DECENT-BRM: Decentralization through Block Reward Mechanisms
Varul Srivastava, Sujit Gujar
TL;DR
This work investigates centralization in Proof-of-Work blockchains driven by mining pools by modeling a two-player game between a new miner and the PoW system under different BRMs. It shows that memoryless BRMs (like Bitcoin) can decentralize only when switching costs vanish and cannot guarantee solo mining incentives, while retentive BRMs (e.g., Fruitchains) lower solo-risk yet fail to prevent pool formation. The authors introduce Decent-BRM, a retentive BRM where rewards are distributed among all prior contributors, which makes solo mining an equilibrium after a delay and yields strong $\rho$-Decentralization for any $\rho>1$. This BRM thus offers a principled mechanism to mitigate centralization and enhance PoW security, with practical implications for the design of future blockchain protocols.
Abstract
Proof-of-Work is a consensus algorithm where miners solve cryptographic puzzles to mine blocks and obtain a reward through some Block Reward Mechanism (BRM). PoW blockchain faces the problem of centralization due to the formation of mining pools, where miners mine blocks as a group and distribute rewards. The rationale is to reduce the risk (variance) in reward while obtaining the same expected block reward. In this work, we address the problem of centralization due to mining pools in PoW blockchain. We propose a two-player game between the new miner joining the system and the PoW blockchain system. We model the utility for the incoming miner as a combination of (i) expected block reward, (ii) risk, and (iii) cost of switching between different mining pools. With this utility structure, we analyze the equilibrium strategy of the incoming miner for different BRMs: (a) memoryless -- block reward is history independent (e.g., Bitcoin) (b) retentive: block reward is history-dependent (e.g., Fruitchains). For memoryless BRMs, we show that depending on the coefficient of switching cost $c$, the protocol is decentralized when $c = 0$ and centralized when $c > \underline{c}$. In addition, we show the impossibility of constructing a memoryless BRM where solo mining gives a higher payoff than forming/joining mining pools. While retentive BRM in Fruitchains reduces risk in solo mining, the equilibrium strategy for incoming miners is still to join mining pools, leading to centralization. We then propose our novel retentive BRM -- \textsf{Decent-BRM}. We show that under \textsf{Decent-BRM}, incoming miners obtain higher utility in solo mining than joining mining pools. Therefore, no mining pools are formed, and the Pow blockchain using \textsf{Decent-BRM} is decentralized.