PHASE 3.08 — NEXT DESIGN NOTE

Phase 3.08 validates the coherent HARD 17 -> STAND intervention with a substantial independently replicated improvement:
Primary +1.889 average bankroll units versus 3.07.
Holdout +1.996.

The remaining replicated evidence is now heavily populated by HARD 16 states.

IMPORTANT DESIGN BOUNDARY
Unlike HARD 17, HARD 16 cannot safely be converted into one unconditional action family. The Phase 3.08 evidence itself shows dealer-dependent Textbook actions:
- HARD 16 vs 10 / Ace -> HIT in the recorded states.
- HARD 16 vs 4 / 5 split-child -> STAND in the recorded states.

Phase 3.09 should therefore analyse HARD 16 as a structured dealer-dependent family. If an intervention is justified, it should map each qualifying visible HARD 16 state to its corresponding Textbook action rather than imposing one blanket HIT or STAND rule.

All architecture, information controls, 30-hand horizon, multiplier monitoring and peak-bankroll instrumentation remain frozen.
