Hello to the Rootstock community!
We’ve been researching a vulnerability in Rootstock’s REMASC reward system that could degrade network performance. We propose the RSKIP-531 as a possible solution.
The Problem: Fee-Withholding Strategies
REMASC’s delayed reward mechanism creates an unintended incentive for strategic miners to withhold transaction fees. Here’s what happens:
-
Strategic Behavior: Miners exclude transactions with fees from blocks they mine but with no rewards to them, delaying the fees to future blocks where they can capture part of the reward
-
Perverse Incentives: Higher fees actually make transactions slower, not faster, as miners have more incentive to delay them
-
No Penalization: Excluded transactions simply roll over with no cost to the withholding miner
The Impact:
-
Reduces network efficiency and user experience
-
Creates unfair advantages for strategic miners
-
Undermines the reliability of fee-based transaction prioritization
Proposed Solution: Immediate Partial Fee Distribution
We are proposing a modification to REMASC that distributes a small fraction (e.g., 3%) of transaction fees immediately to the block producer, while preserving the remaining 97% for the delayed reward system.
This solution:
-
Eliminates the incentive to withhold fees
-
Maintains REMASC’s security benefits
-
Improves network performance and user experience
-
Creates alignment between miner incentives and network health
Important Note: Although this strategy introduces a clear theoretical vulnerability in REMASC’s implementation, it is not currently observed in practice. Implementing it requires changes to mining software, coordination with mempool management, and tracking of matured block eligibility — all of which add engineering complexity. Given the current fee levels, the expected benefit may not justify the effort. However, if network usage and fees increase in the future, the strategy may become more profitable and more likely to be exploited.
Proposed RSKIP: RSKIP: Fee withholding mitigation by patogallaiovlabs · Pull Request #531 · rsksmart/RSKIPs · GitHub