The idea of the longest chain being the one that takes precedence
Dynamic Availability
No notion of finality, though typically ~6 blocks is seen as a safe amount to prevent a reorg.
This number solely depends on number of malicious validators, as shown in some paper ( [[Scalable and Probabilistic Leaderless BFT Consensus Through Metastability]] )
synchronous, where the required difficulty of PoW is dependent on max network delay
the number of required blocks for a transaction increases exponentially with the number of adversarial nodes, with an asymptote at f = n/2 wherein the number is infinte