Scientific record
No methodology is adopted. These are design questions for exploratory work, not a selected task suite or model architecture.
Questions before protocol design
- Specify which aspects of deliberate generation, manipulation, exploration, and evaluation the project will investigate and why they belong to the target capability.
- Identify observable tasks and comparison conditions that separate the target from prediction, simulation, planning, creativity, and Project Fiction.
- Decide whether and how to measure properties such as consistency under changed assumptions. Justify any criterion rather than treating it as part of the definition by default.
- Determine suitable inputs, data requirements, existing-system comparisons, evaluation procedures, and limits on inference from task performance.
- Develop testable hypotheses after the capability boundary is sufficiently clear.
No dataset, benchmark, metric, experimental protocol, or model design is selected here. Record proposals with their actual state in the corresponding project folders. When experiments are warranted, register them and preserve the full reproducibility record described in the experiment registry.