Why we pivoted then
Every item in F1's table is now closed or blocked, which is the rule for leaving a family (issue #29). The last open item, F1.1 (one alphabet in an unknown order on both sides of the table), was settled exactly at every key length but two: the Beaufort form at lengths 23 and 26 fits K4, as it fits most scrambled copies. Only an English climb could go further, and 0120 showed that no climb at this size recovers a planted message, so a further test would be weak. F1.1 was marked blocked with that reason rather than run weakly. F6 is second in the ranking and the ranking is unchanged.
What happened before
- 0119-0124 decided one and two unknown alphabets with no move exactly: no fit at most key lengths, fits at the rest at the scrambled-copy rate; the rare length-19 fit for two alphabets did not survive 1,000 fresh copies.
- 0125 read the key forced at the known places as running English on every dictionary keyword table: inside the copies, below every planted running key.
- 0126-0128 listed and tested the hand schedules for a key word and the tableau rows as keys: dead on the sculpture's tables, at the copies' rate on free ones.
The change we believe is needed
Carry the exact free-alphabet decision of 0119 and 0123 through every move of the type-1 library (F6.1, then F6.2), where 0097 had only a relaxation at three key lengths; then the mixed three-stage routes (F6.3) and the hand padding rules (F6.4). Cells that fit at the copies' rate are left blocked for the same reason as F1.1.