@Dusk_Foundation I judged DUSK from fork count first and that now feels like the wrong place to start.
A chain with zero visible forks can still be carrying weak network conditions underneath. What matters more is how consensus behaves as latency packet loss and validator downtime stack together.
Take 500 ms P95 latency. Alone maybe nothing breaks. Add 5% packet loss and 20% validator downtime and the question changes how much does the odds ratio of a fork rise versus normal conditions and how much consensus delay appears before forked rounds?
The same logic applies to the mempool. DUSK nodes do not need identical pending queues before agreeing on a final block. So one explorer showing 500 pending transactions tells me less than P90/P10 dispersion across 20 nodes or a pressure index built from median queue size × median pending age.
Some disagreement is normal. Distributed systems are messy that part is not alarming.
The real test for DUSK is whether disagreement stays local and temporary or leaks into inclusion odds consensus delay and finality.
I am still watching one uncomfortable point when stress factors combine does DUSK degrade gradually or does reliability fall off a cliff?
#dusk $DUSK
A chain with zero visible forks can still be carrying weak network conditions underneath. What matters more is how consensus behaves as latency packet loss and validator downtime stack together.
Take 500 ms P95 latency. Alone maybe nothing breaks. Add 5% packet loss and 20% validator downtime and the question changes how much does the odds ratio of a fork rise versus normal conditions and how much consensus delay appears before forked rounds?
The same logic applies to the mempool. DUSK nodes do not need identical pending queues before agreeing on a final block. So one explorer showing 500 pending transactions tells me less than P90/P10 dispersion across 20 nodes or a pressure index built from median queue size × median pending age.
Some disagreement is normal. Distributed systems are messy that part is not alarming.
The real test for DUSK is whether disagreement stays local and temporary or leaks into inclusion odds consensus delay and finality.
I am still watching one uncomfortable point when stress factors combine does DUSK degrade gradually or does reliability fall off a cliff?
#dusk $DUSK