Here's an example cipher - I challenge anyone to try to break it. It is a standard homophonic substitution cipher, read and enciphered in the normal reading direction.
- Code: Select all
What makes these generated ciphers unique is that they share very many qualities with the actual Z340:
- Very similar symbol distribution
- Same number of repeated bigrams/trigrams
- Appearance of pivots
- Top 20 L=2 homophone cycles have the same "strength" as the top 20 from Z340
- Same "prime phobia" feature of + and B
- Simulated transcription errors (Up to 20 polyphones where the alternate symbol assignments come from set of visually similar symbols)
- Same number of repeated ngrams when even (or odd) positions removed.
- Same absence of repeated symbols on lines 1-3, 11-13
- Matches Jarlve's non-repeat measurement of the Z340 (4462)
- Same appearance of word-like symbol sequences ("zodiak", "her", "god")
- Same phenomenon of a trigram repeating in the same column, and that intersects with another repeating trigram
- Similar appearance of "box corner" patterns
- Similar appearance of "fold marks" on line 10
- Misspellings and missing words in the plaintext
- There is a section of filler
Here's a visual of some of the patterns in the generated cipher that are similar to Z340:
The hint for the plaintext is that it comes from a Gilbert and Sullivan play.
I fed the cipher into zkdecrypto and azdecrypt. They each were able to recover around 60% to 65% of the plaintext.
A 10-symbol section of the cipher is filler. And there are 20 intentional transcription errors. So, that contributes to 9% of the cipher text being messed up. On top of that there are several intentional misspellings. These factors are probably making it difficult to automatically recover more of the plaintext.
Another hint: Because of the poetic nature of the plain text, some phrases appear more often than normal. Perhaps that is also increasing the difficulty.