Distributed ledgers rely on consensus protocols to commit messages in an agreed order. In practice, this order is often chosen for profit, which raises fairness concerns. For example, in decentralized finance, nodes exploit transaction order to extract Maximal Extractable Value (MEV). This paper systematizes the mechanisms at the consensus layer that constrain message ordering in Byzantine fault tolerant systems. We decompose adversarial behavior into three capabilities that affect ordering and analyze each ordering family against them. We organize existing work into FIFO, random, and blind ordering, and compare the metadata and assumptions each requires. We also introduce a new formalization, the $f$-Robust Median Order, which exposes a gap in existing fairness definitions based on timestamps. Finally, we describe a reusable pattern for adding fair ordering to existing BFT consensus protocols. For the protocols we analyze, fair ordering costs 1.5 to 18 times the communication of vanilla consensus, and ordering computation varies by up to five orders of magnitude across families.

翻译：暂无翻译