Thanks for sharing @jamesray1
The Spectre paper claims that the protocol is ‘secure’ (as defined by them i.e. satisfying Properties 1-3) if the attacker controls less than 50% of the computational power.
I believe Lamprot, Shostak & Pease showed that this is only possible in a world without firewalls, else any distributed network is only safe if an attacker controls less than 1/3rd of the computational power.
(Theorem 3 in https://people.eecs.berkeley.edu/~luca/cs174/byzantine.pdf).
Do Sompolinsky and Zohar simply assume that there aren’t any firewalls? If not, how do they get to 50%?