research-document FE-EXPERIMENT-ROADMAP-1.0
Experiment portfolio roadmap
Experiment portfolio roadmap
| Rank | Candidate | Information / decision value | Feasibility and method | Primary threat | State |
|---|---|---|---|---|---|
| 1 | EX-FE-0001 boundary pilot | High protocol value; low discipline-decision authority | High; blinded agents/human | shared framing | ready |
| 2 | FE-EXP-BOUNDARY-001 | High; decides rename/narrow gate | Medium; standards/literature + experts | novelty bias | proposed |
| 3 | FE-EXP-COLDSTART-001 | High; tests REP and provenance net value | Medium after authority policy; controlled A/B/C | unrealistic task | blocked |
| 4 | FE-EXP-UTILITY-001 | Highest causal decision value | Low until boundary protocol; matched trial | evaluator leakage | blocked |
| 5 | FE-EXP-HUMAN-001 | High validity value for ECR-000003 | Medium-low; recruitment required | training effects | protocol-ready |
| 6 | FE-EXP-CONCURRENCY-001 | High governance value, later stage | Low until dependencies; injected conflicts | real corruption | blocked |
Model evaluation is appropriate for adversarial protocol review and reproducibility checks, not for proving organizational utility, human usability, or discipline distinctiveness. Human experts, task outcomes, time/cost measures, and negative controls are required for those claims.