Harmony’s EPOS model is amazingly innovative, but I think we see where it falls short. The network is getting more centralized by the epoch.
Let me just say that I know this is going to be amazingly unpopular. Binance and whale validators are absolutely going to hate this proposal… but sometimes (like medicine) things that are good for you, taste BAD.
I propose we loosely borrow Harmony’s other amazingly innovative model, it’s token issuance model.
We have a rough idea of the amount of total network participants, counting validators and stakers. Simply come up with a % that we want to be validating. Whatever that % ends up being, we will be presented with two simple outcomes that can be balanced with two simple solutions:
Outcome 1) The network is centralized, the % of network participants validating is too low. Too many people are delegating.
Solution: Cap the amount of rewards given to each validator. Temporarily raise the % taxed so validators themselves aren’t harmed. This will push delegators into becoming validators.
Outcome 2) The network has too many validators. The % of network participants validating is too high (you’ll be able to tell as well, by the network quality). More people need to delegate.
Solution: Simply lower the tax until validators cannot afford to do be validators anymore, and they return to being delegators. This would have to be light handed and very gradual, as you don’t want to crash the network. I propose that lower performing validators get hit with the lower tax hammer a bit harder than the higher performing ones, but all should be hit.
Another solution to outcome 2 is to simply impose a network quality metric threshold that must be met or you lose your validating priveledge until the network starts to become centralized again.
As if I wasn’t unpopular enough for proposing this, I’m going to also say that what should happen immediately is the killing of validators getting multiple slots. One slot, one validator.
Multiple slots simply make the rich get richer far too easily. Sure they can go and spin up new validators anyways, but at least that would make them work for it. Let’s not make it too easy to centralize here.
Also as total network participants increases, we ABSOLUTELY should be aiming for more shards.