Done. The profile recomputed across all 89 examples — your five new positions reinforced existing axes rather than moving them (the shifts are sub-0.1 at this granularity, since they're consistent with your established framework). What they strengthened, directionally:
- Anti-Cheat (3.7) — reinforced three times: label-inflation as a delegitimization weapon, narrative-control-by-omission, technique-vs-program. This is now one of your most densely-evidenced dimensions.
- Process (3.4) — reinforced: liability tracks proximity + control + gross-negligence, not loose correlation.
- Speech Tolerance (3.6) / Pluralism (3.5) — reinforced: ugly rhetoric stays protected; refusing to lump a whole coalition as "nazis."
- Institutional Respect (2.5) / Certainty-Seeking (2.5) — nudged down: you traced the common comparison to primary sources instead of accepting it, and sat comfortably in the calibrated middle.
What the research settled, in one paragraph
Your prior held up. Hitler's worst was eliminationist program on the record years before power (the Mein Kampf poison-gas passage, 1925-26; the 1939 "annihilation of the Jewish race" prophecy that he then executed). Trump's worst is a handful of dehumanizing quotes — "vermin" (aimed at domestic political enemies, not a race), "poisoning the blood" (aimed at immigration, origin-coded but not provably a single-race white-supremacist claim, and he denied the Mein Kampf lineage), "animals" (disputed MS-13 referent). The techniques rhyme — how you rile people up to vote — but the content and program don't, and ~90% of Trump's typical rhetoric maps better onto the American populist line (Jackson → Long → Wallace → Buchanan) than onto Nazism. Your friend's "pre-power Hitler was mild like Trump" doesn't survive primary sources: the ideology was explicit and written before 1933.
Artifacts on disk
- research/hitler_trump_rhetoric/SYNTHESIS.md — the full four-quadrant comparison, three-axis analysis, delegitimization-tool section, and uncertainty zones, all quote-anchored.
- pos1–pos6.md — the six steelman reports.
One honest data gap worth flagging: there is no peer-reviewed corpus/grade-level study of Hitler's German speeches, so all "90% of his speaking" percentages rest on Klemperer's philology and Kershaw's synthesis, not quantitative analysis — they're estimates, more confident-sounding than the underlying evidence strictly warrants. The Trump side (CMU grade-5.7, PNAS Nexus 2024) is genuinely quantified.