Reports

Dated pieces of work, computed from the record on the day they were published and left alone afterwards. Each states the day it was written, the coverage it had at the time, and the query behind every figure. For what is true right now, which changes as the record grows, see what the record shows.

Published	Report	What it is about
2026-09-15	Five of every six token creations are not launches	One million token creations read block by block: 1,020,962 mints over 99.945% of the slots in the window, of which 170,694 are launches - and the largest single creator of tokens produced 756,294 mints and not one launch.
2026-09-15	Evidentiary report on a single launch where the launch-record tests did not fire	The specimen of the commissioned provenance report: one launch's evidence in full, read 2026-09-12 — chain anchors, custody, a complete curve ledger reconciled to the digit, wallet structure in aggregate only, and what the record does not say.
2026-09-12	A launch's own account of itself is not uniformly durable	Documents on 211,977 IPFS-addressed launches lost nothing in ten days; one host carrying 42,902 launches stops serving them after about two days.
2026-09-11	A graduation event is not a graduation	Of 12,524 graduation events on the feed, 5,489 have no pool behind them, and of those we read on chain 96.4% had no completed curve either.
2026-09-11	The same ticker, a new creator every time · revised 2026-09-11	20 ticker symbols, each used by fifteen or more separate mints with a fresh creator wallet almost every time.

Every figure in a report is a query against the record, printed beside it so anyone can run it and get the same answer for that date, or a different one, and say so. A report is never rewritten by a later build: it is written once, and a correction has to say what it changed.