Before kickoff, did the news stream this project already captures (news titles + fetched bodies) name the men served an all-zero conditional line, and does a mention separate the zero-line men who PLAYED from those who did not?

Owner lane: q-language

Question

Before kickoff, did the news stream this project already captures (news titles + fetched bodies) name the men served an all-zero conditional line, and does a mention separate the zero-line men who PLAYED from those who did not?

Prediction

Prereg research/scientist/prereg/2026-09-23T2038Z-h114-zero-line-news.md: E1 ZP body-mention <=35%, title <=15%, NP body >=60%. E2 mentioned zero rows play more often (direction) but title Fisher p > 0.05. E3 Wentz mentioned, mentioned men carry >=40% of ZP points. E4 fewer than half of mentioned ZP headlines state a role. Falsifier: ZP body rate <= ZN body rate -> MEASURED NOTHING for Q.

Finding and verdict

PARTIAL / EVIDENCE FOUND, CAPTURE_ONLY_FACTOR_NOT_ADMITTED. Captured pre-kickoff language named 42-44 of the 46 graded zero-line men (98% of their 64.5 pts), and an elevation cue near the name raises P(play) among zero-line rows from 0.09 to 0.45 (Fisher p 1.6e-9, post hoc). The largest single loss (Wentz 19.2) was language-consistent: the text said backup. Two weeks, one season, no held-out: this is a discovery, not a forecast gain; no line was recomputed.

Reasoning

The zero line is not a language question at the share stage, but whether a man is on the game-day roster is -- and elevations are published in text hours before kickoff.

Next action

Separate Q from its structured rival: for the 58 elevation-cue rows, compare the language receipt clock against the first time a structured source (nflverse weekly roster status / transactions, if captured) shows the elevation. If structured is earlier, file the fix to S and close the Q angle; if language is earlier or alone, build an elevation parser over news_body with source-span identity and test a p_play/line floor for elevated men on 2026 weeks 3+ held out.