alecmuffett / draft-muffett-end-to-end-secure-messaging Goto Github PK

I am not entirely sure what you are trying to achieve with the approximate length. You would need to have a lot of extra bits, well beyond the block size of a block cipher) to make it hard to guess language.

Consider an attacker who knows (through external sources) that a message communicated is in response to a sneeze. The number of bytes in

• Salud
• Bless you
• Gesundheit
• Egeszségedre

don't quite illustrate the point, as all of them would fit in a single block. But with longer texts, the distinctions would matter. After all, The Spanish is five bytes and the Hungarian is 12 or 13 bytes (depending on precisely how the é is encoded.)

Typically, cryptographic notions of secrecy don't include attempting to conceal anything about the length of the plaintext because it is hard to do. So I'd like to know what you are trying to prevent by attempting such a thing.

ref: https://twitter.com/AlecMuffett/status/1393960691959226371

Leaving aside the precise length stuff, I believe that it would be better to define PSCAM in terms of indistinguishability as used in the cryptographic information.

The idea is that it would be any thing that would give an attacker the ability to make a better than 50/50 guess at any bit of the message. You could think of it as any information that would give the attacker an advantage in determining which of two plaintexts is more likely.

For example, an attacker might already have a fair guess that a message is in English, but nothing that is revealed through the system should give them any help in refining that guess. (Again, leaving length aside). The way cryptographers define indistinguishability captures this.

All participants
**MUST** be peers and

I do not know what that means. And I am wondering if it is necessary. (I don't know that it isn't necessary because I don't know what it means.)

I think Email with PGP may deserve a caveat with regards to how transparent to the user it may be. Email senders can use BCC. A user could inspect the message itself with command line tools, but not all email clients themselves will behave the same with BCC messages. The PGP is the transparent part. The email it comes in can be less so.

https://blog.cryptographyengineering.com/2018/01/10/attack-of-the-week-group-messaging-in-whatsapp-and-signal/

I believe that the word "comprise" is misused (in two places) in the document.

I have (locally) a branch with fixes to that and another minor wording infelicity.

Is there a specific reason why the definitions are limited to software? Would "system" be more appropriate terminology.