Out of band fee payments centralize mining

Out-of-band payment for transaction inclusion — accelerator services, private submission deals between large transactors and pools — is a centralizing force on mining.

For (strongest version as argued): Fee revenue paid inside the protocol reaches whichever miner finds the block; revenue paid outside it reaches only pools big enough to attract deals. If out-of-band inclusion becomes a significant fraction of miner income, a small pool structurally cannot compete for it, and hashers rationally point at the pools that can — the same consolidation loop as fee-sniping, arriving earlier. Evidence that this is already normalized: a material share of blocks include large-transactor transactions never observed in the public mempool, implying standing private relationships. In-band mechanisms (replace-by-fee, and fee-rate-based replacement) exist precisely so bidding stays open to every miner.

Against (strongest version as argued): Occasional out-of-band inclusion is benign — a convenience product for stuck transactions and large sequenced payloads — and its market share is small. Direct submission also has legitimate non-fee purposes (privacy, front-running protection, reliability for exchanges), and analogous infrastructure elsewhere (CDNs, certificate authorities) shows intermediated payment layers can coexist with a decentralized base. If in-band fee tools improve (better RBF semantics, package relay), demand for accelerators shrinks on its own; prohibition is neither possible nor needed.

Nuance: The disagreement is about share and trend, not mechanism — both sides accept that if out-of-band revenue became dominant, it would centralize pool economics. Data on private-inclusion share is thin and mostly inferred from non-mempool transactions in blocks; the specific figure cited on stage (roughly 29% of one exchange's mined transactions unseen by the network) is a claim from the event, not an independently verified statistic.

Common misstatements: "Accelerators break consensus" (they operate entirely outside consensus; that is the point of concern). "Private transaction submission is an attack" (it is permissionless behavior; the debated question is its equilibrium effect). "RBF and accelerators are equivalent" (RBF rebids in the open market; accelerators pay a specific pool privately).