nx_learnmine_lib.nx

nx_learnmine_lib.nx -- LEARNING MINING: third-party how-to material as SUPPORTING GUIDES for the /compare rungs (operator 2026-09-06: farm archive.org, YouTube, Vimeo, DeviantArt and the rest for videos, transcripts and text that power the development rungs, so the boards carry third-party feedback and reference frames, not only the seat's own judgement). The evidence law is the operator's: a workflow video is a LIAR-KILLER (it proves the field can do the thing and shows how) or a REFERENCE FRAME (what the result looks like in that month); it is NEVER an oracle, and its month is its bar. This lib is the FIRST rung of the lane: the archive.org advancedsearch listing (output=json) of one creator or query, parsed from its own bytes into item rows and a census (media types, years, subjects) so a seat reads a 200-row digest instead of a quarter-megabyte of JSON. Every later rung (item files, subtitle mirrors through nx_srt and nx_webvtt, the YouTube caption track, the guide rows on <dom>.guides) composes this table; nothing here fetches -- the bytes come from nx_research_fetch under provenance, and the organ only reads what was mirrored. license_tier: ORIGINAL No hw writes (Rule 26).