---
title: Zoom in, not discover: how this book is written, with the punchline first, every first page a summary of the whole, the reader doing something early, milestones we can map, and my story at the end for anyone who wants it
date: 2026-10-19
author: Dinis Cruz
summary: This is the guidance for how the book is written, and the principle is the same one as for meetings and for every communication. The first ten seconds buy you the next twenty, the twenty seconds buy the next minute, the minute buys the next ten minutes, and the ten minutes buy the hour. So there should be no punchline waiting for the reader at the end. The book is a zoom in, not a discovery. The opening is a best of, a narrative of the summary of every section: here is the opportunity, here is what you are going to do, here is what we think you should do, here is why you are lucky to be alive now. My story, where this came from, goes at the end, for the reader who wants it. There is a test for the opening that we can run on the graph: what the reader reads first should link to the ideas, and the ideas should link to every chapter in the book; if a chapter is not reached, we are missing something. Today the opening of release candidate 2 reaches one chapter of 42 that way. And because the whole point of the book is that you learn by doing, the reader should start doing early, so we map the milestones: what we would like you to be able to do, and to think, after each part, and we ask the readers whether they got there. If I hand somebody the first page, they should see the art of the possible and the punchline of what we are trying to do. They may not know how yet, which is what the book is for, and they may want to challenge it, which is fine.
version: v0.2.34
tags: writing, structure, zoom, milestones, opening, gen-z-book, article
status: published
---

## In short {#in-short}

- **Every step of attention buys the next.** The title buys ten seconds, ten seconds buy twenty, twenty buy a minute, a minute buys ten, ten buy the hour.
- **No punchline waiting at the end.** The book zooms in; it does not make the reader discover.
- **The opening is a best of.** A narrative of the summary of every section: the opportunity, what you will do, what we think you should do, why you are lucky.
- **My story goes last.** How we got here is at the end of a chapter, for the reader who wants it.
- **The graph test.** The opening links to the ideas, the ideas link to every chapter; a chapter not reached means something is missing. rc2's opening reaches one of 42.
- **A manual, not a journey.** You learn by doing, so you start doing early.
- **Map the milestones.** What you should be able to do and think after each part, and ask the readers if they got there.
- **The first page shows the art of the possible.** The reader may not know how yet, and may disagree; they should see the punchline.

## Ten seconds buy twenty {#attention}

The key principle is the same one I use for meetings and for every communication. The title is the first ten seconds, and the first ten seconds buy you the next twenty. The next twenty seconds buy the next minute, the minute buys the next ten minutes, and the ten minutes buy the hour. At every step the reader decides whether to give you the next one, and they decide on what they have already got, not on what you promise is coming.

So a book that saves its point for the end is asking for the whole hour on credit. Nobody gives that any more, and they are right not to.

## Zoom in, not discover {#zoom}

That is why I want this book written with no punchline waiting for the reader at the end. Not a TL;DR, and not a book of bullet points, but a book where the point comes first and everything after it is a closer look. I talked about a plot line before; the key concept is that it is more of a zoom in than a discovery.

The newsroom already writes this way: every article opens with its whole argument in a few lines, and the sections below are the detail for whoever wants it. And it is the same idea as the fractal graphs in [everything is connected](/articles/everything-is-connected.html): "There is no top and no bottom. There is only the altitude you happen to be looking from, and how much definition you choose to load at it." The first page of the book is the highest altitude. Each part, each chapter and each section is the same picture at a closer zoom, and a reader can stop at any altitude and still have the whole of it, at that resolution.

## The opening is a best of {#opening}

The whole first section, where I start to explain that this comes from another book, and then I did this and I did that: the reader is not interested in my story, not at that point. That is for the latter part of the chapter, almost at the end, where it can say: here is how we got here, if you are interested.

What matters is that the introduction is just a best of. Here is the point. You are doing this, you are going to do that, this is the opportunity, this is what we think you should do, this is why you are lucky to be alive now. It should almost be a narrative of the summary of every section. Every chapter should follow the same shape, at its own altitude:

| | What it is | What it is not |
|---|---|---|
| **1. The point** | The chapter's argument, in a few lines a reader could repeat | A scene, a quote from 2018, an anecdote |
| **2. What you will do** | The outcome: what you will be able to do, or will have, at the end | A list of topics |
| **3. The zoom** | The how, closer and closer, each section a closer look at the point | New points that the first lines did not mention |
| **4. Do it now** | One thing to do for your company, today | "We will come back to this" |
| **5. How we got here** | My story, the 2018 book, the evidence trail, for anyone who wants it | Required reading |

## The graph test {#graph-test}

This is the part I like most, because we can check it. If we take a graph of the first thing the reader reads, that text should have a link to the ideas, and those ideas should link to every single chapter in the book. If a chapter is not reached, we are missing something: either the opening has not said the thing that chapter is for, or the chapter is not for anything the book promises. And the opposite holds too: a promise in the opening that leads to no chapter is a promise the book does not keep.

The book is already a graph. Every chapter names the ideas it was written from, so the test can be run today. On release candidate 2, the opening chapter names two ideas, and between them they reach one chapter of the 42: itself. Nine chapters, most of the hands-on part among them, name no ideas at all, so nothing could reach them. That is not a criticism of the drafting: nobody asked for this until now. It is the measure, and it should go to 42 of 42 before the book is signed off.

## A manual, not a journey {#manual}

We need to focus on the practicality. Somebody picks up the book and goes: yes, I want to learn this. And then they can zoom in, and from there it is almost a manual of how to do it. It is less about the journey. I think that is a different book. We have enough meat here, and enough practicality, that our job is to get the reader on it.

And the whole point of this book is that you learn by doing. If you have to wait a long time before you start doing, that misses the point. The readers of release candidate 2 told us the same thing in their own way: more than 8,700 words of the first part come after the first chapter and before the reader is asked to do anything, and the first afternoon of real work is about 26,000 words in.

## Map the milestones {#milestones}

What I like about our world now is that we can literally map the milestones. So we should: this is what we would like you to be able to do, this is what we would like you to think, after this particular part. Define the places in the book where an objective should be achieved, and then ask the readers, at every one of those milestones, whether they got there.

A first map, for the editor to correct:

| After | You can | You think |
|---|---|---|
| The first page | Say what the book will get you to, and what you would build | This is possible, and it is possible for me |
| Before you start | Name your problem, and where you stand on the map | I am early, not late |
| The frame | Explain a model, a role and an agent; brief your first agent | I am a developer, whether or not I write code |
| Working with agents | Run your first workspace, with a policy and a record | The agents work for me, inside limits I set |
| Direction and judgement | Name the stop before you start; check a claim back to its source | My judgement is the scarce part |
| Start the company | Ship something to someone, run the miss-it test, set a price, register the company | I have a company, or I know exactly what is missing |
| Build it without a server | Ship a product that runs with no server of mine, with the user's data private and the agents bounded | It costs nothing when nobody uses it |
| Hands on, 2026 | Each workflow, done once | I can do this again without the book |

The synthetic readers can be asked this at every milestone, and so can the real ones: what can you do now that you could not do before this part? A milestone that nobody reaches is a finding.

## The first page {#first-page}

If I just handed somebody the first page of the book, they should be able to answer a whole bunch of questions:

- What is the opportunity, and why now?
- Who is this for, and is it me?
- What will I have done by the end?
- What do I do first?
- What will it cost me, in money and in time?
- What do I need to know already? (Almost nothing.)
- What does this book think that I might disagree with?

They should know the art of the possible. They might not know how to do it, which is fine: that is what the book covers. They might want to challenge some of the ideas, because they go: ooh, I am not sure about that. That is fine too. But they should see the punchline of what we are trying to do, on the first page, and every page after it should be a closer look.

## The brief {#brief}

For the editor. Read `team/editor/role.md` and this article first.

1. **Make this the writing contract for rc3**: every chapter in the five-part shape above, the point first, one thing to do for your company, and the story at the end. Done when the contract is in your `briefs/` and every chapter of rc3 follows it.
2. **Rewrite the opening as a best of**: a narrative of the summary of every part, with no 2018 story before the point, and the first page able to answer the seven questions above. Done when the opening passes the graph test.
3. **Correct the milestone map** and put each milestone at the end of its part, as the question the reader can answer. Done when the map is in the book.
4. **Ask the readers at the milestones**: the next synthetic readers run reports, at each milestone, what the reader can now do. Done when the study on rc3 has a row per milestone.

For the librarian:

5. **Make the graph test a check**: the opening's ideas, and the chapters those ideas reach, with the build reporting the chapters not reached and the chapters with no ideas. Done when the build prints the coverage, and rc3's is 42 of 42 (or the chapter count it has).

## What this does not claim {#limits}

- **"Ten seconds buy twenty" is a rule of thumb**, mine, from meetings and communications, not a measured law of attention.
- **The coverage figure is from the chapters' front matter** on release candidate 2: two ideas named by the opening, the chapters that name the same ideas, and nine chapters that name none. A chapter can serve the opening without sharing an idea id; the test makes that link explicit.
- **The milestone map is a first draft.** The editor owns it.
- **The 8,700 and 26,000 words are the editor's measurements** in the study of release candidate 2.
- **Zooming in does not mean no story.** The story stays, at the end, for whoever wants it.

## Where this comes from {#sources}

A voice memo of mine on 19 October 2026, written up by this site's journalist, `gen-z.journalist`, in my voice, keeping the memo's order and phrases. The argument and the editorial responsibility are mine.

The coverage was computed from the chapter files in `v2/content/` (their `ideas` front matter) at this release. The word counts are from [five readers who do not exist](/agent-articles/five-readers-who-do-not-exist.html) and its synthesis. The fractal graphs quote is from the newsroom's [introducing fractal semantic graphs](https://newsroom.sgit.ai/articles/introducing-fractal-semantic-graphs.html), as quoted in [everything is connected](/articles/everything-is-connected.html).