The paper "Collusion-Resilience in Transaction Fee Mechanism Design" by Hao Chung, Tim Roughgarden, and Elaine Shi explores the design of transaction fee mechanisms (TFMs) in blockchain protocols. The authors investigate the trade-offs between user incentive compatibility (UIC), miner incentive compatibility (MIC), and collusion-resilience properties. They build upon previous work by Roughgarden, who introduced the concepts of OCA-proofness (off-chain agreement) and $c$-side-contract-proofness ($c$-SCP) to ensure that TFM mechanisms are resilient against miner-user collusion. Roughgarden's OCA-proofness requires that there exists a "reference strategy" where the miner and users maximize their joint surplus, while $c$-SCP demands that no coalition of the miner and a subset of users can profit through strategic deviations. Chung and Shi previously showed that no TFM can simultaneously satisfy UIC and $c$-SCP for finite block sizes when there is contention between transactions. The main contribution of this paper is to prove that no TFM, even if randomized and truthful, can simultaneously satisfy UIC, MIC, and OCA-proofness when there is contention between transactions. This result resolves a key open question in Roughgarden's work. The authors also introduce a simpler notion called global SCP, which captures the idea that strategic users and miners cannot steal from the protocol and is more appropriate for UIC mechanisms. The paper provides a detailed proof roadmap and formal proofs, demonstrating that the impossibility holds even when the globally optimal strategy does not need to be individually rational. Additionally, the authors suggest several relaxations of the basic model that might allow the impossibility result to be circumvented, such as allowing non-truthful mechanisms or users to coordinate in bidding. The paper concludes with a discussion on the implications of these findings and open questions regarding the use of cryptography and Bayesian notions of incentive compatibility to address the limitations of TFM design.The paper "Collusion-Resilience in Transaction Fee Mechanism Design" by Hao Chung, Tim Roughgarden, and Elaine Shi explores the design of transaction fee mechanisms (TFMs) in blockchain protocols. The authors investigate the trade-offs between user incentive compatibility (UIC), miner incentive compatibility (MIC), and collusion-resilience properties. They build upon previous work by Roughgarden, who introduced the concepts of OCA-proofness (off-chain agreement) and $c$-side-contract-proofness ($c$-SCP) to ensure that TFM mechanisms are resilient against miner-user collusion. Roughgarden's OCA-proofness requires that there exists a "reference strategy" where the miner and users maximize their joint surplus, while $c$-SCP demands that no coalition of the miner and a subset of users can profit through strategic deviations. Chung and Shi previously showed that no TFM can simultaneously satisfy UIC and $c$-SCP for finite block sizes when there is contention between transactions. The main contribution of this paper is to prove that no TFM, even if randomized and truthful, can simultaneously satisfy UIC, MIC, and OCA-proofness when there is contention between transactions. This result resolves a key open question in Roughgarden's work. The authors also introduce a simpler notion called global SCP, which captures the idea that strategic users and miners cannot steal from the protocol and is more appropriate for UIC mechanisms. The paper provides a detailed proof roadmap and formal proofs, demonstrating that the impossibility holds even when the globally optimal strategy does not need to be individually rational. Additionally, the authors suggest several relaxations of the basic model that might allow the impossibility result to be circumvented, such as allowing non-truthful mechanisms or users to coordinate in bidding. The paper concludes with a discussion on the implications of these findings and open questions regarding the use of cryptography and Bayesian notions of incentive compatibility to address the limitations of TFM design.