Simple and elegant leader change (censorship and network failure resistance). In traditional pBFT consensus like Tendermint if the leader (block proposer) fails or becomes malicious the system will not be able to reach consensus and finalize blocks. Changing leader is extremely complicated and communication intensive. Most implementations have had to change certain assumptions to avoid dealing with leader change. In Tendermint, for example, the developer assumed that the network has strong synchrony (meaning all the nodes share the same well-synced global clock). If a leader fails to reach consensus within the time slot all the validator nodes will move to the next round. HotStuff-based BFT has a built-in leader change algorithm. The complexity of changing the leader is the same as that of creating a regular block. One of the advantages of HotStuff consensus is that it doesn't rely on the synchronized global clock. It assumes the network is partially synchronous, meaning the network may be jammed or partitioned from time to time, but will eventually become synchronized. This brings the tremendous benefit of preventing malicious nodes and censorship in the network. ETH 2's Casper FFG is a variant of BFT consensus with a partially synchronous network assumption with the help of the VRF random beacon. This however requires a different trade off of liveness and consistency. In ETH2 the system continuously generates new blocks even though there are not enough validators to reach a 2/3 majority and finalize the block (the blocks then become unusable as they could be reverted in the future). In Meter's PoV we use PoW to create a source of randomness, while the HotStuff consensus halts committing new blocks if 2/3s majority is not reached. This is much more desirable behavior for financial applications.