Why we pivoted then
Every item in F12's table is closed or blocked. F3 is fourth in the ranking, after F12; the ranking is unchanged.
What happened before
- F12.1 blocked without a run: no image of K4's section of the copper at letter resolution is in the repository.
- 0168: K4 measured the way Stein measured K1-K3, 71 statistics fixed before the run, each against 1,000 shuffles: nothing outside every shuffle.
- 0169: Stein's index-of-coincidence peak at 13 strands is unusual in K1 (p 0.0075, about 0.02 allowing for the strand counts he plotted) and absent in Part II of K2, where he saw it again: a coincidence.
- 0170: a pencil session, K4 written out at seven widths; six patterns noted by eye, none outside 1,000 shuffles.
- 0171: the two known runs as two parts with their own repeating keys: no key length fits either part (copies 0-1 cells).
- 0172: sixty probable words at every place they fit, in the ten settings the known letters leave undecided, each placement decided exactly: K4 fits as often as its scrambled copies (above every copy in 1 of 10 settings; copies 0-3).
- F12.7 closed by subsumption: Stein's column method finds periodic keys on a known table, which 0005's exact sweeps already cover.
The change we believe is needed
- F3 next, in its table's order, starting with F3.1's sandwich from Scheidt's patent (US 6,075,865): c = T2[T1[p + a] + b], T1 and T2 from the standard and KRYPTOS orders and their inverses, then the other combiners.
- Each F3 item is a fixed-table family with few parameters, so it can be enumerated exactly against the 24 known letters, K4 and kept-place copies, without an English model.