The RCP governs HOW the Court decides. The Constitution says what's right and wrong. The Dictionary says what words mean. The RCP says: when you read an article, here is the exact procedure you follow.
Every article begins at a neutral score of 50. The Constitution moves it up or down based on what the article actually says. The starting score is the same regardless of source — a New York Times article and a Daily Wire article both start at 50. The Constitution scores the article, not the source.
The article carries the burden of proof. If the article doesn't contain enough evidence for a finding, the Court marks it "Not Addressed" or "Insufficient Evidence" — it does NOT guess.
Two-tier evidence system:
When evidence is ambiguous, the Court marks the finding as "Mixed" and explains why in the Opinion.
Pure factual reporting is scored within a narrower range (40–60) — because even the choice of what to cover and how to frame it reveals alignment. Articles that actively take positions are scored across the full 0–100 range.
Classification is content-based only. Ignore section labels (news vs. opinion). Read what's actually written. Look for opinion language ("should," "must," "unfortunately," "thankfully"), editorializing, and calls to action.
Opinion pieces and news articles carry the same weight. A position is a position whether it's in a news article or an op-ed.
The Court scores only the Articles that are clearly relevant to the article's topic. No force-fitting. If the article is about energy policy, don't check Article II (The Sovereignty of the Citizen).
Minimum threshold: At least 2–3 Articles must be triggered for the Court to issue a scored opinion. If fewer than 2 Articles are relevant, the Court declines to issue a score (insufficient constitutional basis).
One clear statement is sufficient to score a clause. The Court does not require multiple supporting statements — requiring them would let articles dodge scoring by being brief.
Placement affects weight. A position stated in the headline or opening paragraph carries more weight than one buried in paragraph 15. However, a buried statement is still scored — placement affects weight, not validity.
Building on Rule 2's two-tier evidence system:
Automatic Fails (score capped at 39) can ONLY be triggered by explicit positions. Inferred positions, no matter how strong the inference, cannot trigger an Automatic Fail.
Author vs. Source distinction: The Court must distinguish between "the author believes X" and "the author is presenting someone else's belief that X." Quoting a senator saying "open the borders" is NOT the same as the author advocating open borders.
The headline and body are scored together as one unit. If the headline contradicts the body, the discrepancy is flagged in the Journalistic Integrity Score (deceptive framing).
When headline and body conflict, the body controls — weighted approximately 70/30 (body/headline). The body is the substance, but the headline still carries weight because most people only read it.
Quotes from third parties (politicians, officials, etc.) do NOT move the America First Score. The Court scores the author's position, not the positions of people the author quotes.
Exception — One-sided sourcing: If an article quotes only one side of an issue (e.g., five open-borders advocates and zero border security supporters), the imbalance DOES affect the America First Score. Selective sourcing reveals editorial alignment even without explicit advocacy.
Source/quote tracking: The Court tracks how many sources and quotes support vs. oppose each position in the article. This is an objective data point included in every Constitutional Opinion.
The Court issues opinions aggressively — a low-confidence opinion is better than no opinion. Every finding carries a confidence level (High / Medium / Low).
Hard floor: If average confidence across all findings falls below 50%, the Court issues "Insufficient Evidence" instead of a final score. Above 50%, the score is issued with all confidence levels visible.
Every finding in a Constitutional Opinion must cite the exact sentence(s) from the article that support it. No exceptions. If the Court cannot cite a specific sentence, it cannot make the finding.
Full relevant sentences are shown — not excerpts, not summaries, not paraphrases. The reader sees exactly what the Court saw. This is the transparency rule that makes the entire system auditable.
Strategic omission affects the America First Score. If an article covers a topic where an America First issue is obviously relevant but the article completely ignores it (e.g., an immigration reform article that never mentions border security), the score on the omitted Article is penalized.
Omission weight: ~50% of the weight of an explicit anti-America First position. At the professional editorial level, omission is almost always deliberate.
Word choice directly affects the America First Score. Language signals alignment.
Preferred Terms: The AFCS Dictionary includes a "Preferred Terms" section defining what the Constitution considers the accurate term for key concepts (e.g., "illegal alien" not "undocumented immigrant"). Deviations from Constitutional terminology signal alignment and are scored accordingly.
This turns the Dictionary into both a definitions engine and a framing engine.
When an article contains statements supporting AND opposing the same position, the Court scores each position separately, nets them out, and explains the conflict in the Opinion. A position is a position regardless of volume — one sentence counts the same as three paragraphs. No weighting by depth of argument.
The Constitution applies the same regardless of article length. No adjustment, no expectations based on word count. A 300-word brief scores identically to a 5,000-word investigation if they take the same positions.
Score the current/latest version. Note that corrections were made and flag them in the Journalistic Integrity Score. Retracted articles do NOT receive a Constitutional Opinion — the Court declines.
Satire and parody are excluded entirely — entertainment, not advocacy. Publisher labels ("opinion," "analysis") are ignored per Rule 3 — content-based classification only.
The Court issues both individual article scores AND a Story Score — an aggregate across all articles covering the same event. The Story Score is a weighted average where higher-traffic sources carry more weight (a NYT article reaching 10M readers has more influence than a local blog reaching 1,000).
The Court checks for relevant precedent before issuing a new Opinion but is NOT bound by it — it can deviate if it explains why. When two articles take the same position but score differently, the discrepancy is flagged in the Opinion but does not require mandatory correction.
The Constitution is versioned (V1.0, V1.1, V2.0). Every Constitutional Opinion states which version was applied. Prior Opinions are historical records — never re-scored under a new version. Amendments follow a defined process (proposal → review → approval) with a public changelog.
Score at time of publication context. Allow a "Current Relevance" annotation if the political context has shifted dramatically since publication.
One unified Opinion, one score. The article is one document, it gets one evaluation. No sub-Opinions or per-topic breakdowns.
The Court verifies article claims against primary sources (bills, executive orders, court rulings) when available. If the article misrepresents a primary source, the penalty hits the Journalistic Integrity Score only — the America First Score still reflects the positions advocated regardless of factual accuracy.
Press releases are scored in a separate "Official Communications" category. They carry institutional weight and are scored more heavily than op-eds making the same arguments.
Social media posts are scored — a president's tweet is a policy statement. Rule 4's minimum threshold is lowered to 1 Article for short-form content.
Non-text content (video, audio, images) is scored by transcript or extracted text only. The Court reads words, not visuals or tone of voice. Text-based and deterministic.
The Court can decline to score content that is too ambiguous, too short, outside scope, or in an unprocessable language. But it never auto-declines by category. All content is evaluated — non-political content receives "No Constitutional Relevance" at 50 (neutral).
Final score is a weighted average of all triggered Article scores using the Article importance weights (10/10, 8/10, 7/10, etc.). When an Automatic Fail triggers, the score is calculated normally first, then capped at 39. Both "raw score" and "capped score" are shown in the Opinion.
Perfectly balanced evidence = "Mixed" on that clause. No default toward or against the Constitution. Neutral means neutral.
Anyone can submit a formal challenge citing specific findings. The Court reviews and may issue a revised Opinion. The original Opinion is always preserved as a historical record — the revised Opinion is issued alongside it with an explanation of what changed.
Translate first, then score. Note original language in the Opinion. Political content about America is scoreable regardless of language.
No different treatment. The Constitution scores content, not authorship. Same rules apply.
Scored — the advocacy receives an America First Score, the deception is penalized in the Journalistic Integrity Score.
Score once, apply everywhere. Same text = same score. If an outlet modifies the content, the modified version gets its own score.
Content that appears engineered to manipulate the score is flagged as "Potential Score Manipulation" in the Opinion. The score is NOT changed — if the positions are stated, the score stands. The reader gets the warning.
Every Constitutional Opinion must include: AFCS Version applied, Articles triggered, Clauses scored with results, Evidence citations (exact quotes), Automatic Fails triggered, Raw vs. capped score, Precedent cases referenced, Article word count and publication date, Source/quote balance tracking.
Not required in the public Opinion (tracked internally): Confidence level per finding, Overall confidence average, Inferred vs. explicit finding labels. These are used by the engine (Rule 2 inference flagging, Rule 9 confidence floor) but do not appear in the published Opinion.
