Why we pivoted then
Every item in F3's table is closed. F2 is fifth in the ranking, after F3; the ranking is unchanged.
What happened before
- 0173: the text/key relation from Scheidt and Wack's patent US 6,075,865 with one keyed table between two repeating keys (108 table classes, keys 1-13 each), decided exactly. Nothing fits K4 wherever the known letters give 4 or more checks (161 of 169 length pairs). The 8 pairs with both keys 10-13 long are too weak to decide.
- 0174: two key words joined through a mixed numbering or by multiplication: no consistent case on K4 or on any copy.
- 0175, 0176, 0178, 0179, 0180: numbers from the sculpture as the number half of a key, each with an unknown word of 1-24 letters on 12 table rules. The readings were the Morse E counts, the plate's line lengths, the directions as degrees and compass points, where the misspelt letters sit, the World Clock's ring, and "point" (the decimal point and the punctuation marks). Nothing fits K4 or any of 20 scrambled copies, and every planted text was found.
- 0177: the patent's two-table stack with three keys of 1-10, decided exactly by a compiled search: nothing fits at 4 or more checks.
The change we believe is needed
- F2 next, in its table's order: F2.1 (the forced fragments of 0057 matched against the dictionary), F2.2 (nulls placed by a rule), F2.3 (homophones from a phrase with repeated letters).
- F2's mask changes which letters stand at which places. So each test must state how the 24 known letters constrain it before anything is run.