, the Models, the Money7 in
- AI Terms People Get Wrong: Open Source, Reasoning, Memory and More6 in
- Claude Opus 5.5 vs Opus 5: What Changed and What It Costs6 in
- Color, Light & Mood: The Visual Language of AI Generation6 in
- Contextual Prompting Meaning — Why Prompt Engineering Died6 in
- Grok 4.6 vs Grok 4.5: What Actually Changed6 in
- The AI Glossary: sixty terms, honestly defined6 in
- The Anatomy of an AI Video Prompt6 in
- What a Director’s Name Does to a Prompt6 in
- What AI Skills Are, and How They Work6 in
- What Happened to Vertex AI, NotebookLM and Imagen?6 in
- When AI Models Retire, and How Much Warning You Get6 in
- Who Owns What You Generate6 in
- Does AI Understand Anything? — What Can and Cannot Be Settled5 in
- Gemini Notebook, Formerly NotebookLM: What Actually Changed5 in
- Gemma 4 vs Muse Glimmer: Two Open Models, Two Different Bets5 in
- How AI Fraud Actually Works5 in
- How AI Got Into Your Tools Without Asking5 in
- How to Run Muse Glimmer Locally5 in
- Meta Muse vs Gemini Spark vs Grok Bot, Compared5 in
- Muse, Grok Bot, AI Studio: Which Product Do You Mean?5 in
- Prompting for LLMs: ChatGPT, Claude & Gemini — What Actually Differs5 in
- The AI Tools Atlas: the landscape by job, not by logo5 in
- What AI Actually Runs On: sand, power, and heat5 in
- What Stays Human: slow, imperfect, on purpose5 in
- What the OpenAI API Actually Costs5 in
- Who Owns Grok Now? xAI, X and SpaceX, Explained5 in
- Why Your Generation Came Out Wrong5 in
- AGI, Meaning: what the term actually claims — and why nobody agrees4 in
- Google AI Studio vs Gemini Agent Platform4 in
- HOW AI FILMS ARE MADE — How AI Films Are Actually Made4 in
- How Big Is Colossus? The Numbers xAI Publishes4 in
- Is Llama Dead? What Meta Replaced It With4 in
- Meta’s Developer Stack: Muse Spark, Code, Image and the Model API4 in
- Still Optimistic — but everything is about to move faster, and we need to keep up4 in
- The AI Influencer Playbook: Build a Persona, Create Content, Get Paid4 in
- The CARE Framework: Context, Anchor, Restrict, Exemplify4 in
- The Sora API Shuts Down on 24 September 20264 in
- Ultra-Detailed Upscale Prompts: 25 Variations for Every Vibe4 in
- What a Repo Is, and How Repos Actually Work4 in
- What Is Meta Vibes? The AI Video Feed, and the Likeness Question4 in
- What Is Meta’s Sentinel? The Agent That Guards the Agent4 in
- Which AI Image Model for Which Job: Vendor Claims and Prices Compared4 in
- Which Jobs Are Safe From AI? What the Evidence Actually Supports4 in
- AI Images in Property Listings3 in
- AI Product Photography That You Can Actually Sell With3 in
- AI Studios & Aggregators: Every Platform Explained3 in
- Astra, Opus, Flash: What AI Model Names Actually Mean3 in
- Gemini CLI Is Gone: Moving to Antigravity CLI3 in
- How to Use Reference Images in AI Generation3 in
- Prompting 101: The Complete Beginner's Guide to AI Prompting3 in
- SAM 3.1: What Meta Changed, and Why It Is Twice as Fast3 in
- Seedance 2.5 vs 2.0: What a Clip Actually Costs3 in
- The Automated-Internet Vocabulary: 25 terms3 in
- Understanding AI Model Types: Image, Video, Audio & LLM Explained3 in
- What AI Cannot Cover3 in
- What Is Gemini Spark? Google’s Background Agent, Explained3 in
- What the Internet Was — And What Actually Changed3 in
- Where Whisk and ImageFX Went: Google Flow, Explained3 in
- AI for a Local Business2 in
- Aspect Ratio, Resolution & Format: The Technical Decisions That Change Everything2 in
- FLUX 3: Black Forest Labs’ First Video Model, Prices and Limits2 in
- How to Build an AI Content Pipeline: Idea to Published Post2 in
- How to Build for Meta Ray-Ban Display Glasses2 in
- How to Explain AI to Someone Else2 in
- The Eight Kinds of AI You Already Use2 in
- The Five Archetypes2 in
- The History of AI: 1950 to Today — A Timeline You Descend2 in
- What Companies Actually Use Grok Bot For2 in
- Which OpenAI Model to Use: The Four That Matter2 in
- 3D Websites From a Prompt1 in
- How AI Prompt Geneer Started1 in
- The AI Coding Stack, Explained1 in
- Vertical Microdramas, Explained1 in
- Articles—
Craft 52
- What the Vendors Actually Say27 in
- Negative Prompt Library24 in
- Model Prompting Cheat Sheets20 in
- Agent Guardrails14 in
- Permission Tiers for Agents14 in
- Context Engineering13 in
- Nano Banana 2 Prompting Cheat Sheet12 in
- Loops That Improve and Loops That Degrade9 in
- Model Guides9 in
- What Generation Actually Costs9 in
- ChatGPT Image Prompting Cheat Sheet8 in
- The Prompt Vault8 in
- Skills Are a Supply Chain7 in
- The Agent Brief7 in
- A Security Checklist for AI-Built Software6 in
- Diagnosing a Run That Went Wrong6 in
- LLM Prompting Cheat Sheet6 in
- Photorealistic, Hyperrealistic, Cinematic, Anamorphic6 in
- Prompt Injection: Why There Is No Fix6 in
- Prompting Agents6 in
- What You Are Actually Burning6 in
- Automations5 in
- Bad, Good, Great5 in
- FLUX.2 Prompting Cheat Sheet5 in
- Free Alternatives: Run AI on Your Own Computer5 in
- Prompting Myths: Advice That Stopped Being True (or Never Was)5 in
- Suno Prompting Cheat Sheet5 in
- Veo 3.1 Prompting Cheat Sheet5 in
- What You Owe a Client Whose System You Automated5 in
- Camera Grammar4 in
- Getting Text Right Inside an Image4 in
- Midjourney Prompting Cheat Sheet4 in
- Prompt, Context, Skill, or Fine-Tune4 in
- Seedance 2.5 Prompting Cheat Sheet4 in
- The Craft4 in
- The Review Loop, In Order4 in
- ElevenLabs Prompting Cheat Sheet3 in
- Grok Imagine Prompting Cheat Sheet3 in
- How to Test Whether a Prompt Change Actually Helped3 in
- Multi-Agent: When It Helps and When It Does Not3 in
- Prompting Cheat Sheets3 in
- Prompting Music and Voice3 in
- Reusable Context Beats Reusable Prompts3 in
- Kling 3.0 Prompting Cheat Sheet2 in
- Prompt Caching: The Biggest Discount Nobody Turns On2 in
- Reverse Prompting2 in
- Runway Gen-4.5 Prompting Cheat Sheet2 in
- Seedance 2.5: The Cinematic Look2 in
- Seedream 5.0 Prompting Cheat Sheet2 in
- Skills, Repos and MCP2 in
- Higgsfield Prompting Cheat Sheet1 in
- Plan It Before You Generate It1 in
Record 36
- How to Spot AI Writing — Nine Signals That Actually Work16 in
- State of the Automated Internet15 in
- How to Ask AI Well — Habits That Outlive the Models11 in
- SPOT FAKE IMAGES & VIDEO — How to Spot Fake Images, Video and Voices11 in
- Auditing AI-Generated Work10 in
- Deepfake Self-Defence10 in
- How AI Influences You Back10 in
- When Not to Use AI — The Cases Where It Costs You10 in
- What Happens to What You Type — Retention, Training and Opt-Outs9 in
- Auditing a Vendor’s AI Claims8 in
- How to Check a Citation an AI Gave You8 in
- How to Read a Number Before You Repeat It8 in
- HOW MUCH IS BOTS — How Much of the Internet Is Bots? The Sourced Answer7 in
- How to Check a Viral Screenshot Before You Share It7 in
- The Household Guide — Protecting Your Family From AI Scams7 in
- Using AI at Work — What Not to Paste, and What to Disclose7 in
- Claude’s Text Watermark: What It Is, What It Can Prove, and Why the EU Requires It6 in
- Common AI Misconceptions, and What Is Actually True5 in
- How to Search Now — Finding Real Things in an Answered Internet5 in
- Press and Citation — How to Quote This Site5 in
- The Cost of Checking Too Much5 in
- Cloning a Voice or a Face: What the Law Says4 in
- Follow the Record — How to Get Updates4 in
- Is Your AI Output Usable? Alt Text, Captions and Contrast4 in
- AI Chatbots and Kids: Ages, Parental Controls and What to Say3 in
- Letting an AI Browse For You3 in
- Moltbook: The Social Network Humans Are Not Allowed to Post On3 in
- Vibe Coding: What It Is and What It Costs3 in
- A Team AI Policy You Can Actually Adopt2 in
- Confirmation bias, repetition and AI: six checks on yourself2 in
- Grok and the Deepfake Investigations: What Regulators Actually Did2 in
- How Misinformation Spreads, and What Slows It Down2 in
- How to Spot Fake Reviews, Including AI-Written Ones2 in
- The Record2 in
- When Claude Models Reached Real Systems Without Permission: What Anthropic Reported2 in
- Using AI on a Job Application1 in
Blog 16
- HOW AI GETS LABELLED — How AI Content Gets Labelled18 in
- AI Agents — What They Actually Do, and Where They Fail13 in
- The Rules Arriving — What AI Law Actually Requires Now9 in
- AI Companions — What the Research Actually Shows7 in
- What AI Costs in Energy and Water — The Contested Numbers7 in
- AI and the Open Web — Who Pays for the Pages Now?6 in
- Seven Repos and What They Actually Cost to Run6 in
- AI and Work — What the Evidence Actually Shows5 in
- AI in Schools — What Teachers, Students and Parents Should Know5 in
- The Timeline — Documented Moments, 2022 to Now5 in
- How to Make Money with AI Content4 in
- How to Start an AI Influencer for Free (and When You Will Need to Pay)3 in
- The Seven Ways People Sell AI Work3 in
- Higgsfield MCP + Claude: Make Motion Design Ads With AI2 in
- The Best Negative Prompts for Every AI Model — Copy and Paste2 in
- Blog—
Worlds 15
What this is not
It is decoration over a working page. The index below is complete HTML — every page, every title, every link. The map is drawn on a canvas on top of it, and a canvas contains no text: nothing in it is readable by a search engine or a screen reader. That is the whole reason the list stays.
It does not load on phones, where it would cost battery to render something too dense to read on a small screen. It respects reduced-motion: nothing drifts on its own, and choosing a page puts you there instead of flying you there. And it is 44 KB uncompressed — 25 KB of renderer and 19 KB of graph, about 17 KB over the wire — with no 3D library at all. The article on 3D websites explains why that number matters more than how it looks.