On Thu, Aug 6, 2015 at 8:43 PM, Michael Naber wrote: > How many nodes are necessary to ensure sufficient network reliability? > Ten, a hundred, a thousand? At what point do we hit the point of > diminishing returns, where adding extra nodes starts to have negligible > impact on the overall reliability of the system? > It's not about reliability. There are plenty of nodes currently for synchronization and other network functions. It's about reduction of trust. Running a full node and using it verify your transactions is how you get personal assurance that everyone on the network is following the rules. And if you don't do so yourself, the knowledge that others are using full nodes and relying on them is valuable. Someone just running 1000 nodes in a data center and not using them for anything does not do anything for this, it's adding network capacity without use. That doesn't mean that the full node count (or the reachable full node count even) are meaningless numbers. They are an indication of how hard it is (for various reasons) to run/use a full node, and thus provide feedback. But they are not the goal, just an indicator. -- Pieter