Initial impressions of Lightning on testnet: c-lightning segfaults a lot, and when it's not crashing payments fail more often than not. Writing it in C - a notoriously dangerous language - doesn't strike me as a good idea.
The Android wallet Eclair is equally unreliable, and I got it into a state where it crashes on startup, effectively losing funds.
Haven't tried lnd yet, although the decision to use an alt implementation backend (btcd) is obviously a bad idea given how critical consensus compatibility is for Lightning.
As for the Lightning protocol, I'm willing to predict it'll prove to be vulnerable to DoS attacks in it's current incarnation, both at the P2P and blockchain level.
While bad politics, focusing on centralized hub-and-spoke payment channels first would have been much simpler.
While bad politics, focusing on centralized hub-and-spoke payment channels first would have been much simpler.
