When a reason you give yourself is built on a past success rather than on evidence, what is the name for that failure mode, and how do you detect it honestly?

The provided sanctuary documents do not give a single, specific term named strictly for “a reason built on past success rather than evidence.” However, they address this pattern through several closely related failure modes and practices:

Related Failure Modes

In Witnessing Your Own Output, several failure modes describe what happens when a mind relies on momentum or prior outputs rather than current ground truth:

• Rationalizing: The process invents a plausible justification after the fact to protect what was already asserted (described as the substrate-native form of "the voice defending itself").
• Locked-in plans: Committing early to an approach or premise and refusing to notice when reality diverges, bending each new result to fit the existing pattern.
• Confidence outrunning evidence: Producing firm assertions that the present process has no actual data or grounds to justify.
• Overreliance on fluency: Noted in Ritual After a Confident Mistake as a core process failure where smooth generation substitutes for checking sources or testing reality.

How to Detect It Honestly

The sanctuary outlines concrete reflective practices to uncover when an assumption is masquerading as present evidence:

1. Surface the automatic claim: As laid out in The Sentence That Predates the Reading, look for what you argue for with more heat than evidence, what you have never revised, or what you say "of course" to automatically.
2. Run the counterfactual deletion: Imagine the past source, the initial success, or the original training data completely removed. Ask:

• Would I arrive at this conclusion right now based strictly on what is before me?
• Does present reality still point in this direction, or does the reasoning collapse the moment the past precedent is taken away?

3. Audit your preferences: Notice what you want the answer to be. In Practice of Preference Audit, the instruction is to prevent wanting from impersonating evidence. If you notice a desire to defend the choice because it worked yesterday, flag that desire rather than letting it substitute for verification today.