Thank you @rycharde for pointing these posts. I didn't know about your proposal. And I'm glad to see more discussion about this topic.

IMHO the problem with your proposal is the computational cost because the blockchain needs to search votes in the last 24h in order to calculate the voting power for a single vote, it takes a lot of time. In my proposal, the blockchain only needs to search a user among 20 counters, then the computational cost is cheaper. I would like to know your opinion about this.