Slashing and Rewarding in MCR Settlement Contracts

[l-monninger]

Summary

MCR implements an asynchronous PoS mechanism for agreeing on finalized state. At the time of initial writing, the formal modeling needed to ensure safety, liveness, and persistence of the rollup built on this mechanism are underway. There are, however, open discussion of relevant phenomena amongst developers should improve the quality and ease of planning and implementation.

The intent of this discussion is to cover various phenomena related to our PoS design, ultimately being married with implementation details from Movement Labs Research to generate tickets in this repository.

PoS and Provability

The Proof of Stake literature generally consists of mechanisms derive from or inspired by PBFT. PBFT provided the first practical Byzantine Fault Tolerant algorithm utilizing a quorum of $\frac{2}{3}$ of replicas to ensure agreement and validity conditions across view changes.

Proof of Stake extends this quorum to a notion of stake-based voting power the intent of which is broadly to incentivize honest behavior (or deter dishonest behavior) via penalties and rewards--slashing and rewarding.

Amongst the most well-known PoS algorithms are the following:

• Casper FFG
• HotStuff
• Tendermint
• Ouroboros

You may note, with the exception of the Ouroboros, these papers do not try to prove properties of stake-weighted voting w.r.t. a unified model of loss of the PoS heuristic. Mechanisms for ensuring safety, liveness, and persistence are generally addressed individually.

PoS Phenomena and Implementation

There are a number of phenomena that may affect the properties of PoS system. In the replies to this initial comment, I will lay as many as necessary out to generate discussion of how they may be handled.

Again, most of these phenomena require rigorous modeling and the details will be generated from research. However, having a fundamental understanding of what is implemented may prove critical to success in implementation and maintenance.