Is Chaocipher a secure cipher under ciphertext-only attack?

Chaocipher was invented by John F. Byrne in 1919. The algorithm was recently revealed – see Moshe Rubin's Chaocipher Revealed, the Algorithm (PDF).

While a known plaintext attack successfully finds the keys, nobody has been able to put forward a general solution to this cipher. Is that possible?

-
Any cipher than suffers key-recovery under known plaintext is horrifically weak under modern requirements. Nowadays, we tend to require that an attacker cannot even recognise the ciphertext (compared to random data), even if they can choose the plaintext to encrypt and even ask for the decryption of some messages of their own choice –  figlesquidge Apr 8 at 9:40
Ciphertext-only attacks virtually always assume some a priori distribution of plaintexts; otherwise all keys are equally probable. –  Dmitry Khovratovich Apr 8 at 10:20
Thank you both form your responses. I have a follow-up for figlesquidge please. Given that the cipher text from Chaocipher shows some deviation from randomness in one respect (which it does!) how then does the cryptanalyst go about solving the cipher? He knows the algorithm, but has no idea of the two 26-character keys. –  user2256790 Apr 8 at 13:26
Because the key is relatively large [ $2\log_2(26!)>176\text{ bits}$ ], and since the cipher is not trivially bad, a ciphertext-only attack can only be carried with significant amount of ciphertext corresponding to redundant plaintext. If we consider there is 2 bit/letter of exploitable redundancy in English text, and IF the cipher was perfect, we would need about 90-letter ciphertext to have any hope of solving it. Are there large Chaocipher challenges around? (or course it is easy to make some). –  fgrieu Apr 9 at 8:24
There are some lengthy Chaocipher ciphertexts here: mountainvistasoft.com/chaocipher/Chaocipher-ASCII-versions.htm I (and several others) have broken some of these with a known plaintext attack. But nobody (to my knowledge) has developed a method to solve using a ciphertext attack. Such a method would be of great interest. –  user2256790 Apr 27 at 15:55

While a known plaintext attack successfully finds the keys, nobody has been able to put forward a general solution to this cipher. Is that possible?

You really have to go back in time to learn that ChaoCipher has been subject to some cryptanalysis before the description of the algorithm/device was published. As an example: Here’s one of the oldest papers I found related to the cryptanalysis of the ChaoCipher which was first published in 2003… long before the algorithm was published in 2010 and long before the cipher was subject to a complete break.

If you do a bit of research, you’ll find a few more papers. Most cryptanalytic looks at ChaoCipher have shown ample weaknesses in the distribution, producing anomalies which could be exploited.

Having used the algorithm description to code a C version, I quickly discovered it works somewhat like a substitution cipher that incorporates a primitive version of an which scrambles the S-box. Nothing you would want to use in times of modern cryptography. It’s more like a most-primitive rotor machine.

Anyway… getting back to the core of your question: is it possible that no one has created a general “this breaks it from every side” solution?

Sure! One reason can be found in the fact that a counts as a complete break (and in this case, a practically feasible one too). Finding additional attack vectors can be fun as a hobby, or it may be something you might want to check on while writing a thesis about ChaoCipher… but it’s not that interesting for most cryptanalysts to dive in deeper when a cryptographic algorithm already was subject to a complete break. Instead, cryptanalysts will be more interested in finding weaknesses in more modern (currently used) ciphers. Compared to modern cryptography, ChaoCipher can’t hold up.

On the other hand, there are ample websites that talk about its internals… leaving the chance that one day, someone may find ways to break it from all sides – after investing enough time and efforts.

-