Research · Four-arm workflow · Part 4 · Slide 12 · Tier timeline
A later on/off test wave. The C1/C2/C3 test grid was a different, earlier experiment.
Full sentences · matched to the numbered chips
21 Aug — SYNDROME_LEAD — KEEP
On 21 Aug, SYNDROME_LEAD (start from the syndrome) was KEEP. It is the only tier kept.
22 Aug — EDGE_CUTTING — TURN OFF
EDGE_CUTTING (only split different actions) was turned off. Missed reportables rose 6 → 9. Any rise is a hard-fail rule. This is C1's idea. It was measured separately, earlier. It is currently off.
LOOKAHEAD — TURN OFF
LOOKAHEAD (look-ahead ranking) was turned off.
ABSTAIN_SAFE — no measurable effect
ABSTAIN_SAFE (safe I-don't-know) had no measurable effect.
End mark
Four tiers were measured. Exactly one was kept. EDGE_CUTTING is OFF today. Nothing is deployed.
A later on/off test wave. The C1/C2/C3 test grid was a different, earlier experiment.
Four tiers measured on the ship-or-stop test. Only SYNDROME_LEAD is KEEP. EDGE_CUTTING is OFF today. Nothing is deployed.