Regarding putting trust on the table as a design solution to possible attacks that aren't happening, wouldn't it be wise to start with whatever trust solutions common networks already use and mutate that to this situation?
Example: KYC, black/whitelisting, reputation scoring, permissioned/private subnets, scoring/tiers Of course, none of us actually want to design formal KYC into LN, but it really is the same premise: "identifying" your counterparty and assessing the amount of risk you would like to allocate. Permissioned access, reduced public listening, out-of-band credentialing, etc, etc. Networking issues like these might not even be appropriate to be handled at the LN level. Possibly better to use a multipurpose context layer (gets into ID systems, namespaces, WoTs). Sorry if I'm oversimplifying the topic, but it is frustrating to watch devs argue about the edge of an edge of a knife no one is using, and then bikeshed every imperfection... Cryptographic punishment schemes aren't swiss army knives.
_______________________________________________ Lightning-dev mailing list [email protected] https://lists.linuxfoundation.org/mailman/listinfo/lightning-dev
