About this book
This book is a projection of graphs.sgit.ai: the
same content, in a reading order, in three formats. The chain has three links and one
source of truth: the chapter text is authored in markdown (each chapter is
fetchable at https://graphs.sgit.ai/content/<chapter>.md), the site pages are rendered
from it, and this book is generated from those pages by
admin/build/gen_book.py. The introduction is projected the same way from
the site's front page, so the book opens exactly where the site does. Nothing can
drift: the build fails if a page lags its markdown or the book lags a page. That is the book's own argument — documents
are projections of graphs — applied to the book itself.
It was written by Dinis Cruz together with a team of AI agents, from voice memos developed into more than 1,300 structured briefs inside a corpus of some 3,300 documents; the full statement of how, with the numbers and their sources, is in the closing chapter, the author's interest.
The chapters therefore speak as the site speaks: where a page says “this site”, it means graphs.sgit.ai, of which this book is a view — and where a page says “this page”, the book says “this chapter”, because the projection rewrites self-references the same way it rewrites links. The site is the living version; the book is the version you can read on a train.
Four ways to read it, same content in all four — the two PDFs are regenerated together from the same chapters on every release, and each carries the site version it was generated from on its cover:
- Chapter pages: one chapter per page, with the table of contents beside you.
- One single page: the whole book in one HTML file, for reading straight through, searching with ctrl-F, or fetching once.
- The screen PDF: the single page printed at US Letter in the site's own design. Colour, one column, comfortable on a tablet.
- The print PDF: a real 6″ × 9″ print interior with gutters, folios and a paginated contents; the edition a print-on-demand service takes.
The print edition, and the formats deliberately not offered
There are two PDF editions, generated together from the same chapters. The screen edition is the site's own design at US Letter, the one to read on a tablet. The print edition is not a printout of the web pages — it is a print interior in the standard technical-book format, ready for print-on-demand (KDP and equivalents):
- 6″ × 9″ trim, the standard trade size, with no bleed — nothing runs to the page edge.
- Mirrored margins: 0.75″ inside (the gutter), 0.5″ outside — above KDP's minimums for a no-bleed interior of this length (0.375″ / 0.25″).
- Folios and running heads: page numbers sit bottom-outside; the verso head carries the book title, the recto head the chapter title; front matter and part pages carry neither.
- A paginated contents — the page numbers in it are computed by the typesetter at render time, so they cannot disagree with the pagination. The same no-drift rule as everything else here.
- Justified, hyphenated, monochrome, with every font embedded. Colour exists only in the screen editions; ink is black.
The cover is generated alongside the interior — as SVG, by
gen_cover.py: the front cover (the artwork on
this page), and the full print
wrap — back, spine and front with bleed — whose spine width is computed from the
interior's page count, so when the book grows the spine follows and the build fails if
it does not. The graph on the cover is not decoration: it is a true subgraph in the
book's own edge vocabulary, six edges reading as sentences and one ghosted, unanswered
edge — the palette rule the book teaches, applied to its own cover.
One thing is deliberately not here: a Kindle edition — a fixed-layout PDF is the wrong upload for Kindle in every case; the right artefact is a reflowable EPUB, which this pipeline could generate from the same chapters, and which is queued rather than pretended to.
Edition: site v0.3.5, 21 August 2026. All content is released under the Creative Commons Attribution 4.0 International licence (CC BY 4.0). The raw source documents behind every chapter are published at graphs.sgit.ai/documents/.