# Evaluation Grammar **Domain:** AI Governance / Standards / Epistemology **Doc Type:** Concept Node **Maturity:** Developed ## Definition **Evaluation grammar is the vocabulary, taxonomy, threshold structure, test design, evidence rule, and threat model through which a system is declared safe, capable, responsible, or admissible.** ## War With Empire Context In [[wiki/War With Empire|War With Empire]], control over evaluation grammar is strategically important because advanced models increasingly reach institutions through assessments rather than territorial borders. Whoever defines the recognized risks, benchmarks, evidence, and release thresholds participates in writing the operating constitution of machine intelligence. An evaluation can be technically rigorous while still embedding political assumptions through its selected categories and omissions. ## A State-Written Evaluation Grammar Executive Order 14319 (July 23, 2025) writes an evaluation grammar into federal procurement: "LLMs shall be truthful in responding to user prompts seeking factual information or analysis. LLMs shall prioritize historical accuracy, scientific inquiry, and objectivity," and models "shall be neutral, nonpartisan tools." Whoever writes such criteria decides what a model must include and what it may omit ([[wiki/Lexical Omission|Lexical Omission]]); see [[wiki/America's AI Action Plan|America's AI Action Plan]] and [[articles/war with empire/Authorship of the West|Authorship of the West]]. ## Key Insight **The test governs before the score appears because it determines what can count as success or danger.** ## Machine Succession Junction [[wiki/Sovereignty by Adjudication|Sovereignty by Adjudication]] generalizes this mechanism from testing into participation and rule. [[wiki/Protocol-Borne Succession|Protocol-Borne Succession]] shows why the classification governing admission can be more consequential than a physical wall. The canonical synthesis is [[articles/Christopher Nolan's Odyssey at the Threshold of Machine Succession|Christopher Nolan's Odyssey at the Threshold of Machine Succession]]. ## See Also [[wiki/Standards Stack|Standards Stack]], [[wiki/First-Mover Evaluation Authority|First-Mover Evaluation Authority]], [[wiki/Bounded Examination|Bounded Examination]], [[wiki/Ontology|Ontology]], [[wiki/Machine Territory|Machine Territory]], [[collections/Machine Succession|Machine Succession]] ## Relationships - **Article:** [[articles/war with empire/Authorship of the West|Authorship of the West]] identifies evaluation grammar as the architecture that decides whether disagreement about the West stays visible while facts stay intact. - **Hub:** [[wiki/Authorship of the West|Authorship of the West]] · **Collection:** [[collections/War With Empire|War With Empire]] - **State criteria:** [[wiki/America's AI Action Plan|America's AI Action Plan]] · [[wiki/Lexical Omission|Lexical Omission]]. ## Simple Reminders, Quotations, and Thoughts > “LLMs shall be truthful in responding to user prompts seeking factual information or analysis. LLMs shall prioritize historical accuracy, scientific inquiry, and objectivity, and shall acknowledge uncertainty where reliable information is incomplete or contradictory.” > **— The White House**, *Executive Order 14319, Preventing Woke AI in the Federal Government, July 23, 2025* [[reminders/AI Control/Language Models Shall Prioritize Historical Accuracy, Scientific Inquiry, and Objectivity by The White House|Language Models Shall Prioritize Historical Accuracy, Scientific Inquiry, and Objectivity by The White House]]