The first time the graph finished loading, it had arranged itself into a spiral. Nobody designed that. The layout is a physics simulation: every node pushes the others away, every connection pulls its two ends together, and the whole thing settles wherever the forces cancel out. The shape that came out looks like a galaxy because a galaxy is solving the same problem. Dense bright cores where the mass concentrates, long arms where it thins, and a lot of dark between.
That was the first evidence we had that the book holds together. You cannot fake this shape. If the chapters were a list of separate essays, the graph would look like twenty-one islands. It does not.
What you are looking at
Click any node and a panel opens with what it is, where it appears, and what it touches. Click a legend colour to isolate one kind and watch everything else fade. Scroll to zoom, drag to move, press / to jump to the search box. Size means editorial weight, so the biggest circles are the ideas the book leans on hardest. Line thickness means the same thing for connections.
The graph speaks English, Spanish, Romanian and Portuguese. It remembers which one you picked on any other page of the site.
Three ways in
You finished a chapter and want the roots
Search the chapter number. The panel shows every thinker and concept that chapter stands on, and each one is a door. Ch.19 leans on Arendt, and Arendt connects to Labour, Work and Action, and that concept runs into Ch.14 through a different argument about power. You arrived by reading. You leave by a route the chapter never mentioned.
You have a question and no chapter
Start at the concept instead. Search attention, or memory, or trust. The graph will tell you which chapters carry it and which thinkers shaped it. This is the honest way to use a book you have not read yet, and it works from any direction.
You want to wander
Pick anything. Follow whatever connection looks strange. There is no wrong path because there is no destination, and the graph will keep handing you a next node for as long as you keep clicking. Some evenings that is the right way to read.
Then take it somewhere
Here is where the graph stops being a diagram. Every chapter page on this site carries a prompt built for that chapter and a button that opens it in Claude, ChatGPT, Gemini or Perplexity. You can copy it, edit it, throw half of it away. Bring the node you were looking at. Argue with it.
The book's own method has three moves and only three. You generate the starting point. The AI reframes rather than answers. You decide. Run that on any node in this graph and you will end up somewhere neither of us wrote, which is the entire argument the book has been making for twenty-one chapters. Intelligence is a property of the configuration. Two nodes, and a third thing between them that belongs to neither.
We have a name for that third thing in the book. Reading about it and doing it are different activities, and only one of them changes you.
Send back what you find
If a conversation you had off the back of this book went somewhere good, we would like to read it. Some of what readers build will be sharper than what is in these chapters, and the honest thing to do with a better idea is to say where it came from.
Sharing is a deliberate act, never a background one. Nothing you type into Claude or ChatGPT reaches us, because those conversations happen on their servers and not on this site. If you want us to see one, you send it. You choose whether to be credited by name or stay anonymous, and you can withdraw it later. Write to us while the submission form is being built.