I was tracing how Dusk's consensus actually reaches finality, and the "one block, one confirmation" framing started to feel incomplete.
Succinct Attestation works in rounds, and each round can run through multiple iterations if a committee fails to produce or validate a block in time. On the surface, that's just a retry mechanism fine, the network self corrects. But when I followed what a failed iteration actually costs, it's not just an abstract retry. Every failed iteration reshuffles the committee, resets the voting window, and pushes the timestamp of finality further out. That delay doesn't show up anywhere as a single number it's distributed across whoever is waiting on that block: a settlement counterparty, a bridge relay, an app checking confirmation status.
So the "finality time" people quote is really a best-case figure. The real cost is conditional it compounds with network conditions, validator responsiveness, and how often committees actually finish their job on the first try. Nobody carries that risk evenly; the person waiting on the transaction absorbs it, not the protocol.
I'm still not sure how that failure rate behaves once transaction volume and committee turnover both scale up together.
#dusk $DUSK @Dusk
Succinct Attestation works in rounds, and each round can run through multiple iterations if a committee fails to produce or validate a block in time. On the surface, that's just a retry mechanism fine, the network self corrects. But when I followed what a failed iteration actually costs, it's not just an abstract retry. Every failed iteration reshuffles the committee, resets the voting window, and pushes the timestamp of finality further out. That delay doesn't show up anywhere as a single number it's distributed across whoever is waiting on that block: a settlement counterparty, a bridge relay, an app checking confirmation status.
So the "finality time" people quote is really a best-case figure. The real cost is conditional it compounds with network conditions, validator responsiveness, and how often committees actually finish their job on the first try. Nobody carries that risk evenly; the person waiting on the transaction absorbs it, not the protocol.
I'm still not sure how that failure rate behaves once transaction volume and committee turnover both scale up together.
#dusk $DUSK @Dusk
