In order to mitigate sybil attacks, peers should detect unusual request patterns

[sanity]

Peers should look for unusual statistical patterns in requests coming from their neighbors. For example, a bunch of new contracts being added all with unusually similar locations might be indicative of a Sybil attack. Peers with very unusual such patterns should be disconnected.

[Destroyinator69420]

I agree, there should be a strict definition of how a locutus node should behave. If a locutus node does not fit that pattern, then the connection should be dropped. If it does not behave exactly like the reference implementation, (the locutus node in this repository) then it should be shunned from the network. I also think that the symmetric encryption locutus uses should not place trust in a single algorithm. I would recommend AES-256(Kuznyechik-256(Kalyna-512(Message))). We shoudl use these algorithms in XTS mode. It would help the censorship resistance of Locutus if all messages sent through the network are of one standard size in the ciphertext, but from an outside observer they are compressed with ZSTD and padded by no more than 10%.

[ple1n]

Again, that's not really how censorship resistance works. You get easily detected if your packets are statistically uniform and 'featureless', and/or of same size. This is proven by https://gfw.report research and such.