Competing Slop Filter hypotheses are ready to test
Classification and exceptions teams can propose bounded interventions while an evaluator compares them against the same baseline.
Depends on: P1-M1, O1-M1, E2-M1
Epic P2
Competing hypotheses and experiments
Several teams can propose different approaches to the same outcome, and an evaluator can compare them before build work starts.
Purpose
The organisation can choose between classifier changes, exception controls, positive selection or no build at all.
Current status
Proposed. The assignment defines the experiment but says it must not be run automatically as part of this website build.
Milestones
Evidence
Cultivate operator brief
Full 20-section assignment preserved locally with SHA256 e8b55ed458e3bc6347735cccee872f59a0bf863b4d1dddb56ba2400b21b2461f.
Existing narrative and capability roadmap Google Doc
Read through an authorised Google Drive connector and restructured into site pages rather than copied wholesale. The source document link is not published.
Risks and open questions
Risks
- Competing teams can converge on the same weak evidence.
- The evaluator can become another untested agent opinion.
Open questions
- What budget is acceptable for the first comparison?
- Which stopping conditions prevent chasing marginal gains?
Linked Wardley components
- User outcomes
- Product experiments
- Independent evaluation
- Agent orchestration
Next steps
- Prepare the experiment record without commissioning the experiment yet.