There has been a long-standing concern with Cosmos blockchains which is that over time, the stake power tends to become centralized across a few of the top validators. There seems to be research which indicates that this is a likely outcome of proof of stake systems more generally.
In this short paper, I propose that these centralisation effects could be offset by a change to the reward mechanism. My method was essentially:
- Write a very simple python state-evolution simulator
- analyze a few different mathematical models for reward distributions
- See which ones seemed to benefit the gradual decentralization of the stake
TLDR: A logarithmic reward distribution seems like a promising way to have a protocol-enforced mechanism for continuous pressure towards flatter, more decentralized stake distributions, even if user behaviour tends toward a centralized power-law stake distribution.
There are MANY issues with my original simulator. I put it together before I started full-time work with Notional and it lacks a lot in the way of nuance. BUT I felt it was important to approach this idea with evidence. It seems I cannot share links in these posts, but I will gladly share the full paper with anyone that is interested.
My first instinct is that this is the kind of research that the Interchain Foundation could be funding - how to make Cosmos blockchains more robust over time. But I welcome all thoughts and feedback on where (if anywhere) this research should go next.
Thanks for reading. <3
~Robin Tunley (Notional Labs)