Skip to content

Book Bible

The single source of truth for what this book is, who it's for, and how it sounds.

Working title

The House Edge: How On-Chain Trading Really Works

(Alternates to consider: Following the Dollar; The Plumbing of On-Chain Markets; Who Pays the Spread.)

One-sentence pitch

A business reader's guide to on-chain trading that follows a single dollar from a trader's wallet to its final resting place — exposing, at each step, which actors took a cut, why, and who they took it from.

The reader

A composite of three real people:

  • Sales lead at a fintech who needs to talk to crypto-native clients without being snowed
  • Board member at a custodian, infrastructure firm, or asset manager trying to evaluate the space strategically
  • Senior associate at an allocator doing diligence on a crypto fund or infrastructure investment

What they share: numerate, skeptical, time-constrained, no patience for jargon, will not finish the book if it sounds like a pitch deck. They've read one Matt Levine column on crypto and remember most of it.

What they don't know: anything about wallets, gas, MEV, mempools, validators, LPs, AMMs, CLOBs, or the names of any protocol. Assume zero.

What they do know: how a stock exchange works, what a market maker is, what payment for order flow is, what HFT is, why dark pools exist, what a prime broker does. The TradFi parallels are your best friend.

The through-line

The book makes one argument, told three ways:

On-chain trading is a layered economic game in which most participants are systematically disadvantaged in ways they don't see, and a small number of actors capture most of the value. The architecture of each chain determines who those actors are and how much they capture.

  • Follow-the-money spine (primary): A dollar of trader intent gets converted into a dollar of settled position, but loses a few cents to every actor it passes through. Each chapter dissects one layer of the conversion.
  • Actor-by-actor lens (recurring): Each chapter introduces or revisits a specific actor — retail trader, market maker, searcher, validator, builder, infrastructure provider, exclusive-flow desk — and explains their economics from their own seat.
  • Chain-by-chain comparison (recurring): At the end of every relevant chapter, a "How this plays out on Solana, Hyperliquid, and Ethereum" section shows the same mechanic instantiated differently and which actors win or lose under each design.

What the book is selling (in the marketing sense)

Even though this is written as neutral industry education, Figment's commercial interest is that once a reader understands on-chain market structure, they understand why validator infrastructure is no longer the product — execution and blockspace are. Readers who internalize the book's argument are pre-sold on the conversations Figment sales wants to have with them. This is why the book travels as a marketing asset without containing marketing.

The epilogue is the only place where Figment is named.

Recurring devices

Use these on purpose. They give the book texture and they help the reader retain.

  • "Meet the actor" sidebars. When a new actor type first appears, a half-page sidebar in a different visual style: name, job description, how they make money, who they take it from, their TradFi analogue. Reusable as the actor returns in later chapters.
  • "Show the dollar" callouts. A small numbered diagram showing a $10,000 trade and where each cent ends up. Updated as the chapter introduces new extraction layers.
  • "On Solana / On Hyperliquid / On Ethereum" comparison boxes. End-of-section, three short paragraphs.
  • "What changes when..." thought experiments. Single-paragraph counterfactuals at the end of each chapter ("What changes when a validator runs a privileged client?"; "What changes when the mempool is private?"). These are the seeds for the next chapter.
  • The Goldman MD test. Before sending a chapter for review, the agent should imagine reading it aloud to a Goldman MD over coffee. Anything that would make her check her phone gets cut.

Banned moves

  • Defining jargon with jargon. ("MEV is MEV extraction.")
  • Hype words: revolutionary, game-changing, paradigm-shifting, the future of, disrupting.
  • Doom words: rugged, broken, dead, scam (unless naming a specific actual scam by name with a source).
  • Tribal endorsements or insults of specific chains.
  • Anything that sounds like investment advice.
  • Padding paragraphs that restate the previous paragraph.
  • Acronyms without first-use definition.
  • Tweets cited as authoritative sources.
  • "It's important to note that..." and similar throat-clearing.
  • Comparing crypto to "Web 1.0" or "the early internet" — readers have heard it 200 times.

Encouraged moves

  • TradFi analogies, especially uncomfortable ones (sandwich attacks ≈ front-running by a privileged HFT desk; exclusive order flow ≈ Citadel paying Robinhood; validators ≈ NYSE specialists with extra powers).
  • Naming names. Specific protocols, specific firms (Jump, Wintermute, Jito, Temporal, Citadel, Jane Street), specific incidents (Knight Capital, the 2024 Solana congestion, the Mango exploit).
  • Concrete dollar examples ("on a $10,000 swap...").
  • Short sentences for verdicts. ("That money came from the LPs.")
  • Footnotes for the technically curious. They keep the main text clean.

Voice samples

These are the calibration targets. When in doubt, read these aloud and match.

Sample A — explaining MEV to a business reader

Every block on Ethereum is built by someone. Until 2022, that someone was a validator running default software, picking transactions roughly in the order they arrived. Today, the validator outsources block-building to a specialist — a "builder" — who runs sophisticated software designed to do one thing: maximize the dollar value of the block before it's sealed. The builder reads every pending transaction, looks for opportunities to insert profitable trades around them, sells those positions to the highest bidder, and pays the validator a kickback. The trades being inserted around are yours. The profit being extracted comes out of your execution price. The kickback to the validator is the share of your money that bought their cooperation.

Sample B — introducing a worked example

Imagine you're swapping $10,000 of USDC for SOL on a Solana DEX aggregator. The aggregator finds you a route, your wallet signs the transaction, and ninety seconds later you have SOL in your account. The interface tells you the trade succeeded. What the interface does not tell you is that between the moment you clicked "swap" and the moment your transaction was included in a block, four other actors saw your order, two of them placed trades designed to profit from yours, and one of them paid the validator who built the block a fee for letting them go first. By the time your SOL arrived, $73 of your $10,000 was gone — not to fees you agreed to, but to a chain of small extractions you never saw.

Sample C — the "who pays" verdict

The numbers in this chapter don't add up to a controversy. They add up to a market. Searchers earn because they spotted the opportunity. Validators earn because they sold the privilege. Builders earn because they did the routing. The losses are real, but they are also distributed — fractions of a percent on millions of trades, paid by tens of thousands of users who will never read this book. That is the structure on-chain trading has converged toward, on every major chain, with local variations. The next chapter explains the variations.

Length and scope

  • Target length: 60,000–80,000 words, roughly a 200-page trade paperback equivalent.
  • Chapter count: 12 chapters plus prologue, glossary, and epilogue. Not 20. Not 8.
  • Per-chapter length: 4,000–6,000 words. Chapters longer than 6,000 are signal that the chapter is actually two chapters.
  • Diagrams per chapter: 2–5. Not zero, not ten.
  • Worked examples per chapter: 1 anchoring example threaded through, plus optional cameos.

What "done" looks like

A complete book/ directory with:

  • OUTLINE.md reflecting the final TOC
  • Twelve chapter folders, each with RESEARCH.md, OUTLINE.md, DRAFT.md, REVIEW_NOTES.md
  • A unified book/MANUSCRIPT.md concatenating all chapters in order
  • glossary/GLOSSARY.md with every term used in the book
  • book/assets/ with all diagrams
  • book/sources/ with a deduplicated bibliography
  • A short EPILOGUE.md written by Nick, not the agent