The House Edge

Appearance

Chapter 1 — Outline (Phase 2) ​

Status: OUTLINE (Phase 2 — written by agent, awaiting Nick's review) Date: 2026-05-13 Working chapter: 01 — What Is a Trade, On-Chain Carries forward from: RESEARCH.md

Working assumptions (carried over from Phase 1) ​

Defaults from RESEARCH.md §6, treated as binding for Phase 3 unless Nick overrides.

  1. Alpenglow treatment: Use the current production figure (Solana finality at ~12.8 seconds) as the chapter's anchor number. One clean forward-looking sentence near the end of the finality section notes that Alpenglow, which targets ~150ms finality, is live on testnet as of mid-2026 with mainnet activation expected later in the year. The Bible's "no breaking news" rule binds: the chapter should still read correctly in 2027 if Alpenglow ships on schedule, and still read correctly if it slips.
  2. Ch 2 back-edits: Patched as part of this chapter's Phase 3 work. Two corrections — Hyperliquid taker fee 3.2 → 4.5 bps, finality "~70ms" → "200ms median end-to-end / 900ms p99."
  3. Worked example: Candidate C — side-by-side Coinbase + Solana, both walking Alice's $10,000 USDC→SOL swap through the three phases.
  4. TradFi parallel weight: Heavy. The chapter is standalone for a reader who hasn't met Chs 2 or 4.
  5. Failed-tx framing: Quote the 70% Solana headline once with the 92% bot caveat right next to it.
  6. CEX anchor: Coinbase (Advanced Trade specifically, since that's where its CLOB lives).
  7. Chapter length: Target ~3,800 words including footnotes — lighter than Ch 2 (5,053) and Ch 4 (4,892). The chapter is structural setup; it doesn't need the same mechanical depth.

Title and subtitle ​

Chapter 1 — What Is a Trade, On-ChainThree phases that happen to every trade, and what the chain does to each one.

Cold open ​

When the New York Stock Exchange moved its settlement cycle from T+2 to T+1 on 28 May 2024, the transition took ten years to negotiate and required coordinated upgrades across hundreds of firms and the DTCC clearinghouse. The change meant that an equity trade executed on a Tuesday would now finalise on Wednesday rather than Thursday — a one-day reduction in the gap between an action and its irrevocability. On the same Tuesday afternoon, hundreds of thousands of trades on Solana and Ethereum finalised in seconds. This chapter is about what those on-chain trades actually look like: the three phases every trade passes through, what the chain does at each phase, and where the costs go in between.

(Why this open: the T+1 transition is a specific, dated, primary-sourced event the reader can verify; the contrast between "ten years to shorten settlement by one day" and "seconds to final" is vivid; and the closing sentence lays out the chapter's job in the Ch 4 pattern.)

What this chapter answers ​

  • What is a trade, in plain terms — and what makes "on-chain" different from "on Coinbase"?
  • What are the three phases every trade passes through, on every venue, and what does the chain do at each phase?
  • What is the public mempool, and why is its existence the precondition for everything that follows in this book?
  • What does the trader actually pay — at each phase, on each kind of venue — and to whom?

Section list (one-sentence summaries) ​

  1. Cold open — T+1 vs on-chain seconds; sets up the timing-and-mechanics question.
  2. What this chapter answers — the four questions above.
  3. The setup (≈500 words). The three phases as a universal concept: every trade has intent, settlement, and finality. The TradFi version (Coinbase Advanced Trade, NYSE, DTCC); the on-chain version (Solana, Hyperliquid, Ethereum and L2s). Plant: "we are about to follow Alice through both."
  4. The worked example (≈350 words). Alice runs the same $10,000 swap twice — once on Coinbase Advanced Trade, once on Solana via Jupiter — and we watch each one phase by phase, with the chapter showing what is identical and what is different. The example carries the chapter; each subsection of §5 returns to it.
  5. The mechanics, in detail (≈2,200 words; three H4 subsections — one per phase):
    • Intent (≈700 words). On Coinbase: Alice opens the app, types in $10,000, sees the route, clicks Buy. On Solana: she opens a wallet, signs a transaction Jupiter has constructed. What "intent" means on each side, who can see it (no one but Coinbase on the CEX side; potentially anyone running an RPC node on the Solana side, with the Gulf Stream caveat), and what it costs (nothing yet — intent is free on both venues). The mempool concept is defined here, in cameo, with the explicit forward link to Chapter 3.
    • Settlement (≈800 words). On Coinbase: the matching engine runs an anonymous FIFO match against the on-exchange CLOB; the trade fills against another user's resting order (or against Coinbase's house book on Pro-style venues that have one). On Solana: Alice's transaction enters the slot leader's pipeline, is included in a block (~0.6 seconds), and Alice's $10,000 has now been exchanged for some amount of SOL. The chapter walks the costs at settlement: on Coinbase, the spread plus the explicit taker fee (combined ~0.5% retail / 0.045% institutional); on Solana, the base fee plus priority fee plus any Jito tip plus the slippage Alice paid against the pool (collectively in the dollars, not the cents, on a $10,000 trade). Coinbase does not use payment-for-order-flow for retail spot — useful clarification, since the Bible favours TradFi parallels and PFOF is the one the reader might assume by analogy.
    • Finality (≈700 words). On Coinbase: instant for the user (the trade is reflected in their account immediately); legally, the cash-and-asset transfer between Coinbase's accounts and external counterparties may settle on T+1 if Coinbase chose to use an external clearinghouse. On Solana: probabilistic — confirmed (~0.6 seconds, 66% stake vote), finalized (~12.8 seconds, 32 slots past). On Ethereum L1: ~12.8 minutes via Casper FFG. On Hyperliquid: ~200 milliseconds end-to-end median. On Ethereum L2s: ~250 milliseconds to ~2 seconds for sequencer-soft confirmation, with the seven-day fraud-proof window for canonical L1 settlement on optimistic rollups. The chapter lands the structural point: finality is not a single number; it is a curve from "the venue says it's done" to "no one can roll it back." Alpenglow is named in one sentence, framed as shipping but not yet shipped.
  6. How this plays out on each chain (≈350 words). The required chain-comparison box. Three paragraphs: Solana (12.8s finality production, Alpenglow shipping to ~150ms; gas now ~$0.017 average; ~67% of user txns pay priority fees; failed-tx rates 2024-peak 70% but ~92% bot-driven); Hyperliquid (200ms median end-to-end finality, no public mempool, 4.5 bps base taker fee, the architecture is a CLOB rather than chain-with-AMMs); Ethereum and L2s (12.8 minutes to FFG finality on L1, ~$0.25–$0.39 typical Uniswap swap, L2 swaps ~$0.01–$0.30; the 7-day fraud-proof window for optimistic rollup canonical settlement).
  7. Who wins, who loses, why (≈250 words). A structural verdict — the chapter is not adversarial. Winners: traders who use the right venue for their job (which depends on size, asset, and tolerance for latency); the venues themselves, all of which capture fees in different ways; the validators and sequencers who include the trades. Losers: traders who route to the wrong venue (large size into a thin AMM, small size onto a CEX with high explicit fees, a chain whose finality timeline doesn't match their settlement needs). The honest answer to "is this bad": no — the three-phase structure works on every venue, and the differences are mechanical rather than moral. The next several chapters are about who is privileged at each phase and what those privileges cost everyone else. Chapter 1's verdict is restraint: do not draw conclusions until we have the actors on the board.
  8. What changes when… (one paragraph). The transition: what changes when the trader's intent is public before settlement happens — when anyone running an RPC node can see what Alice is about to do and act on it first? That is Chapter 3.
  9. Footnotes and sources — numbered with URLs and access dates. Approximately 12 footnotes.

The worked example and where it threads through ​

Alice's $10,000 USDC→SOL swap on (a) Coinbase Advanced Trade and (b) Solana via Jupiter, February 2026.

SectionBeat
§3 (Setup)Plant: "we'll follow Alice through both — what is identical, what is different."
§4 (Worked example)Setup only: Alice clicks Buy on Coinbase, then opens her wallet and clicks Swap on Jupiter. Both interfaces tell her the trade succeeded within seconds. The chapter asks: did the trades happen the same way?
§5 IntentOn Coinbase: she sees a price quote and a confirm button. On Solana: she sees a route, a slippage tolerance, and a signature prompt. Both intents are equally "stated"; one is visible only to Coinbase, the other potentially visible to anyone watching the RPC layer.
§5 SettlementOn Coinbase: her order matches against another user's resting bid; the spread and the explicit fee are the bill. On Solana: her transaction is included in a block 0.6 seconds after she signed; the base fee, priority fee, and pool slippage are the bill. Both venues took roughly the same time; they did very different things.
§5 FinalityOn Coinbase: Alice's app shows the trade as done instantly. On Solana: the chapter walks the four numbers (processed 0.4s / confirmed 0.6s / finalized 12.8s; Alpenglow shipping to ~150ms).
§6 (Chain comparison)The same trade on Hyperliquid (200ms), on Ethereum L1 (~12.8 minutes to FFG), on Base (~2s sequencer + 7-day fraud window).
§7 (Verdict)The structural conclusion: no venue is universally better; each handles the three phases differently, and the right choice depends on what the trader needs.

Diagrams needed ​

Three.

  1. D1 — The three phases of a trade, parallel tracks (Mermaid sequence). A sequence diagram with two participant groups: Coinbase-track (Alice, Coinbase Matching Engine, Coinbase Account) and Solana-track (Alice, Solana RPC, Slot Leader, Block). Both tracks walk left-to-right through intent → settlement → finality. The visible difference: Solana's track has a "Pending state, visible to RPC observers" arrow in the middle that Coinbase's track does not.

  2. D2 — Alice's cost at each phase, on each venue (markdown table). Rows: Intent / Settlement / Finality. Columns: Coinbase Advanced Trade ($), Solana via Jupiter ($). Cells: explicit dollar amounts (using approximate 2026 numbers — taker fee on Coinbase ≈ $25 on a $10K trade at 25 bps; on Solana ≈ $0.02 base fee + $0.30 priority fee + ~$50 pool slippage on a clean major-pair swap). The table lets the reader see what each phase costs concretely.

  3. D3 — Finality timeline by chain (Mermaid xychart or static SVG). A horizontal timeline showing wall-clock seconds, with markers at: Hyperliquid 0.2s, Solana confirmed 0.6s, Solana finalized 12.8s, Ethereum L1 ~12.8 minutes (≈768s — break the timeline scale or use log-axis), Coinbase instant-user-final / T+1-external-final. Visual punchline: the on-chain finality range spans three to four orders of magnitude across chains.

Glossary terms this chapter introduces ​

To be appended to GLOSSARY.md in Phase 3.

Defined in full (first appearance):

  • Trade — the act of exchanging one asset for another at an agreed price.
  • Transaction — the on-chain instruction that, when executed, settles a trade (or transfers funds, or interacts with a smart contract).
  • Intent — the trader's stated desire to make a trade, before the trade is matched or executed.
  • Settlement — the moment the trade is finalised at the venue level (matched on a CEX; included in a block on-chain). The TradFi sense (T+1 cash-and-asset transfer at a clearinghouse) is named and contrasted.
  • Finality — the moment the trade is irrevocable; can be instant (on a CEX, from the user's perspective), probabilistic (on most chains, with chain-specific time windows), or deferred (on optimistic rollups, where canonical settlement waits seven days).
  • Wallet — the user-facing software that holds the trader's private keys and signs transactions on their behalf.
  • Gas — the unit of compute work a transaction consumes, priced by the chain in its native token; the user pays it whether the transaction succeeds or fails.
  • Block — the unit of transaction batching on most chains; transactions are grouped into blocks and validated together.

Cameo (one inline sentence; full glossary entry but no extended prose treatment):

  • Validator / Sequencer (full sidebar in Ch 5)
  • Mempool (full treatment in Ch 3)
  • Slot leader (Solana-specific; one inline mention)
  • TPU / Gulf Stream (Solana-specific; mentioned only as "Solana's pre-leader transmission path")
  • Casper FFG (Ethereum-specific; one inline mention)
  • Sequencer soft confirmation (L2-specific; one inline mention)
  • Fraud-proof window (optimistic rollups; one inline mention)

Backward:

  • Prologue (not yet drafted): Alice's $10K swap is introduced. If the prologue is not ready when Ch 1 enters Phase 3, the chapter must briefly establish Alice and her swap inline — one or two sentences carry the freight.

Forward:

  • Chapter 2 (Where Liquidity Lives, drafted): the pool Alice's swap lands in is the subject of Ch 2. The chapter can reference Ch 2's findings (the prop-AMM displacement, the passive-LP role) only in passing — Ch 1 must stand alone for a reader who hasn't read Ch 2 yet.
  • Chapter 3 (The Mempool and What Replaces It, not yet drafted): the mempool cameo in §5 Intent is the direct lead-in. §8's "What changes when…" tees up Chapter 3 explicitly.
  • Chapter 4 (The Searcher, drafted): the searchers who watch the mempool are Chapter 4. Ch 1 plants visibility as a concept (one sentence) but does not develop it.
  • Chapter 5 (Validator and Builder): the validator who included Alice's transaction is named here in cameo.

Diagrams generated where in Phase 3 ​

  • D1: inline Mermaid sequence diagram in §5.
  • D2: inline markdown table at the bottom of §4 or top of §5 Settlement.
  • D3: inline Mermaid xychart-beta plus a static SVG saved to book/assets/01_trade/finality-timeline.svg for the pandoc pipeline.

Ch 2 back-edits scoped for Phase 3 ​

While drafting Ch 1, two corrections to the Ch 2 draft need to land:

  • Ch 2 §5b (line ~88): "HyperBFT, which produces blocks at approximately seventy-millisecond intervals" → revise to: "HyperBFT, which produces ~200ms median end-to-end latency (~900ms p99) for a co-located client" (with citation to Hyperliquid's own HyperCore docs).
  • Ch 2 footnote 16: "Hyperliquid taker fee 3.2 bps" → revise to current docs: "Hyperliquid taker fee 0.045% (4.5 bps) at the base tier on perpetuals, scaling to 0.024% at top tier" (with citation update).

I'll make these patches as part of Ch 1 Phase 3 work and note them in Ch 1's REVIEW_NOTES.md and Ch 2's REVIEW_NOTES.md revision log.

Phase 2 is complete. Per the user's pattern of compressed reviews, Phase 3 begins immediately.