Experiment 257
D1b (issue #77). 0229 measured design 2's power per run for one free alphabet and D1 qualified a one-run decider (Alexis, 2026-10-28, #100). Here the same measurement for two free alphabets with 0144's compiled tempering, unchanged: F1.2's sampled lengths 13 and 16, 24 clean and 24 Sanborn-style plants each, 6 kept-place copies per plant, ten full runs per text with each run's best recorded. Two lengths so the power can be checked for sameness across the class before one measured power is used for every two-alphabet setting (F1.4's 1,346 two-alphabet settings, F1.2 13).
What the method does with the 24 known letters.
For the hardest version of the cipher, with two scrambled alphabets, our search tool's success rate was measured on only a handful of hidden test messages, and the counts disagreed between tries. Here we hide 96 test messages, half plain book English and half written Sanborn's way, and count how many the tool finds after one search, two, and so on up to ten. If one search finds about as many as ten, the tool becomes about thirty times cheaper, and two thousand untested settings come within reach.
The carved letters go in at the top; the reading comes out at the bottom.
D1b: the cheaper judge for the hardest version, two scrambled alphabets; does one search find as many hidden messages as ten?
96 planted messages, 6 scrambled copies each, ten searches per text, every search's best kept; 180 s per text for ten.
One search finds 63, ten find 71: one is not enough; four (66) are within 5. Power the same at 13 and 16 (30 and 33 of 48 at one search).
What this experiment was built on, from the lineage page: 17 experiments before it, back to the root of its line. Its number is in bold.
This experiment produced no candidate: the search ruled everything out before a full text was built. Its product is the bounded negative under Result and bounds below.
Redo the method on a few letters with pencil and paper.
The same steps run by the code on the real letters.
D1b measures how fast design 2 finds a planted message under two free alphabets. For one plant and its six copies:
run 1 2 3 ... 10
plant best of runs 1..r
copies best of runs 1..r, each
found at r when the plant's best is above every copy's best.
Over 96 plants at key lengths 13 and 16: one run qualifies when it finds within 5 of what ten runs find.
Family, key space, the constraint under test, and the clues it reads.
F1.5 design 2 (0143, 0144, 0146) closed F1.2's lengths 8, 13 and 16 on power from 4-5 clean plants each; 0226 and 0228 redrew L13 on Sanborn-style plants (9/12, 8/12) and on clean ones (2/5): two-alphabet power is in doubt and rests on too few plants. Two questions on one set of climbs, as 0229 for one alphabet: (1) design 2's ten-run power at 13 and 16 on clean and Sanborn-style text (judge bench classes xstops, k123, sub3; A6, C17, C19), 24 plants per class and length; (2) the power of the first r runs, r = 1-10. D1's rule (Alexis, #100): a cheaper decider qualifies when its power pooled over the cells (here 96 plants) is within 5 of the reference on the same plants. Sameness: the one-run power at 13 and at 16 within 5 of 48 of each other. Clue entries C27(c) (two alphabets: Sanborn's 'slightly different' tableau on both sides).
Written before the run; the result decides it.
Ten-run power at 13 about 0.6 clean and 0.6 Sanborn-style, at 16 a little higher (fewer checks keep less of the copies' freedom out, but longer columns read better). As for one alphabet, power flat in the number of runs: one run within 5 of 96 of ten, so a one-run decider qualifies for two alphabets; 13 and 16 within 5 of 48. Zero candidates.
The harness's record: candidates, gates, loss, and the negative result this experiment leaves behind.
D1b: pooled power by runs {1: 63, 2: 64, 3: 64, 4: 66, 5: 66, 6: 67, 7: 72, 8: 71, 9: 72, 10: 71}; one run 63/96 against ten 71/96; one run at 13 30/48, at 16 33/48; qualifying runs [4, 5, 6, 7, 8, 9, 10]; one-run decider does not qualify
What was searched, as the harness recorded it.
How this experiment sits among the others is on the lineage page; its family's open hypotheses are on the families page.