Casper CBC: Others
Proposal mechanism
@sproulM_
What is your current preferred block proposal mechanism? Or, how do you think the optimal round-robin proposing behaviour for CBC Casper should be implemented?
@VladZamfir
I like a deterministic protocol with queueing for validator slots determining the round robin order (if we decide to limit the number of slots), for simplicity/elegance
For something that other people will like, I am into Vitalik's private entropy solution
Formalization
Estimator$ \mathcal{E}type depends on state type$ \Sigmabut$ \Sigmadepends on$ \mathcal{E}
Hence, we should define the type of estimator with a supertype of $ \Sigma(like in our Isabelle formalization)
Removing equivocations or double-counting
Max-driven with removing / double counting equivocation = majority-driven with removing equivocation
In pBFT-based consensus, it is not so simple
In CBC, 1 observed equivocation contributes to finality as much as 1 new locked validator
Random thoughts
Does "verifiability of equivocation" weaken attackers (and improve some metrics)?
With this, CBC can slash more kinds of Byzantine behaviors, but not all.
Attackers can lie that they do not receive some messages (this can not be differentiated with network latency)
Without this, attackers can tell a lie about the ordering of his messages
"Future" state?
Actually, future states include in-flight messages due to the delay