Hm, well things that are "overthought" would simply introduce a
   helpful redundancy in the system itself, allowing for backup
   trajectories of edges to reach the proper nodes in case of
   catastrophic failure.

   After all, any system that can be subject to a cascading
   failure, is NOT a properly designed system. You _need_
   independent pathways that only pass messages and NOT be
   dependent upon parallel processes... because then, well, they're
   not really parallel are they? They're shared.

   .... yeah, I guess you're right.