Post ASwsM4KyXRxTEzKn0C by Rucknium@pleroma.rucknium.me
 (DIR) More posts by Rucknium@pleroma.rucknium.me
 (DIR) Post #ASwsM4KyXRxTEzKn0C by Rucknium@pleroma.rucknium.me
       2023-02-22T20:23:53.013051Z
       
       0 likes, 0 repeats
       
       Is there any way to show (probabalistically) that a given byte sequence is encrypted instead of plaintext when the keys are not available to the verifier? Other than diehard-style statistical tests? #Monero is considering ways to restrict the use of the tx_extra field in its transactions.@matthew_d_green @socrates1024 @str4d or @filippo may know.
       
 (DIR) Post #ASxKkpwnySL6ZFcxe4 by str4d@abyssdomain.expert
       2023-02-23T01:30:13Z
       
       1 likes, 0 repeats
       
       @Rucknium @filippo @matthew_d_green @socrates1024 As I said in the #Zcash Discord, it would be possible to concretely prove this with a ZKP, as the transaction creator knows the encryption keys. The cost is implementing the encryption scheme inside a circuit (as well as the general-purpose ZKP cost if you don't already have it).Otherwise you're generally limited to looking for known plaintext structure (e.g. check if it's valid UTF-8), as the ciphertext usually has no visible structure.
       
 (DIR) Post #ASxKktwF8UGawfOZKC by str4d@abyssdomain.expert
       2023-02-23T01:33:10Z
       
       1 likes, 0 repeats
       
       @Rucknium @filippo @matthew_d_green @socrates1024 However, it's not necessarily impossible to detect ciphertexts directly, because the usual security definitions for encryption (e.g. IND-CPA or IND-CCA2) are about distinguishing the plaintext that was encrypted, and do not hide _that_ encryption occurred. https://crypto.stackexchange.com/a/74440 gives a nice overview of this (and shows how ciphertext-indistinguishability-from-random implies some more usual security definitions).
       
 (DIR) Post #ASxKkxfNH99MVVX9aS by str4d@abyssdomain.expert
       2023-02-23T01:33:22Z
       
       1 likes, 0 repeats
       
       @Rucknium @filippo @matthew_d_green @socrates1024 The relevant literature here would be on DPI systems and detection avoidance. See also this Noise mailing list thread: https://moderncrypto.org/mail-archive/noise/2018/001456.htmlBut it's still going to be much easier to detect plaintext structure, if you know the options are "it's encrypted, or it's not, but there is still going to be data present".