“Plaintext autokey, chained: the 24 known letters force the rest, lag by lag”
0003 only checked the cribs against each other. Here each known letter is also used as the key for the letter one lag later, and as the ciphertext's partner one lag earlier, and the forced letters are chained until nothing new follows. For small lags that fixes the whole text.
In an autokey cipheran autokey cipher A cipher that uses the message itself, or the scrambled text itself, as the secret word, each letter enciphered with the letter a fixed distance before it.the messagethe message The 97 letters of English that Jim Sanborn wrote and enciphered. is its own secret word: each letter is enciphered using the letter some fixed distance before it. That makes the 24 known lettersthe known letters Sanborn has confirmed that positions 22 to 34 of the message read EASTNORTHEAST and positions 64 to 74 read BERLINCLOCK. very powerful. Each one was used to encipher a letter further on, so that letter can be worked out; and each one was itself enciphered by a letter further back, so that one can be worked out too. Repeating this spreads outward until, for short distances, the whole message is known. We try every distance and every way of using the tablethe table The right-hand half of the sculpture, 26 rows of the alphabet each shifted one place further, starting from the word KRYPTOS., and read what comes out.
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 chain
In a plaintext autokey each letter is keyed by the message letter 13 places before it. So each known letter reveals the letter 13 places later and the one 13 places earlier, and those reveal more.
where chains meet
Step by step the chains reach every place, and two routes to one place must give the same letter. At lag 13 they disagree under every way of using the table, so lag 13 is dead.
the bound
Lags 1 to 25 and 30 to 52 die under all 24 ways of using the table. Lags 26 to 29 and 53 upwards survive but force too little to build a reading.
2
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 in section 7.
3
Walkthrough, by hand
Redo the method on a few letters with pencil and paper.
Take lag 13 and the sculpture's alphabet with Vigenere.
Position 22 is E and its ciphertext is F; under the table, the key that turns E into F is R, so position 9 is R.
Position 35 was enciphered with key E, the letter at 22; the ciphertext at 35 is S, and S decrypted with key E is the letter at 35.
Now 9 and 35 are known, so 48 and 61 follow, and so on, up and down by 13 until the chain leaves the text.
Do the same from every one of the 24 known letters.
Where two chains meet, the letters must agree, or lag 13 is dead under this convention.
If every position fills, read the text aloud.
The first 13 letters were enciphered with the primer; the key letter that maps each of them to its ciphertext is the primer, which should be a word.
the worksheet: the first steps by hand, on squared paper
4
Worked example, generated
The same steps run by the code on the real letters.
lag 13, KRYPTOS alphabet throughout, Vigenere: chaining from the cribs
consistent: False; letters forced: 28 of 97
.....................EASTNORTHEASTS............N............K..BERLINCLOCK............K..........
position 22 (E, cipher F) forces position 9 = R; key E at 22 decrypts cipher O at 35 to Q
5
Hypothesis, for a cryptanalyst
Family, key space, the constraint under test, and the clues it reads.
Plaintext autokey (the NSA's 1992 guess, C11a; a Scheidt-era hand system, C5): c[i] = E(p[i], p[i-L]) for i >= L after a primer of length L. A crib letter p[i] forces p[i-L] through c[i] and forces p[i+L] through c[i+L]; each forced letter forces two more. The closure from the 24 cribs covers every residue class modulo L that a crib position touches, so for L <= 13 the whole text is forced, for L up to about 24 most of it, and for larger L two or three stretches. 0003 found lags 1-11 and 30-52 inconsistent inside the cribs; lags 12-29 and 53-96 were undecided. Here, under each of the 24 conventions and each lag 1-96, we propagate to closure, reject on any conflict, and for a fully forced text derive the primer and submit the candidate. Partially forced texts are scored as English on their forced runs and compared with the same procedure on shuffled K4 letters. Bounds: complete for plaintext autokey with a primer of length L, any L 1-96, under these conventions.
6
Prediction
Written before the run; the result decides it.
Every lag from 1 to 25 either conflicts somewhere in the chain or yields a fully forced text that is not English, scoring below -6 per quadgram against -3.6 for English. Lags above 25 leave stretches that score no better than the shuffled control. No candidate passes the English standard; a fully forced text may pass the four gates by construction and will then stand as a champion with a loss of 3 to 4, which is a gibberish champion, not a solution.
7
Method
For each convention and lag L, start with the 24 cribs as forced letters.
For each forced letter at i: if i >= L, force p[i-L] as the key that maps p[i] to c[i]; if i + L < 97, force p[i+L] by decrypting c[i+L] with key p[i].
Repeat until nothing new appears; any conflict kills the pair.
If all 97 letters are forced, derive the primer k[i] for i < L and submit the text.
Otherwise score the forced runs of four or more letters as English.
Run the same procedure on ten shuffles of K4 for a control.
8
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
1129 consistent (lag, convention) pairs; 0 fully forced texts submitted. A fully forced text passes the crib gate by construction; only its English score matters.
What was searched, as the harness recorded it.
family
plaintext autokey with primer, chained closure from cribs