On 12/31/2017 10:01 AM, Steven Schear wrote:
I'm still hoping a new cryptocoin/fork implements a client-determined miner selection capability similar to what we suggested in that 2013 paper I've mentioned here.
I cannot immediately find your link to your paper. Obviously, at full scale we are always going to have immensely more clients than full peers, likely by a factor of hundreds of thousands, but we need to have enough peers, which means we need to reward peers for being peers, for providing the service of storing blockchain data, propagating transactions, verifying the blockchain, and making the data readily available, rather than for the current pointless bit crunching and waste of electricity employed by current mining. The power over the blockchain, and the revenues coming from transaction and storage fees, have to go to this large number of peers, rather than, as at present, mostly to four miners located in China. Also, at scale, we are going to have to shard, so that a peer is actually a pool of machines, each with a shard of the blockchain, perhaps with all the machines run by one person, perhaps run by a group of people who trust each other, each of whom runs one machine managing one shard of the blockchain. Rewards, and the decision as to which chain is final, has to go to weight of stake, but also to proof of service - to peers, who store and check the blockchain and make it available. All durable keys should live in client wallets, because they can be secured off the internet. So how do we implement weight of stake, since only peers are actually sufficiently well connected to actually participate in governance? To solve this problem, stakes are held by client wallets. Stakes that are in the clear get registered with a peer, the registration gets recorded in the blockchain, and the peer is gets influence, and to some extent rewards, proportional to the stake registered with it, conditional on the part it is doing to supply data storage, verification, and bandwidth.