Splitting for Stability: a Modified Reward Scheme for Transaction-Fee-Only Regimes

arXiv:2611.13252 · cs.CR, cs.GT · Submitted 2026-11-09 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Security Radio. Generated commentary on the latest security and cryptography papers.

Nadia: We're looking at the paper: "Splitting for Stability".

Elias: Here's the gist of the paper. Splitting for Stability: a Modified Reward Scheme for Transaction-Fee-Only Regimes proposes a reward-splitting mechanism to mitigate instability caused by winner-take-all reward structures in fee-dominated regimes Because Bitcoin’s block subsidy declines over time, miner revenue will come to depend increasingly on transaction fees In fee-dominated regimes, winner-take-all reward structures can create incentives for undercutting and fork-based mining strategies, leading to instability and reduced transaction throughput To prevent this instability, we propose a reward-splitting mechanism in which the winning block producer receives an α-portion of the block reward, while the rest is distributed among other miners This incentivizes mining on previous blocks with higher rewards, while preserving the incentive to produce blocks with high rewards.

Nadia: Novel Reward Structure The proposed reward structure splits each block reward between the winning miner and a set of all other atomic miners This structure is designed to disincentivize forking by reducing the immediate reward earned by the winning miner, while subsidizing atomic miners who participate in the protocol.

Elias: Analysis of Profitability The analysis derives a local profitability condition for one-step Function-Fork deviations, showing how the winnerside reward, loser-side subsidy, and honest hash power jointly determine whether undercutting is profitable The analysis recovers the canonical 2/3 undercutting threshold of Carlsten et al.

6: under their stylized exponential model.

Paper summary: Nadia: So we're looking at this paper called "Splitting for Stability: a Modified Reward Scheme for Transaction-Fee-Only Regimes." Basically, it’s about Bitcoin in the future when the block subsidy drops, meaning miners rely almost entirely on transaction fees.

Elias: Right. And in those fee-dominated regimes, the traditional winner-take-all reward structure can actually encourage miners to undercut each other or even fork the chain because they want to maximize their immediate take and reduce competition.

Priya: So if you don't have a mechanism to stop that, you could see instability and less transaction throughput on the network.

Nadia: Exactly. The paper proposes a reward-splitting mechanism where the winner of a block gets some portion, say alpha, and the rest is distributed among all other participating miners. This is meant to encourage miners to keep mining on previous blocks with higher rewards while still giving them an incentive to produce blocks with high rewards themselves.

Elias: It’s a structural change in how the reward flows through the system. They evaluate this by looking at one-step function-fork deviations and deriving a local profitability condition that shows how the winner's share, the loser's subsidy, and honest hash power all work together to decide if undercutting is profitable.

Priya: I’m curious what kind of results they found when they compared this new structure against the standard model.

Nadia: The simulation findings show parameter regions where honest behavior dominates those fork-based deviations, meaning the splitting can reduce the level of honest participation needed to keep aggressive forking strategies in check compared to what you see in previous winner-take-all studies.

Elias: They have some key theoretical results too. Theorem four establishes a local undercutting threshold, saying that a one step deviation is only weakly profitable if the winner's share is less than or equal to the honest hash power divided by the loser's subsidy, which they call Bi sub i less than or equal to Ai.

Priya: That sounds complicated. What does that mean for someone who just uses Bitcoin and doesn't study proofs?

Nadia: It means that deviation is only profitable if there’s a lot of honest hash power compared to the winner's reward and the loser's subsidy, which is what they call Bi sub i less than or equal to Ai. However, proposition eight shows that just splitting by itself doesn't lower that local cutoff below the winner-take-all value.

Elias: And then theorem twelve shows that smoothing out this splitting actually improves the local cutoff in a symmetric environment, with the value Bi sub i alpha delta is equal to Bi sub i alpha zero plus m squared of m minus one times alpha delta over one minus delta.

Priya: So, what does this mean practically for network health? Does it really make things more stable?

Nadia: The strategy learning experiments suggest that splitting can substantially alter the strategy landscape seen in earlier work, and smoothing, specifically, improves both the local bound and the observed threshold, moving it from zero point three six to approximately zero point two three in environments without smoothing.

Paper summary: Elias: That's interesting because it suggests that even outside those regions where honest behavior dominates deviations, network health remains higher than in the standard case.

Priya: How does this translate into a real system? Can we actually implement this idea?

Nadia: They have a concrete Bitcoin instantiation describing how atomic miner designation and redistribution can be enforced using information already available in the blocks and coinbase transactions. This directly implements the rolling-window splitting and smoothing redistribution mechanisms they tested.

Elias: But they did have some limitations to mention, which is important for us to consider. One limitation is that they used a fixed one percent hashpower threshold to determine what an atomic miner is, which isn't how you'd actually implement that in a deployed system.

Priya: And what about the strategy learning experiments? Were they as general as we need them to be?

Nadia: Those experiments inherit the fixed strategy space studied by Carlsten et al., which means they might not cover every possible mining behavior out there.

Elias: Looking ahead, future research could test how sensitive these results are to different distributions of power and try to find parameter choices that simultaneously suppress undercutting, avoid encouraging selfish mining, and stay stable even when the total mining power is changing over time.

Priya: And what about the incentives for selfish miners? Does this splitting mechanism actually make it worse if there are multiple non-colluding selfish miners around?

Nadia: That’s an open question they plan to explore, specifically whether the reward mechanism continues to avoid worsening those selfish mining incentives in environments with multiple non-colluding selfish miners.

Elias: So, to wrap up this paper, the authors presented a reward-splitting scheme aimed at reducing the incentives for fork-based mining strategies in fee-only regimes. Analytically, they showed that deviation is profitable only when honest hash power is small relative to the winner's reward and loser's subsidy.

Priya: And experimentally, they found parameter regions where honest behavior dominates those fork-based deviations, showing that even outside those regions, network health stays higher than in the standard case.

Nadia: They also identified a simulation-derived default compliance threshold for suppressing function-fork behavior that beats both the sixty-six percent bound from Carlsten et al. and the thirty percent bound reported by Budinský et al. for FSRC, and they verified that orthodox selfish mining isn't made worse by their mechanism.

Elias: The paper calls this work "Splitting for Stability: a Modified Reward Scheme for Transaction-Fee-Only Regimes." It’s a proposal to handle the instability caused by winner-take-all rewards when transaction fees become the main source of miner revenue.

Priya: So, what does this mean for how we think about Bitcoin security in these fee environments?

Nadia: It suggests that changing the reward structure isn't just about tweaking numbers; it’s about creating a mechanism that actively shifts the incentives away from destabilizing behaviors like undercutting when miners are competing primarily on fees.

Conclusion: Nadia: So we're wrapping up this look at "Splitting for Stability: a Modified Reward Scheme for Transaction-Fee-Only Regimes."

Elias: Basically, the authors are proposing a way to fix instability in Bitcoin when it relies on transaction fees instead of block subsidies.

Priya: They’ve found that splitting the reward doesn't just look good on paper; their simulations show parameter regions where honest miners dominate those fork-based deviations.

Nadia: It means there are specific settings where you can actually keep the network stable even when people are trying to cheat the system by undercutting.

Elias: The theoretical results they found, like that local undercutting threshold, give us a mathematical way to see exactly how much incentive is needed to make deviation profitable.

Priya: But we have those caveats, you know? They used a fixed one percent hashpower threshold for defining what an atomic miner is, which isn't how real systems work.

Nadia: And the strategy learning experiments they ran were based on a specific set of mining behaviors from a previous study, so we gotta be careful about how general those findings are.

Elias: The big picture here is that this isn't just a tweak to the reward structure; it’s an attempt to build in resilience against aggressive, fee-driven cheating.

Priya: It moves the conversation from just looking at potential attacks to designing mechanisms that actively discourage those attacks from taking hold.

Nadia: So, what does this mean for how we think about Bitcoin security when the main revenue stream is fees?

Eric Brigham, Baigang Chen, Nicholas Hopper

University of Minnesota

cs.CR, cs.GT

Submitted: 2026-11-09

Updated: 2026-11-09

License: http://creativecommons.org/licenses/by-nc-nd/4.0/

Importance score: 80/100

The gist: Splitting for Stability: a Modified Reward Scheme for Transaction-Fee-Only Regimes proposes a reward-splitting mechanism to mitigate instability caused by winner-take-all reward structures in…

Key concepts

Winner-Take-All Reward Structure
In a standard system where one miner gets most of the reward, it can encourage others to try and undercut their work or fork the chain. This creates instability because miners are incentivized to cheat rather than cooperate, potentially reducing network throughput.
Reward Splitting Mechanism
This new scheme divides the block reward: a portion goes to the winning producer, and the rest is distributed among all other atomic miners. This change aims to make honest participation more profitable than aggressive undercutting or forking.
Local Undercutting Threshold
This is a mathematical condition that determines if it's profitable for a miner to try and produce a competing block immediately after another one. The paper analyzes how the split reward and subsidies affect this threshold, showing when undercutting becomes financially viable.

Terminology

Summary

Splitting for Stability: a Modified Reward Scheme for Transaction-Fee-Only Regimes proposes a reward-splitting mechanism to mitigate instability caused by winner-take-all reward structures in fee-dominated regimes Because Bitcoin’s block subsidy declines over time, miner revenue will come to depend increasingly on transaction fees In fee-dominated regimes, winner-take-all reward structures can create incentives for undercutting and fork-based mining strategies, leading to instability and reduced transaction throughput To prevent this instability, we propose a reward-splitting mechanism in which the winning block producer receives an α-portion of the block reward, while the rest is distributed among other miners This incentivizes mining on previous blocks with higher rewards, while preserving the incentive to produce blocks with high rewards. p. 0

Where the paper says it

In fee-dominated regimes, winner-take-all reward structures can create incentives for undercutting and fork-based mining strategies, leading to instability and reduced transaction throughput. To prevent this instability, we propose a reward-splitting mechanism in which the winning block producer receives an α-portion of the block reward, while the rest is distributed among other miners; this incentivizes mining on previous blocks with higher rewards, while preserving the incentive to produce blocks with high rewards.

Novel Reward Structure

The proposed reward structure splits each block reward between the winning miner and a set of all other atomic miners This structure is designed to disincentivize forking by reducing the immediate reward earned by the winning miner, while subsidizing atomic miners who participate in the protocol.

Analysis of Profitability

The analysis derives a local profitability condition for one-step Function-Fork deviations, showing how the winnerside reward, loser-side subsidy, and honest hash power jointly determine whether undercutting is profitable The analysis recovers the canonical 2/3 undercutting threshold of Carlsten et al. [6] under their stylized exponential model. p. 2

Where the paper says it

Our analysis first recovers the canonical 2/3 undercutting threshold of Carlsten et al. [6] under their stylized exponential model, then extends this style of calculation to our atomic-miner reward split.

Simulation Findings

Simulations show parameter regions in which honest behavior dominates fork-based deviations Suitable reward-splitting parameters can reduce the honest participation needed to suppress aggressive forking strategies relative to the winner-take-all structure studied in prior work.

Key Theoretical Results

Theorem 4 establishes a local undercutting threshold, stating that the one-step deviation is weakly profitable if and only if y Bi ≤ Ai Proposition 8 demonstrates that splitting alone cannot lower the local cutoff below the winner-take-all value. Theorem 12 shows that smoothing strictly improves the local cutoff in a symmetric environment, with Bi(α, δ) = Bi(α, 0) + m2(m − 1) · α δ / (1 − δ). p. 9

Where the paper says it

3 in their main configuration. The corollary shows that the qualitative effect is already present in the local model, and explains it: deferring reward mass monotonically lowers the one-step cutoff (e.

System Health Implications

The strategy-learning experiments suggest that splitting can substantially alter the strategy landscape observed by Carlsten et al. [6] Smoothing, by contrast, improves both the local bound and the observed threshold, moving it from 0. p. 8

Where the paper says it

Finally, (Gap − Rem) + is nonnegative and strictly positive exactly on {Gap > Rem}, so its expectation is positive iff pG > 0. 2 A General Threshold Theorem ▶ Theorem 4 (Local undercutting threshold).

36 to approximately 0.23 in unsmoothed environments.

Practical Implementation

A concrete Bitcoin instantiation describes how atomic miner designation and redistribution can be enforced using information already available in blocks and coinbase transactions. This construction directly implements the rolling-window splitting and smoothing redistribution mechanisms considered in our experiments. p. 5

Where the paper says it

For controlled threshold experiments, atomicity is assigned by the experimental setup so that we may study the effect of reward-splitting parameters without also varying which miners qualify as atomic. 2 Even and Proportional Redistribution Let A denote the set of atomic miners and let w ∈ A denote the winning miner of a canonical block.

Limitations of the Study

Limitations include using a fixed 1% hashpower threshold to determine atomicity, which cannot be directly implemented as stated in deployed systems. Furthermore, the strategy-learning experiments inherit the fixed strategy space studied by Carlsten et al. p. 0

Where the paper says it

Simulations show parameter regions in which honest behavior dominates fork-based deviations. In the cases where forking becomes dominant, the learned strategy is the least aggressive variant studied.

[6].

Future Research Directions

Future work could evaluate the sensitivity of our results to different distributions and identify parameter choices that simultaneously suppress undercutting, avoid encouraging selfish mining, and remain stable under dynamic mining power distributions. The study also suggests exploring whether the reward mechanism continues to avoid worsening selfish mining incentives in the presence of multiple non-colluding selfish miners. p. 0

Where the paper says it

Together, these results suggest that reward redistribution can mitigate fee-driven fork incentives without introducing stronger selfish mining incentives, providing a possible direction for incentive design in fee-era proof-of-work systems. 2012 ACM Subject Classification Theory of computation → Algorithmic game theory and mechanism arXiv:2611.

Conclusion Summary

In this work we presented a reward-splitting scheme with the goal of reducing incentives for fork-based mining strategies in PoW transaction-fee-only regimes. Analytically, we showed that deviation is profitable only when honest hash power is small relative to the winner-side reward and loser-side subsidy. p. 0

Where the paper says it

We evaluate the proposed reward structure through analysis and simulation. Our analysis derives a local profitability condition for one-step Function-Fork deviations, showing how the winnerside reward, loser-side subsidy, and honest hash power jointly determine whether undercutting is profitable.

Experimentally, we identified parameter regions where Petty Honest behavior dominates fork-based deviations, and showed that even outside these regions network health remains higher than in the standard case. p. 0
Where the paper says it

Our analysis derives a local profitability condition for one-step Function-Fork deviations, showing how the winnerside reward, loser-side subsidy, and honest hash power jointly determine whether undercutting is profitable. Simulations show parameter regions in which honest behavior dominates fork-based deviations.

Our simulation-derived Default Compliance threshold for suppression of function-fork behavior significantly improves on both the 66% bound identified by Carlsten et al. p. 2
Where the paper says it

Contribution 2: Analysis We analyze the incentive effects of the proposed reward structure by comparing honest publication against a one-step Function-Fork deviation. Our analysis first recovers the canonical 2/3 undercutting threshold of Carlsten et al.

[6] and the 30% bound reported by Budinský et al. p. 4
Where the paper says it

We analyze how this allocation affects undercutting profitability and evaluate the resulting strategy choices of heterogeneous miners, including the effects of smoothing. The work most closely related to ours is the fee-redistribution smart contract framework of Budinský et al.

[5] for FSRC. Since we also verified that orthodox selfish mining is not exacerbated by our mechanism, it appears to effectively mitigate undercutting incentives in the fee-only regime. p. 0
Where the paper says it

Together, these results suggest that reward redistribution can mitigate fee-driven fork incentives without introducing stronger selfish mining incentives, providing a possible direction for incentive design in fee-era proof-of-work systems. 2012 ACM Subject Classification Theory of computation → Algorithmic game theory and mechanism arXiv:2611.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Reward redistribution can mitigate fee-driven fork incentives without introducing stronger selfish mining incentives by splitting each block reward between a winning miner and a set of eligible non-winning atomic miners. This mechanism incentivizes mining on previous blocks with higher rewards while preserving the incentive to produce blocks with high rewards.

  2. The AI system can be trained to adopt strategies that are less aggressive when reward splitting parameters are set in regions where honest behavior dominates fork-based deviations, as suggested by the analysis and simulations. This is achieved by tuning the winner-side reward fraction alpha to favor stable outcomes over aggressive undercutting strategies.

  3. The AI system can be designed to maintain higher canonical fee throughput and lower stale block rates when it learns a strategy that is Function-Fork 1, which is the least aggressive fork-capable variant studied. This suggests that even less aggressive forking strategies do not necessarily result in substantial transaction-fee stagnation.

  4. The AI system's required level of Default-Compliant participation can be significantly reduced by utilizing the proposed reward mechanism, as simulations show it can lower the observed threshold from 66% to approximately 23% in unsmoothed environments. This demonstrates that immediate redistribution among eligible atomic miners is beneficial on its own.

  5. The AI system can utilize a rolling-window redistribution mechanism enforced through block validation checks, where eligibility for subsidies is determined by payment script hashes appearing in preceding blocks within a defined window. This allows for the enforcement of atomicity and smoothing without requiring an auxiliary smart contract system.

Abstract

As Bitcoin's block subsidy declines over time, miner revenue will come to depend increasingly on transaction fees. In fee-dominated regimes, winner-take-all reward structures can create incentives for undercutting and fork-based mining strategies, leading to instability and reduced transaction throughput. To prevent this instability, we propose a reward-splitting mechanism in which the winning block producer receives an α-portion of the block reward, while the rest is distributed among other miners; this incentivizes mining on previous blocks with higher rewards, while preserving the incentive to produce blocks with high rewards. We evaluate the proposed reward structure through analysis and simulation. Our analysis derives a local profitability condition for one-step Function-Fork deviations, showing how the winner-side reward, loser-side subsidy, and honest hash power jointly determine whether undercutting is profitable. Simulations show parameter regions in which honest behavior dominates fork-based deviations. In the cases where forking becomes dominant, the learned strategy is the least aggressive variant studied. The analysis and simulations suggest that suitable reward-splitting parameters can reduce the honest participation needed to suppress aggressive forking strategies relative to the winner-take-all structure studied in prior work. Finally, we show that the modified reward scheme does not appear to worsen selfish mining incentives. Together, these results suggest that reward redistribution can mitigate fee-driven fork incentives without introducing stronger selfish mining incentives, providing a possible direction for incentive design in fee-era proof-of-work systems.

Sources

Related papers