PUBLICATIONS · REPORTS / ESSAYS / ANALYSIS / NOTES

Everything we publish is open, dated, and written to still be worth reading in five years.

Reports are long-form surveys of a quadrant. Essays argue a single idea. Analysis follows the data. Notes are fieldwork in progress.

2026-06-02

Intent is the new order book

Solver networks now route a majority of large DEX flow. We measure concentration across 90 days of fills and ask whether intent systems re-created the intermediary they were meant to remove.

Analysis
2026-05-18

Attestation as a market primitive

Remote attestation lets strangers trust hardware they don't own. The financial applications begin with custody and end with markets whose matching engines can prove their own fairness.

Essay
2026-04-30

Sync engines and the end of the request

A technical reading of five production sync engines — replication semantics, conflict behavior, offline windows — and why the request/response web is quietly becoming a replication web.

Report
2026-04-12

The FHE cost curve, revisited

Fully homomorphic inference was 10,000× slower than plaintext in 2021. We re-run our benchmark suite on 2026 hardware and schemes, and chart what the curve implies for private finance.

Analysis
2026-03-27

Stablecoins are the eurodollar of software

Offshore dollars rebuilt global banking outside its regulators once before. The parallel is closer than it looks, and instructive about where settlement migrates next.

Essay
2026-03-09

Field note: running a light client where the internet is bad

Verification on commodity hardware over unreliable links is the real test of decentralization. Notes from three weeks of measurements outside the data-center map.

Note
2026-02-20

Who is liable when the agent trades?

Autonomous agents holding balances collide with a legal system built for persons. A survey of the doctrine, the precedents, and the contractual workarounds emerging in practice.

Analysis
2026-01-15

Annual survey: the state of the frontier, 2026

Our yearly reading of all six quadrants — what compounded, what stalled, and the three theses we retired in public.

Report
DON'T REFRESH THIS PAGE

New publications reach the Letter first.

Subscribers read reports a week before they're posted here.