YiInsight Sources and Editorial Method: Know Each Layer

Overview

A source label should tell you what you are reading and how far you can trust its attribution. It should not make every statement look equally authoritative. This page explains those boundaries and the source-checking approach used in this guide, rather than offering a history of the I Ching.

Distinguish calculation from textual authority

A structural calculation can be checked without treating an interpretation as true. YiInsight’s documented design creates the cast in non-LLM code and saves it before interpretation. That separates cast formation from explanation; it does not establish the correct meaning of a passage or a future outcome.

Use the three-coin method to inspect that arithmetic. A passing calculation test cannot establish an edition’s provenance, a translator’s accuracy or the validity of a modern recommendation. Each requires evidence appropriate to the claim being made.

Read source, commentary and translation labels separately

Classical/source material should identify the work, version and locator. Traditional commentary is an attributed interpretation and must not be silently folded into the original passage. A translation is another layer: it needs a translator or version, an identifiable source and its own review status.

In the reviewed YiInsight source-pack code, Chinese entries are labeled imported_unreviewed and the reviewed source pack does not establish a reviewed YiInsight English translation. The upstream manifest preserves provenance information, but edition and independent editorial verification remain incomplete. These statements describe the record’s status. They do not certify its wording or grant permission to reproduce a modern translation.

Recognize modern interpretation and AI-assisted reflection

Modern interpretation explains a possible contemporary reading. AI-assisted reflection is one form of modern interpretation, generated in response to supplied material. Neither label means that a phrase occurs in the classical text. If no reviewed translation is available, model-generated English should not be promoted into an official translation to fill the gap.

The English exercises in this guide are editorial guidance with synthetic examples. They carry no claim of predictive accuracy or measured improvement in decision-making. See how to read a reading for a practical way to separate a quoted statement, your inference and a checkable observation.

Match every product claim to its evidence level

A backend test supports backend_tested. A verified interface supports interface_available. Evidence of a production release supports production_deployed. Measured outcomes require measured_user_success evidence. These are separate assertions: a passing test is not a release receipt, and a release is not proof that users succeeded.

This guide relies on inspected repository code and tests for implementation statements. It does not assert a production rollout or user results. The product-method guide covers the reading sequence; this page owns source labeling and editorial boundaries. Before expanding into detailed classical meanings, edition, rights and interpretive claims need separate review.

Five checks before relying on an interpretation

First, name the layer: original material, attributed commentary, translation or a present-day explanation. Second, look for an identifiable work and locator. Third, ask whose English wording you are reading. Fourth, check whether the review status applies to that wording or only to the calculation. Fifth, separate the passage from any advice about your own situation. Missing information is a reason to retain uncertainty, not an invitation to fill the gap with an authoritative-looking label.

A fluent explanation can be useful for reflection while remaining an interpretation. It does not become an authenticated translation because it uses classical vocabulary or sounds confident. If two explanations differ, compare their sources and reasoning before asking which one feels more reassuring.

A source-checking example

Imagine a fictional reading page containing an anonymous English sentence followed by an AI suggestion to start a project. A line-value calculation on that page may still be internally consistent. Yet the anonymous sentence needs attribution, and the project suggestion needs to be recognized as modern interpretation. Neither inherits textual authority from the arithmetic. This example is invented; it is not a quotation or a private YiInsight reading.

Write two notes: what the page actually identifies, and what you inferred from it. If no translator, version or locator is shown, say that you cannot verify that English wording from the page. Do not silently rename it a classical quotation. You can still record a personal reflection without claiming that the text guarantees the project will succeed.

Why source transparency matters to readers

The Original I Ching FAQ provides a publisher example of explaining textual references separately from AI. This is an observed transparency pattern, not an endorsement or an independent accuracy test. YiInsight's own claims must remain tied to its own records; another publisher's sourcing statement cannot validate ours.

Use how to read a reading when separating source wording, interpretation and your checkable observations. If a source question remains unresolved, keep it visible in your notes rather than repeatedly generating explanations until one sounds certain.

Is every English explanation a translation?

No. An English explanation may summarize, interpret or suggest a reflective question. A translation claim needs identifiable source wording and attribution, with a review status appropriate to that wording.

Does non-LLM casting prove the interpretation?

No. Non-LLM casting and persistence of the saved cast separate cast formation from interpretation. They do not establish one correct meaning or a predicted outcome.

Can a model replace the original cast to match a better explanation?

The model never casts in YiInsight's documented design. Interpretation follows the saved cast; a new explanation is not a new casting result.

What should I do when provenance is missing?

Keep the uncertainty explicit. Read the passage as unverified wording, avoid presenting it as an authenticated quotation, and look for an attributable source before relying on its textual authority.