[GNUnet-developers] Ideas around proof-of-stake

[Jeff Burdges]

There are some interesting papers working on BFT problems for proof-of-stake 
blockchains now.  I’ll mostly refrain from linking those, but instead highlight 
some building blocks that themselves do not require stake, slashing, chains, 
etc.


There is a remarkably simple not-actually-a-blockchain protocol from UCL folk 
that ensures nodes broadcasts the same messages to all other nodes, but does 
use slashing somewhere:  https://arxiv.org/pdf/1809.01620.pdf
It's similar to the slashing free scheme 
https://dspace.mit.edu/bitstream/handle/1721.1/107927/MIT-CSAIL-TR-2017-004.pdf 
but with many senders.

Also, there is an older paper also by Silvio Micali that provides a more 
general agreement scheme:
https://people.csail.mit.edu/silvio/Selected%20Scientific%20Papers/Distributed%20Computation/BYZANTYNE%20AGREEMENT%20MADE%20TRIVIAL.pdf
   Both of these last two are components of Algorand, btw.


There are also interesting things being built with verifiable secret sharing 
(VSS) and publicly verifiable secret sharing (PVSS).   In brief the difference 
goes, there is extra protocol complexity VSS to manage some accusation phase 
since the shares’ correctness can only be verified by the recipients, but VSS 
agrees on scalars so the result can be used to produce many signatures or 
whatever faster, while PVSS agrees on curve points, like maybe a BLS signature, 
which each require more of the PVSS protocol.  There were bugs in previous VSS 
protocols, but they appear to all be fixed here:  
https://link.springer.com/content/pdf/10.1007%2F3-540-48910-X_21.pdf


Best,
Jeff

signature.asc
Description: Message signed with OpenPGP