Proposal for a random encrypt method

[ayanamist]

Proposal for a random encrypt method

[sorz]

I don't think difference exists among different ciphers, in terms of "traffic characteristic".
AFAIK, they all keep generating (pseudo) random bits and then XOR with plaintext.

[librehat]

Unnecessary overhead for the protocol without any obvious benefits. Again, the encrypted data in shadowsocks is not possible to decrypt without knowing the method and key.

If the key is compromised, this method doesn't really work since you have the method in datagram.

Plus not all shadowsocks ports are created equal (different support level for encryption methods)

[ayanamist]

It emphasis on dataflow obfuscation in order to decrease the risk of being detected by dataflow analysis, but not on encryption itself.
By changing method for each connection, it will increase much more chance to survive. There are reports that the ip port used as ss will be slow down or blocked after some time usage.

[madeye]

To me, it looks no helpful to obfuscate data in that "random" way. Instead, it will introduce great overhead on server side.

The state-of-the-art way of obfuscation is obfsproxy, which hides encrypted traffic in "normal" traffic, not "random" traffic. So, if you really want to obfuscate traffic of shadowsocks, obfsproxy is a better choice.

[ayanamist]

I know you guys want to keep UNIX philosophy, but don't concern only on self-hosted. There is no selling ones which provides additional protect for shadowsocks like obfsproxy.

[madeye]

I think it has nothing to do with UNIX philosophy.

As mentioned above, the major concern from us is that your proposed random approach may not help to obfuscate traffic but introduce large overhead. Actually, IV in shadowsocks protocol is just for randomizing keys. And it works quite well.

[ayanamist]

No, neither IV nor OTA will remove or decrease the traffic characteristic of shadowsocks connection.

[Mygod]

Veto on this proposal. Actually why isn't this closed yet?

[madeye]

Closed as we have #26 now.