Notes from the field.

State.OS reads how an asset is actually quoted across the market right now: what reference it's measured against, how many pools carry it, how far those pools disagree with each other, and how much liquidity backs each quote. It does not tell you what to trade. There is no buy or sell signal anywhere in the product, no ranking of opportunities, and no synthesized "best" anything — what it shows is structure, the shape of the market at a given block, and it leaves the interpretation to you.

The four surfaces

The product is split by the kind of question each surface answers, not by asset class or feature area. Markets answers a breadth question: across every tracked asset, which ones currently show the widest dispersion, the thinnest liquidity, the oldest reference? It's built to be scanned, filtered and sorted.

Topology answers a depth question: for one asset, what does its actual pool structure look like right now? It trades breadth for a fuller picture — a graph instead of a row, an inspector instead of a cell.

Replay answers a time question. Markets and Topology are both single-block snapshots; Replay is the only surface that shows how a reading arrived at its current state, stepping through a fixed, recorded sequence of prior blocks.

Field Notes — this section — isn't a fifth data surface. It's the vocabulary the other three assume you already know: what "dispersion" means precisely, what a state label commits to, what a metric is actually computed from. Every term used elsewhere in the product links back to an entry here.

What counts as a "read"

It's worth being precise about where the numbers actually come from, because the product deliberately blends two different kinds of data and the distinction matters. Reference prices for real tickers are live — proxied from a public market-data feed and polled once a second — so a headline price you see is an actual current quote, not a fixture.

The pool-level structure around that reference — how many venues carry the asset, their individual quotes, liquidity, and how those relate to each other — is modeled: generated deterministically from the asset's symbol so the same asset always produces the same shape of pool topology, rather than pulled from real on-chain pools. Every mock data file in the codebase says so directly in a comment. The product's honesty about this boundary is itself part of the read-only stance — nothing here claims to be a live venue integration it isn't.

Why this matters

Most trading-adjacent tools collapse observation and action into the same surface: a price next to a buy button, a spread next to an execute call. That framing quietly biases what gets shown — you tend to surface the numbers that justify the action next to them. State.OS removes the action. What's left is just the reading, and the reading is judged only on whether it's an accurate description of the market's current shape.