Review of: The Byzantine Generals Problem Summary: A mechanism for distributed computers to arrive at a correct concensus in the face of failures is developed using a Byzantine Generals allegory. The most important ideas in this paper are: -In the Byzantine Generals Problem the goal is to have the different units communicate with each other to decide upon a common desirable plan. This is an allegory for the collaboration in a distributed computer system. -Using non-signed messages there must be fewer than 1/3 traitors in the group to come to the correct agreement, even if the agreement is approximate. -If all loyal generals can communicate with each other via other loyal generals and they sign their messages then they can reach agreement if there are 2 or more of them. -All of the assumptions used in the Byzantine Generals Problem hold true with real distributed computer systems, as long as distributed clocks can be synchronized within some error. This means the same class of problems are solved for distributed computer systems. The largest one or two flaws in the paper are: -The paper went into great detail on how to use unsigned messages to come to a correct concensus as long as there were fewer than m traitors amongst 3m+1 generals. It was also shown that if signing was employed at least 2 generals have to be loyal, as long as they are connected. Signing seems to be vastly superior and obviates the more expensive non-signing problem. There was no discussion on why one would design a system without signing. Relevance of paper to modern systems: In distributed systems errors will occur. The Byzantine Generals Problem goes a step further by showing a mechanism that overcomes maliciously incorrect errors as long as they're kept below a certain threshold and a certain network connectivity is in place. There is an upper bound to the number of errors sustainable in the system. One good application of this information is in P2P networks. As long as the RIAA controls less than 1/3 -1 of the P2P nodes the file swapping may continue.