Skip Penalty Ranking Assumption
An earlier assumption that skipping a video should directly reduce its credibility score.
Why it was archived
Skipping has too many plausible causes to function as a reliable direct negative credibility signal.
What it was
A ranking penalty connected to early exit or skipping.
An attempt to treat low exposure as negative evidence.
What changed
The direct skip penalty was removed.
Watch behavior became contextual evidence rather than automatic punishment.
The model retained uncertainty about why a user exited.
What it taught
Absence of engagement is not automatically evidence of poor quality.
Prior knowledge, irrelevance, time constraints, and interface behavior can all create the same trace.
Negative evidence requires as much justification as positive evidence.