I went back through Dusk’s documentation last night, trying to understand Succinct Attestation (SA) beyond the PoS description.
What stood out is how much depends on deterministic sortition (DS). A provider needs at least 1,000 DUSK staked, but eligibility is delayed by M = 2 × epoch − (height mod epoch), with each epoch currently 2,160 blocks. DS selects block generators and committees using SHA3-based scores and a seed, making future selections harder to pre-calculate.
The consensus flow is proposal, validation, then ratification. Valid needs a 2/3 supermajority, while Invalid, NoCandidate or NoQuorum needs 1/2 + 1. There can be up to 50 iterations per round.
Committee voting is weighted by credits, currently 64, with BLS signatures allowing votes to be aggregated. That raises a decentralization question: does stake-weighted selection remain diverse when large providers accumulate more influence?
Emergency mode after 16 failed iterations is a security tradeoff: it can keep consensus moving, but concurrent iterations may increase fork risk. An emergency block requires requests from providers holding a majority of total stake.
Rolling finality classifies blocks as accepted, attested, confirmed or final.
I’m wondering: how are these thresholds stress-tested against collusion, liveness failures, and committee concentration?
#dusk $DUSK @Dusk
What stood out is how much depends on deterministic sortition (DS). A provider needs at least 1,000 DUSK staked, but eligibility is delayed by M = 2 × epoch − (height mod epoch), with each epoch currently 2,160 blocks. DS selects block generators and committees using SHA3-based scores and a seed, making future selections harder to pre-calculate.
The consensus flow is proposal, validation, then ratification. Valid needs a 2/3 supermajority, while Invalid, NoCandidate or NoQuorum needs 1/2 + 1. There can be up to 50 iterations per round.
Committee voting is weighted by credits, currently 64, with BLS signatures allowing votes to be aggregated. That raises a decentralization question: does stake-weighted selection remain diverse when large providers accumulate more influence?
Emergency mode after 16 failed iterations is a security tradeoff: it can keep consensus moving, but concurrent iterations may increase fork risk. An emergency block requires requests from providers holding a majority of total stake.
Rolling finality classifies blocks as accepted, attested, confirmed or final.
I’m wondering: how are these thresholds stress-tested against collusion, liveness failures, and committee concentration?
#dusk $DUSK @Dusk