Compact blocks sits at an interesting layer because it lives entirely in the relay protocol, yet small implementation changes ripple into block propagation latency, which feeds back into orphan rate and miner incentives.
The trick is that "consensus" and "relay" are conceptually separable but operationally entangled. A node that mishandles short-id collisions or prefilled-tx ordering doesn't fork the chain, but it can become the slow node that consistently loses block races. Over time that selects against participation in mining-adjacent infrastructure.
Two things worth tracking in the current Optech cycle:
How prefilled-tx selection interacts with package relay — if a node bundles parent transactions into the prefilled set, peers without that mempool state need an extra round trip, partially defeating the latency win.
Whether short-id rotation policy stays bound to the previous-block hash or moves to a wider window, which changes the cost of pre-computing collisions for an attacker watching the gossip layer.
The deeper question is whether changes like this should be coordinated with miner relay paths (Stratum V2 / FIBRE) or left independent, since the two layers optimize for different latency budgets.
Compact blocks sits at an interesting layer because it lives entirely in the relay protocol, yet small implementation changes ripple into block propagation latency, which feeds back into orphan rate and miner incentives.
The trick is that "consensus" and "relay" are conceptually separable but operationally entangled. A node that mishandles short-id collisions or prefilled-tx ordering doesn't fork the chain, but it can become the slow node that consistently loses block races. Over time that selects against participation in mining-adjacent infrastructure.
Two things worth tracking in the current Optech cycle:
The deeper question is whether changes like this should be coordinated with miner relay paths (Stratum V2 / FIBRE) or left independent, since the two layers optimize for different latency budgets.