lightning's source routing system is brittle like this, and it's difficult to even gather useful data about the network, as well as propagate it effectively to mitigate this kind of fault
there's no doubt that there needs to be further development of the protocol
If more noobs will understand that is NOT really necessary that everybody should be a "public routing node", the network will be much better as it is today.
Most of force closing channels happen EXACTLY because of these shity nodes. And if we can limit to minimum necessary the force closed channels, LN will work quite well.
Having a LN structure by levels and users choose their own level based on their own capacity, will make things to go smooth and well.
High level: big routing nodes, professional ones, well maintained, huge liquidity. Providing high availability and liquidity for big routes.
Medium level: LSPs, well connected to the high level, providing liquidity and routes for private nodes and users.
Low level: edge nodes, private nodes, mobile private nodes, regular users.
But nooo, every noob now want to run a routing node with a shity RPi over Tor, "to earn sats and help the network", when in fact are only clogging the network.
i don't think #lightningnetwork is the final destination for L2s
i haven't looked deeply into Ark but my first impression is kinda favourable
i think that the problem we face is how to do trustless relaying of messages across the network that don't involve falsified UTXOs and can fall back to alternative routes when a planned route fails
it's a cryptography problem, and i'm almost certain there is already a specialised hash function that exactly does what we need, just nobody has figured out that it can do this
you also rage against the AI, i hate that shit, and you know it operates by a specialised hash function right? the AI hash is called a "proximity hash" and it allows you to plot multiple paths through a graph that generate a valid structure that humans recognise (they are like human memories, in fact)
something very close to this will enable us to do trustless path routing and the days of stuck payments will be over, just like QUIC and other new reliable transports have ended (most of) the connection reset by peer problems
actually, just articulating that first point about failure resistance
already right now, if LN protocol had a notion of forking it could be literally like lightning and if it forks into a dead end it can step back one, two, three etc hops and try alternatives
the messages would be a lot bigger but they would have resistance
i started working on this idea with Indra last year, how to do forks, the https://github.com/indra-labs/indra repo has code for "forking" messages, that encode multiple routes for a packet so it can even duplicate across and iirc i even made a join message
