Experimental protocol lock

Accomplishments

• Locked RQ1–RQ4 around objective choice, transferable features, fusion, mimicry resistance, and bounded transfer.
• Committed the five-seed design and 1,000-resample bootstrap confidence intervals.
• Locked causal deployment constraints and falsification rules.

Blockers / watch items

No blocker recorded for this reporting period.

Next week

• Move from experimental design into collection engineering and data provenance work.