speed of transactions does not need to be fixed

10 minutes and the latency of the global internet are a perfect match and lightning state channels (two way transactions that can be securely updated to match a flow of sats in either direction) solve the latency problem, they just bring with them the problem of the brittleness of a source routed network system where a transaction path may not work between when the state of the network was acquired and when a vital hop in the path became congested or went offline

solving these issues can be done other ways, and i have pondered a lot the idea of atomic multipath redundant pathfinding, so transactions literally travel light lightning bolts across the paths that are open and the first one that reaches the end settles it, instead of the standard single path or atomic multipath (AMP) patterns where only one failure in all of the hops in the path causes a failure

AMP only makes bigger payments possible, it did not add redundancy