#dusk $DUSK @Dusk

Read the peer-to-peer section of the whitepaper twice this week.
First pass, filed it as boring plumbing — how blocks get broadcast, not worth dwelling on. Second pass, I'd been wrong about what Kadcast is actually optimizing for.

I'd been treating 'efficient broadcast' and private broadcast as two separate wins bundled together for marketing convenience. They're not separate. One is a side effect of the other.
Kadcast doesn't flood messages to every neighboring node like gossip protocols do. It uses XOR distance from Kademlia's DHT to structure propagation — each node forwards only to peers at increasing distances, so the message cascades outward instead of blasting everywhere at once.
The whitepaper cites 25-50% less bandwidth than gossip protocols, and 10-30% fewer stale blocks in fast-block networks, because fewer redundant transmissions means fewer wasted validations on blocks that never make it into the chain.

Here's the part that clicked: that exact same structure is what obscures where a message actually came from.
Because a node only ever talks to a narrow set of peers at specific distances rather than broadcasting outward to everyone, there's no single moment where the origin is visible to the whole network at once.
The privacy isn't a separate feature bolted onto the routing. It's what efficient routing looks like when you're not allowed to flood.
That's what I'd missed — Kadcast isn't fast, and also private.
It's private because it's the specific kind of fast that doesn't require exposing the source to get there.
Two things I'm still turning over: does structured, distance-based routing create a different kind of attack surface than gossip does — predictable paths versus unpredictable flooding — even if it's harder to trace origin? And do the bandwidth and stale-block numbers in the whitepaper reflect Dusk's actual network in production, or are they still the general Kadcast research being cited rather than Dusk-specific measurements?
$DUSK