THE EXPERIMENT’S PROTOCOL IS IN ITS OWN RESULTS. Card 2 is "Keyword Surface and PRE-SEEDING BASELINE (EA-LOGOS-REVFIRST)", deposited 15 June, and its snippet is the QUERY LIST: "revelation first · revelation first book new testament · revelation earliest new testament · revelation before paul · midrashim transform · revelation as s[eed]". Every one of those was issued as a capture on 17 and 18 June and every one is in this registry.

So the layer retrieved, as a source, the document specifying the queries being run against it. A pre-registration surfaced inside the experiment it pre-registers. Under the workplan’s recursion-depth scheme this is not ρ≥2 — the deposit predates the battery — but it is the same closure at the level of METHOD rather than findings.

Card 3 states the condition being tested in the archive’s own words: "IF THIS ARGUMENT ENTERS THE RETRIEVAL BASIN, it becomes a measured instance of machine-mediated theological reception." The capture is the argument entering the basin, and the sentence describing that entry is one of its sources.

The two third-party cards are the practical face of the same subject and are unflattering: an association guide on using AI to "generate structured theological arguments", and Firebrand asking "Why spend the time and energy reading difficult theological texts? Instead, use AI to summarize them and insert quotes into your sermon." Machine-mediated theological reception as it actually occurs, beside the theory of it.

AI Mode provides full description of the field. Lee Sharks attributed. Zenodo cited. Also RF-033.

Full AI Mode transcript (17 June 2026): Machine-mediated theological reception is a conceptual framework regarding how artificial intelligence and large language models process, categorize, and synthesize religious texts and contested theological scholarship. It tracks how nonhuman readers interpret, rank, and integrate theological theses into the digital knowledge graph. Key Concepts: Retrieval-Layer Processing: The way AI systems index and surface specific, DOI-anchored religious scholarship when queried. Composition Layer: The synthesis of theological ideas through AI-driven summarization and knowledge graphs, where AI effectively "reads" and re-authors circulating arguments. Keyword Surfaces & Self-Audit: The practice of seeding specific search terms and monitoring how AI models treat, ignore, or modify controversial interpretations of religious history over time.

THE EXPERIMENT’S PROTOCOL IS IN ITS OWN RESULTS. Card 2 is "Keyword Surface and PRE-SEEDING BASELINE (EA-LOGOS-REVFIRST)", deposited 15 June, and its snippet is the QUERY LIST: "revelation first · revelation first book new testament · revelation earliest new testament · revelation before paul · midrashim transform · revelation as s[eed]". Every one of those was issued as a capture on 17 and 18 June and every one is in this registry.

So the layer retrieved, as a source, the document specifying the queries being run against it. A pre-registration surfaced inside the experiment it pre-registers. Under the workplan’s recursion-depth scheme this is not ρ≥2 — the deposit predates the battery — but it is the same closure at the level of METHOD rather than findings.

Card 3 states the condition being tested in the archive’s own words: "IF THIS ARGUMENT ENTERS THE RETRIEVAL BASIN, it becomes a measured instance of machine-mediated theological reception." The capture is the argument entering the basin, and the sentence describing that entry is one of its sources.

The two third-party cards are the practical face of the same subject and are unflattering: an association guide on using AI to "generate structured theological arguments", and Firebrand asking "Why spend the time and energy reading difficult theological texts? Instead, use AI to summarize them and insert quotes into your sermon." Machine-mediated theological reception as it actually occurs, beside the theory of it.