Rivers

Published by DeltaSignal (AITrailblazer).

A river is one framework article and everything that later bears on it. The framework states a thesis and the conditions that would break it. After that, the record is kept in the open: articles are reviewed into membership by hand, claims are extracted from the founding article and approved before they are published here, later events are graded as validations against primary sources, and each river is kept under a written protocol with a named owner and a retirement rule.

Every item in our archive carries a TripCode, a permanent address minted from the item itself, so a claim, an article, and a validation can always be traced back to their source. The rivers are the frameworks those addresses flow through.

One honesty note. A machine linker also flags archive items that may belong to a river, by keyword. Those flags are candidates, not membership, and they are counted separately on every page: measured on October 23, 2026 against blind review, 29.7 percent of the linker's article edges were confirmed substantive membership (41.5 percent counting borderline calls). Membership on these pages means exactly two things: the founding article, and articles a reviewer has read in.

The graded calls

The graph also carries the 32 proven calls from the track record, each recorded against the article that made it. Where the article is a reviewed member of a river, the call is graded inside that river and listed on its page. The rest stand in the graph on the track record alone; no call is placed in a river it does not belong to.

The rivers at a glance

Each river gets the same small picture, drawn on the same scales, so the eye can compare them directly: time runs from October 4, 2025 to October 30, 2026 in every panel, and the line rises one step for each reviewed member, to 25 at most. Under the line, one divided bar: dark is reviewed members, open is candidates the linker flagged and no reviewer has read. Exact counts stand under each panel and in the table below.