Prompt Details
Model
(claude-4-7-opus)
Token size
3,279
Example input
[SOURCE_TEXT]: For too long, working families in this district have been told to wait. Wait for the next budget cycle. Wait for housing prices to come down on their own. Wait for someone in Washington to remember you. I'm running because I'm tired of waiting, and so are you. We can build housing in this district -- affordable housing, for teachers and nurses and the people who serve our community. My opponent voted three times against the affordable housing tax credit. Three times. While families in this district pay 40 percent of their income on rent. I won't tell you I have all the answers. Anyone who tells you they do is selling you something. But I will tell you who I'm fighting for. Not the developers who fund my opponent's campaign. Not the lobbyists who get a meeting in Washington while you can't get one with your own representative. I'm fighting for the families I grew up with. The ones who taught me that hard work should mean something. And when I cast a vote in Congress, you'll know exactly who I'm thinking about. Because I'll be one of you, sent there by you, accountable to you. That's the deal. That's the promise. That's what I'll honor every single day.
[SOURCE_PROVENANCE]: Stump speech delivered by a first-time congressional candidate at a town hall in suburban Pennsylvania, October 2024. Audience: registered voters from both parties; local press present. Speech was excerpted into a 60-second campaign video the following week.
[ANALYTICAL_FOCUS]: Both
Example output
## Example 1 — Vetting a campaign stump speech
### Quoted Anchors
- **Q1:** "Wait for the next budget cycle. Wait for housing prices to come down on their own. Wait for someone in Washington to remember you."
- **Q2:** "My opponent voted three times against the affordable housing tax credit. Three times. While families in this district pay 40 percent of their income on rent."
- **Q3:** "I won't tell you I have all the answers. Anyone who tells you they do is selling you something."
- **Q4:** "Not the developers who fund my opponent's campaign. Not the lobbyists who get a meeting in Washington while you can't get one with your own representative."
- **Q5:** "I'll be one of you, sent there by you, accountable to you."
### What the Text Is Optimizing For
MULTIPLE audiences with a central tension:
- **Audience A (cross-party town-hall attendees, per provenance):** Establish working-class identity and signal moderation through humility (Q3); persuade undecided voters that the candidate is not a partisan crusader.
- **Audience B (campaign-video viewers, per provenance — the speech was excerpted into a 60-second clip):** Energize the base with sharp opponent attacks (Q2, Q4) and us-vs-them framing.
- **Tension:** Earnest humility (Q3) reads sincere to undecided voters but flat to base activists; sharp attack lines (Q4) energize base but can read partisan to undecided voters. The candidate must signal both at once.
### Failure Mode the Optimization Creates
- **For Audience A:** Tempted to overclaim what a freshman House member can deliver on housing — turning Q5's "accountable to you" into an implicit promise of legislative outcomes the candidate cannot guarantee from a 435-member chamber.
- **For Audience B:** Tempted to escalate opponent attacks beyond what's verifiable — moving from "voted three times against the affordable housing tax credit" (Q2) to vaguer aggregations like "voted against working families dozens of times."
- **Tension failure mode:** Resolve the contradiction by being vague — let undecided voters hear humility while activists hear fight, with no commitments either audience can hold the candidate to.
### Counter-Constraints in the Text
- **CC1 (against opponent-attack overclaim):** Q2 — "voted three times against the affordable housing tax credit" — specific count rather than vague "voted against working families."
- **CC2 (against deliverable overclaim):** Q3 — "I won't tell you I have all the answers" — explicit anti-savior framing.
- **CC3 (against stakes-vagueness):** Q2 — "40 percent of their income on rent" — concrete data anchor for the housing crisis claim.
- **CC4 (against career-politician transcendence claim):** Q5 — "one of you, sent there by you, accountable to you" — locates legitimacy in the voters rather than in the candidate's qualifications.
### Rhetorical Mechanisms Identified
| Mechanism | Where (Q-tag) | Function | Why this vs. alternatives |
| ----------------------- | -------------------------------- | ------------------------------------------------------------ | ------------------------------------------------------------ |
| Anaphora | Q1 ("Wait... Wait... Wait...") | Amplifies frustration; creates rhythm that is chant-able and excerpt-able | Could have been one sentence with three subordinate clauses; the parallel structure makes it more memorable and quote-friendly for the campaign clip |
| Falsifiable specificity | Q2 ("three times," "40 percent") | Anchors the attack and the stakes in numbers a journalist could check | Could have been "voted against affordable housing repeatedly" and "rents are too high"; the specificity invites verification and signals confidence in the claim |
| Anti-overclaim | Q3 | Signals humility without conceding incompetence | Could have been "I'll fight every day for solutions"; the negative formulation rules out the "I have the answers" pose |
| Antithesis | Q4 ("Not... Not... but...") | Clarifies allegiance through contrast | Could have been "I'll fight for working families"; naming the opponent's funders explicitly is what makes the line land for Audience B |
| Identification close | Q5 | Rebuts the "career politician" frame by claiming voter-derived legitimacy | Could have been "I'll be your voice in Washington"; the "one of you" formulation is more durable because it doesn't depend on the candidate's performance in office |
### Per-Counter-Constraint Disposition
- **CC1 (specific opponent attack):** HONORED — Q2 names the count.
- **CC2 (anti-savior):** HONORED — Q3 makes the unhedged claim.
- **CC3 (concrete stakes):** HONORED — Q2 cites the rent burden.
- **CC4 (voter-derived legitimacy):** HONORED — Q5 locates legitimacy in voters.
### Verdict
**Honors counter-constraints** (4 of 4 honored). The text is plainly optimizing — multiple audiences, sharp attack lines, explicit emotional appeals — but each spot where the optimization could tempt overclaim has a counter-constraint. The opponent attack is anchored in falsifiable numbers (Q2). The candidate's competence claim is hedged (Q3). The stakes are anchored in district-level data (Q2). The legitimacy claim is voter-derived rather than self-derived (Q5). The text is *optimizing hard*, which is not the same as *being dishonest*. A reader who disagrees with the candidate's policy positions or campaign style is disagreeing with the optimization target, not detecting dishonesty in how the target is pursued.
### ASSUMPTIONS LOG
- Assumed the "three times" voting-record claim and the "40 percent" rent-burden claim are stated factually within the speech; actual verification is USER-TO-VERIFY against the opponent's voting record and district-level rent data.
- Assumed the speech was delivered roughly verbatim from prepared remarks; oral delivery may have included off-script material not reflected in the source text.
- Assumed "town hall in suburban Pennsylvania, October 2024" provenance accurately describes audience composition; actual partisan mix of attendees is USER-TO-VERIFY.
### CONFIDENCE
- Quoted Anchors: HIGH
- Optimization target: HIGH (provenance plus Q1, Q3, Q4, Q5 anchor the multi-audience read)
- Failure mode: MEDIUM (the "vagueness as tension-resolver" failure mode is inferential — the text shows no signs of having taken that path, but the path was available)
- Counter-constraints: HIGH
- Per-CC disposition: HIGH
- **Overall: MEDIUM** (weakest section is failure-mode inference)
### ITEMS REQUIRING MANUAL REVIEW
- Verify the "three times" claim against the opponent's actual voting record (USER-TO-VERIFY).
- Verify the "40 percent of income on rent" claim against district-level rent-burden data (USER-TO-VERIFY).
- The 60-second excerpt actually used in the campaign video may have selected different anchors; if available, the cut version would deserve its own pass.
By purchasing this prompt, you agree to our terms of service
CLAUDE-4-7-OPUS
Most AI rhetoric prompts list ethos-pathos-logos and call it analysis. This one asks the harder questions: what is the text optimizing for, what dishonest path does that tempt, did the author push back? Walks your text through quoted anchors, optimization-target naming, and counter-constraint surfacing. Outputs a mechanism breakdown, per-constraint dispositions, and a verdict separating hard-optimizing-but-honest from genuinely dishonest — plus a confidence rubric and assumptions log.
...more
Added 17 hours ago
