[Cryptography] A new, more efficient consensus protocol

[Vincent Strom]

Dear Peter,

Thanks for the kind words. Your suggestions is quite interesting but I think it would have some issues:

1. It assumes a universal clock i.e. synchronization in the network. Because everyone has to wait a fixed amount of time to receive h(x_i) and then again for a fixed amount of time to receive x_i.

2. It is possible some nodes, with poor connectivity, may not receive one of the above or both of the above in the specified period. As a result, their computation of the winner would be incorrect. And they will not accept the block that the majority nodes accept as valid. In other words, a complete consensus might not be achieved. This problem would probably worsen exponentially with the number of nodes in the network as you anticipated.

3. There is another, perhaps more serious problem. When a new node joins and is presented with a blockchain with multiple forks. It would not know which history to trust because nothing distinguishes one history from the other. This is because the winner does not get a permanent and universally verifiable proof of her winning which she could attach to her block (I don't know if such a proof exists in this protocol.).

The protocol presented in the paper does not need synchronization and the winner has a universally verifiable proof (the lottery ticket). When a new node joins the network, the longest valid chain will invariably be the agreed upon chain.

Regards,
Vincent
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://www.metzdowd.com/pipermail/cryptography/attachments/20210115/27c3c015/attachment.htm>