Existence of a TFM that is DSIC, MMIC, and OCA-proof

Determine whether there exists a transaction fee mechanism for blockchain transaction selection that simultaneously satisfies dominant-strategy incentive-compatibility (DSIC) for users, myopic miner incentive-compatibility (MMIC) for the miner, and off-chain agreement (OCA) collusion-proofness.

Background

Roughgarden proposed that “good” transaction fee mechanisms (TFMs) should satisfy user truthfulness (DSIC), miner incentive-compatibility (MMIC), and collusion resistance to off-chain agreements (OCA-proofness). This foundational question asks whether such a mechanism can exist.

This paper shows that side-contract-proofness (SCP) is strictly stronger than OCA-proofness and that the SCP-based impossibility results do not directly answer the OCA-proofness question. The authors then prove a deterministic impossibility (the only deterministic DSIC+MMIC+OCA-proof mechanism is trivial), and rule out several natural randomized classes (e.g., scale-invariant and CTPA), while also providing efficiency bounds for randomized mechanisms. They note that, despite these barriers, it may still be possible that a randomized mechanism achieves DSIC, MMIC, and OCA-proofness, leaving the general existence question unresolved in the randomized setting.