Methodology
Every rhetoric score on this site comes from the same four-step process, run the same way for every member of Congress.
Collect
Pull a real, sourced statement — a floor speech, interview, post, or article.
Classify
Figure out who it's actually aimed at: the other party, their own party, or no one.
Score
Rate three things: blame toward the other side, substance, and respect for their reasoning.
Average
Combine the three into one 0–10 rhetoric score.
This is not an ideology score
A member isn’t scored lower for being further left, further right, more radical, or more moderate than some imagined center. Arguing hard for a strongly ideological position scores well here, as long as the argument stays professional.
What gets graded is professionalism: engaging the substance of an issue, instead of blaming the opposing party as a bloc, attacking a person’s character instead of their record, or treating your own position as beyond question. No one’s politics are provably 100% correct — a member who argues hard for their side while still treating the other side’s reasoning as legitimate is being professional. That’s the bar, regardless of party or how far left or right someone sits.
The full technical breakdown
Click any box below for the exact rule behind that step.
1 · Collect a statement
Every statement comes from a real, linkable source: Senate/House floor speeches (via Congress.gov's Congressional Record), interviews, official written statements, social posts, and news coverage. Nothing is paraphrased or invented — every score links back to exactly where the statement came from.
Why this exists
The rhetoric score is generated with help from an AI language model following this fixed, published rubric — not a subjective editorial judgment. It can still get things wrong: sarcasm, a lawmaker quoting an opponent to rebut them, or one long speech that’s harsh in one place and conciliatory in another are all genuinely hard cases. That’s why every score links back to its original source, and why there’s a Report a problem option on every page — real flagged mistakes are the best way this rubric gets better over time.