blockchain - 区块链共识 51% vs 50%

Question

区块链共识算法的所有解释都将 51% 作为确认节点的最小百分比。为什么较小的百分比不起作用，只要它严格大于 50%？这是为了避免共识动态中的亚稳态问题，这些问题可能是由于百分比太接近 50% 导致其决定性降低？如果是这样，为什么不将其设为 52%？在某些权衡中是否有什么特别之处可以使 51% 变得特别？

标签： blockchainconsensus

Answer 1

51% is just a notation for (n/2 + 1) where n: the total number of nodes participating in the consensus algorithm. I agree that it is a bad notation, more precisely it should be something like 50% + ε.

In academic papers you might find relations with the number of Byzantine nodes f in the system. So for PoW consensus, n > 2f + 1 (which is equivalent to the statement that at more that half of the nodes should be "honest"). For PBFT consensus in permissioned blockchains, it will be n > 3f + 1 (which translates to honest nodes should be more than double of the Byzantine nodes, or using the "bad" notation, 66.6666% + ε).