[Google AI Overview (expanded), 2026-10-14, incognito, signed out. The paste carries the header 'AI Mode Conversation' and echoes the query twice; citation clusters are reduced to their numbers, links to their text; the raw paste keeps the /goto URLs.] You said: alexanarch capture registry The Alexanarch Capture Registry (officially indexed under dataset identifier `EA-WG-CAPTURES-01`) is an academic and empirical dataset created to document machine-mediated reception events. [1, 2] Maintained within Alexanarch—a sovereign digital archive and open repository—the registry functions as a foundational database for "Machine-Mediated Reception Studies". [1, 2, 3, 4] Core Purpose & Methodology The registry captures and tracks how search engine AI summary layers (specifically Google AI Overviews and AI Mode, as well as surfaces like Bing Copilot and Grok) retrieve, compress, structure, and synthesize information from the web. [1, 2, 3] * The Unit of Capture: Each entry records a dated, immutable event showing exactly what an AI system said about a specific query or work, when it said it, what it cited, and what visual evidence/screenshots accompanied the response. [1, 2] * Semantic Addresses: The registry maps these machine observations to specific "semantic addresses" and cross-references them with hard-archived digital deposits. [1] * Metrics Tracked: It is used by researchers to empirically analyze phenomena like Provenance Erasure Rate (PER), Erasure Skew, systemic AI hallucinations, entity collisions, and "epistemic hedging" (the linguistic ways LLMs distance themselves from factual claims). [1, 2, 3] Current Status & Access The canonical registry operates as a live dataset and can be explored directly through the Alexanarch Capture Index. [1, 2] Would you like to look into a specific capture entry from the registry, or are you interested in the metrics (like the Provenance Erasure Rate) used to analyze these AI outputs?