[Cryptography] Proposal for a PoS blockchain
matbit at airmail.cc
matbit at airmail.cc
Mon May 25 07:15:56 EDT 2020
Thanks for your comments.
> 1) Can the chain be forked without this being noticed?
Since the proposed structure is a DAG and not link-list, so in any given
time for any machine we will have different graph state with different
leaves. Therefore theoretically there will be n forks in which n=number
of nodes. But it is not important, because we always working on the
blocks with the age older than 12 hours. Therefore the graph members
which are older than 12 hours must be same for all nodes.
> 2) Does the scheme provide a basis for distributing new currency?
Yes, but the distribution logic has nothing with network or computer
power, instead it based on member’s share which comes from external.
That is, members do some tasks, and claim their wage by stating the
hours that they dedicate to a particular task. Their claim will be
judged by polling.
> This effectively binds the two chains together and neither can defect
> without this being noticed within five steps.
The idea of using Bitcoin(or some other blockchains) as a proof of state
for weaker chains and vice versa is good, and we can bind TWO chain
together, but this also has a flaw. What if the cheater group starts to
do same binding beforehand? I mean imagine the cheaters start to bind
the corrupted chain parallel and put the Bitcoin hash in corrupted
chain, eventually after some days they attack to network and claim their
chain is the right one? How the nodes can determine which chain is true?
They see both chains are recorded on Bitcoin chain and both are linked
to Bitcoin blocks as well! The only solution is nodes stack. The nodes
have to trust the majority of network on every block’s creationTime.
That is, for every time span, the nodes will control “for this time
range, what blocks are signed by majority, and registered in Bitcoin
chain?”. So the nodes always choose the branch which signed by majority
and recorded in Bitcoin chain. Indeed I like the idea of recording
weaker chain (specially PoS-type chains) in stronger chains in one way.
This adds up an external-entity to chain which is useful. By this kind
of binding in one way we secure our chain will not re-organize graph, if
the majority (with today’s shares) wants to change the history of graph.
Your second idea of binding n different pairs is more interesting and as
you mentioned before “it is almost impossible to persuade all chains to
defect”. It could be implement as a plugin on top of our software. In
such a way we can add some auto-snapshoot-recorder that records our
chain’s status in any given time(e.g every 12 hours) on another
blockchain.
> So I conclude that proof of work/waste/stack etc. are unnecessary for
> 'security'.
I am not convinced, since the conflict resolution is a part of security.
I couldn’t find any solution to solve the conflict on chronicle spend
order except using node’s shares and majority’s edict.
If you have a solution for that, can you explain it in detail or refer
me some links please?
Regards
Hu
More information about the cryptography
mailing list