The Ethereum community has started arguing again. Core researchers proposed an EIP-8363 draft, claiming that the more you stake, the more you should gradually burn the portion of block rewards that would otherwise be given to validators. The DeFi big shots behind Aave—Kulechov and the gang—immediately went on the offensive. I’m not taking sides, but I can’t help feeling something: in consensus protocols, how to pay the people who maintain the network is one of the most troublesome problems.
In section 3.9 of the @Dusk whitepaper, Dusk thinks about this issue pretty carefully, because right out of the gate it admits a counterintuitive bug.
It calls it the “future block proposer incentive problem.” Since the lottery has already determined who will propose blocks for every round in this batch, the proposers scheduled later have an incentive to quietly cause the earlier rounds to fail—if the earlier blocks break down, the rewards fall into their laps.
Dusk’s solution is interesting: four steps in sequence. First, it pays voters separately, making honest voting more profitable than betting on yourself later to propose blocks. Next, if a proposer wants to claim the full reward, they have to collect all the known votes—forcing them not to hide votes. Even harsher: in each round, it removes the next round’s block proposer from the voting list, blocking the path for them to abstain and then “pick up the leftovers.” Finally, it clamps down by limiting the number of iterations per round, so fewer “future selves” can come in and stir things up.
The reward structure is also straightforward: newly minted DUSK plus transaction fees are distributed with 80% to block proposers, 10% to the voting committee, and #dusk kept for itself at 10%.
Here, $DUSK ’s role is the clearest—it’s both the carrot and the stick. Rewards come from newly minted DUSK, while punishment via slashing directly burns the staked DUSK of the misbehaving party.
I’ve given this design’s underlying human paradox a name: “the curse of queuing.” The clearer it is who stands behind you in line, the more the people behind want you to fall first. A good mechanism has never been about eliminating self-interest; it’s about steering everyone’s self-interest to work in the same direction. DYOR—don’t just take my word for it.
In section 3.9 of the @Dusk whitepaper, Dusk thinks about this issue pretty carefully, because right out of the gate it admits a counterintuitive bug.
It calls it the “future block proposer incentive problem.” Since the lottery has already determined who will propose blocks for every round in this batch, the proposers scheduled later have an incentive to quietly cause the earlier rounds to fail—if the earlier blocks break down, the rewards fall into their laps.
Dusk’s solution is interesting: four steps in sequence. First, it pays voters separately, making honest voting more profitable than betting on yourself later to propose blocks. Next, if a proposer wants to claim the full reward, they have to collect all the known votes—forcing them not to hide votes. Even harsher: in each round, it removes the next round’s block proposer from the voting list, blocking the path for them to abstain and then “pick up the leftovers.” Finally, it clamps down by limiting the number of iterations per round, so fewer “future selves” can come in and stir things up.
The reward structure is also straightforward: newly minted DUSK plus transaction fees are distributed with 80% to block proposers, 10% to the voting committee, and #dusk kept for itself at 10%.
Here, $DUSK ’s role is the clearest—it’s both the carrot and the stick. Rewards come from newly minted DUSK, while punishment via slashing directly burns the staked DUSK of the misbehaving party.
I’ve given this design’s underlying human paradox a name: “the curse of queuing.” The clearer it is who stands behind you in line, the more the people behind want you to fall first. A good mechanism has never been about eliminating self-interest; it’s about steering everyone’s self-interest to work in the same direction. DYOR—don’t just take my word for it.