Why we pivoted then
Every F13 row is closed or blocked. F13.4f (0241) and F13.5 (0243) closed at full power: when the key is drawn from a word list, the ciphertext alone decides it (plants 32 of 32 and 40 of 40), and K4 sits inside its shuffles both times (p 0.42, p 0.099). F13.4f's second half, F13.4g (Chaocipher) and F13.6 were blocked at the desk with probes: two dictionary alphabets are beyond any screen without the known letters, the left wheel of a Chaocipher is not singled out by 97 letters, and the A13 battery sees no mask on readings chosen by English. D1 and D2 still wait on Alexis (#100).
What happened before
- F13.1-F13.3 set the bits: 97 letters of English carry about 173 bits by quadgrams, 189 with MIX8m.
- F13.4a-e (0234-0240): fixed tables, dictionary tables, one free alphabet, short moves and Hill decided from the ciphertext alone where their plants are found; free keys and free alphabets beyond about 100 bits are not.
- 0237-0239: one replicated excess at table-then-move, key length 2, junk reading; Alexis's call (#101).
The change we believe is needed
F13's lesson carries straight into F14: a key made of words is a few dozen bits where a free alphabet is 88, and those cells can be decided. F14 takes Scheidt's mask in its keyed forms, with the known letters back in use: F14.1 first (two dictionary key-word alphabets plus a key word at key lengths 19-26), cost estimated before any run.