“F1.14's gate: 0267's 24 plants (hand key schedules on two free alphabets) and 12 kept-place copies each, climbed ten times, power after r runs for r = 1-10”
F1.14 (rerun of F1.4's two-alphabet residue at ten runs) waits on this gate: does ten runs lift 0267's plants by 3 or more of 24 over four? 0267's plants are regenerated exactly (its frozen setting scan and plant draws), and each plant and 12 kept-place copies (from the plant's own witness, as 0267) are climbed with ten independent runs of 0267's generalised tempering, every run's best kept, as D1b (0257) did for periodic keys.
Last run, our search found only half of the test messages we hid, in a version of the cipher where Sanborn steps through his key wordthe secret word The word or string a cipher needs besides the method. with a twist and uses two scrambled alphabets. Searching each text ten times instead of four might find more. Before paying for the full ten-times rerun, we check on the same 24 hidden messages whether ten searches find at least three more than four.
Words in the margin are defined on the glossary page.
1.1
The method at work
The carved letters go in at the top; the reading comes out at the bottom.
the idea
0267 found half its hidden messages; would ten searches per text instead of four find more, worth a $25 rerun?
the check
0267's 24 messages regenerated exactly, 12 shuffled copies each, ten searches per text, every search's best kept.
the bound
Found after 1, 4 and 10 searches: 14, 12, 14 of 24. The power is flat: the score, not the search, limits it. F1.14 closed.
2
Experiment lineage
What this experiment was built on, from the lineage page: 20 experiments before it, back to the root of its line. Its number is in bold.
keptreverted, in its family's colour×error or timeoutreverted under today's rulesbuilt on a parent in the same familybuilt on a parent in another family
3
Best candidate against the known letters
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.
4
Walkthrough, by hand
Redo the method on a few letters with pencil and paper.
Take one of 0267's hidden test messages, enciphered with a skipping key schedule and two scrambled alphabets.
Make 12 copies with the 73 letters at unknown places shuffled.
Search each of the 13 texts ten separate times for its most English-looking reading; write down each search's best.
For r = 1 to 10, compare the message's best of its first r searches with each copy's best of its first r.
Count the messages that beat all 12 copies at r = 4 and at r = 10.
the worksheet: the first steps by hand, on squared paper
5
Worked example, generated
The same steps run by the code on the real letters.
F1.14's gate: does searching ten times instead of four find more hidden messages?
one hidden message and its 12 shuffled copies, each searched ten times; best of the first r searches:
found at r when the message's best beats all 12 copies' bests.
24 messages (0267's own); run the full rerun only if r = 10 finds at least 3 more than r = 4.
6
Hypothesis, for a cryptanalyst
Family, key space, the constraint under test, and the clues it reads.
0267 found 12 of 24 plants (8 clean, 4 Sanborn-style) with D1b's four runs on hand key schedules (0128's skip, stutter, back and forth, triangle, square, restart per line) over two free alphabets (Quagmire IV), against D1b's 66 of 96 for periodic keys at 13 and 16. The settings here have up to 31 slots, so the climb has more key letters to find and may profit more from extra runs than D1b measured (+5 of 96 from four to ten). The gate set in F1.14's row: run F1.14 (about $25 on AWS) only if ten runs find at least 3 more of these 24 than four. Clue entries C20(d), C27(c).
7
Prediction
Written before the run; the result decides it.
Four runs find 11-13 of 24 (0267: 12); ten runs find 1-3 more (D1b's +5 of 96 scaled, a little more for the longer keys). Gate likely not met. Zero candidates.
8
Method
0267's run() steps 1-2 unchanged from its frozen attack.py: every 0128 schedule decided exactly on K4 on two free alphabets (4,706 fit), then its 24 plants (12 clean, 12 Sanborn-style) drawn on fitting settings with its seed, each with an exact witness.
Per plant: the plant and 12 kept-place copies of its ciphertext, each climbed with 10 independent runs of 0267's climb (one run per seed), every run's best score kept.
Found after r runs: the plant's best of runs 1..r above all 12 copies' bests of runs 1..r.
Gate: found(10) - found(4) >= 3.
9
Result and bounds
The harness's record: candidates, gates, loss, and the negative result this experiment leaves behind.
0candidates returned
0passed every gate
0/24known letters, best
–lowest loss
nonechampion before
F1.14 gate: 0267's plants found after 4 runs 12/24, after 10 14/24 (gain 2, gate 3: not met, F1.14 closed); power by runs [14, 14, 12, 12, 13, 13, 13, 14, 14, 14].
What was searched, as the harness recorded it.
item
F1.14 gate
settings fit k4
4,706
plants match 0267 settings
24
plants
24
own copies
12
null per plant
1/13
decider
0267's climb (0144's tempering generalised to any slot map, 16 x 625 x 1000 per run) from the plant's exact witness
runs
10
power by runs
1
clean
8
sanborn
6
total
14
2
clean
7
sanborn
7
total
14
3
clean
6
sanborn
6
total
12
4
clean
6
sanborn
6
total
12
5
clean
7
sanborn
6
total
13
6
clean
8
sanborn
5
total
13
7
clean
8
sanborn
5
total
13
8
clean
9
sanborn
5
total
14
9
clean
9
sanborn
5
total
14
10
clean
9
sanborn
5
total
14
gain 10 over 4
2
gate
3
gate met
no
record 0267 plants
8/12 clean, 4/12 Sanborn-style at four runs, copies one at a time
each
0
cls
clean
setting
B skip m=26 phi=18 n=8
slots
8
found at 4
yes
found at 10
yes
letters right 10
53
1
cls
clean
setting
B skip m=9 phi=0 n=23
slots
23
found at 4
no
found at 10
no
letters right 10
29
2
cls
clean
setting
B skip m=16 phi=15 n=19
slots
19
found at 4
no
found at 10
no
letters right 10
36
3
cls
clean
setting
B skip m=28 phi=19 n=19
slots
19
found at 4
yes
found at 10
yes
letters right 10
41
4
cls
clean
setting
B skip m=19 phi=5 n=26
slots
26
found at 4
yes
found at 10
no
letters right 10
36
5
cls
clean
setting
B skip m=13 phi=10 n=13
slots
13
found at 4
no
found at 10
yes
letters right 10
49
6
cls
clean
setting
B skip m=25 phi=7 n=12
slots
12
found at 4
no
found at 10
yes
letters right 10
38
7
cls
clean
setting
B skip m=17 phi=7 n=20
slots
20
found at 4
yes
found at 10
yes
letters right 10
35
8
cls
clean
setting
B skip m=25 phi=12 n=8
slots
8
found at 4
yes
found at 10
yes
letters right 10
47
9
cls
clean
setting
B skip m=17 phi=3 n=14
slots
14
found at 4
no
found at 10
yes
letters right 10
34
10
cls
clean
setting
B skip m=29 phi=6 n=12
slots
12
found at 4
yes
found at 10
yes
letters right 10
40
11
cls
clean
setting
B skip m=21 phi=3 n=14
slots
14
found at 4
no
found at 10
yes
letters right 10
39
12
cls
xstops
setting
B skip m=28 phi=11 n=12
slots
12
found at 4
yes
found at 10
yes
letters right 10
42
13
cls
xstops
setting
B skip m=30 phi=18 n=21
slots
21
found at 4
no
found at 10
no
letters right 10
36
14
cls
xstops
setting
B skip m=27 phi=3 n=24
slots
24
found at 4
no
found at 10
no
letters right 10
38
15
cls
xstops
setting
B skip m=11 phi=0 n=26
slots
26
found at 4
yes
found at 10
yes
letters right 10
39
16
cls
xstops
setting
B skip m=26 phi=2 n=17
slots
17
found at 4
yes
found at 10
yes
letters right 10
41
17
cls
xstops
setting
B skip m=27 phi=22 n=20
slots
20
found at 4
no
found at 10
no
letters right 10
28
18
cls
xstops
setting
B skip m=27 phi=22 n=25
slots
25
found at 4
no
found at 10
no
letters right 10
34
19
cls
xstops
setting
B skip m=22 phi=0 n=21
slots
21
found at 4
no
found at 10
no
letters right 10
35
20
cls
xstops
setting
B skip m=23 phi=22 n=26
slots
26
found at 4
yes
found at 10
yes
letters right 10
34
21
cls
xstops
setting
B skip m=3 phi=1 n=21
slots
21
found at 4
yes
found at 10
yes
letters right 10
33
22
cls
xstops
setting
B skip m=27 phi=2 n=12
slots
12
found at 4
yes
found at 10
no
letters right 10
44
23
cls
xstops
setting
D back and forth n=15 ends=repeated phi=0
slots
15
found at 4
no
found at 10
no
letters right 10
34
texts
312
core seconds
37,508
compute seconds
1,249
resumed units
0
exhaustive
no
cost estimate
312 texts x 10 runs x about 13 s = about 11 core-hours, plus 0267's setting scan; on AWS about 30 min, about $0.30
How this experiment sits among the others is on the lineage page; its family's open hypotheses are on the families page.
Rendered from the record on 2026-11-05 17:14 UTC; updated after every experiment.