I know I may seem overly cautious, but sometimes it is worth to think countless times about risks than to relay on assumption that those risks will never happen. I am proud to be part of Stellar network advocates from its beginnings but topic I want to rise is fault tolerance of it.
In essence Stellar consensus does what Internet does: users are all interconnected most of the time based on fact that ISPs creates their individual routing slices to other ISPs etc. If ISPs fails we might get isolated for a moment, but it is generally accepted amd nobody expects truly 100% uptime. Sometimes 99.5% is enough, other times 80% is good enough too.
Stellar consensus does what is technically possible to create new financial network but it seems that technical controls are not enough to make complete model that guarantee 100% connectivity at all times. I had discussed some forking scenarios before.
Topic of interest for me:
What complementary controls could be implemented in Stellar network to make it bullet-prove?
Logic is that if model will be good enough it will be implemented, as there seems to be lack of non technical controls that might be helpful.