Good morning everyone. I want to discuss a particularly overlooked issue during a bull market—the network layer.

Many people when talking about performance start to focus on TPS, block time, consensus speed, but only after running full nodes, experiencing on-chain "gas is fine but it's still stuck," or facing sudden instability in confirmation times do you realize a truth: slowness isn't necessarily the fault of consensus; it could very well be that messages aren't propagating through the network at all.

Decentralization requires more nodes, that's true. The problem is, if the underlying layer still uses the old gossip protocol—where nodes randomly forward messages to a few neighbors—the more nodes there are, the more exaggerated the duplicate messages become. Imagine what happens if a group blindly forwards the same meme repeatedly? Instead of speeding up information flow, the network gets overwhelmed by a "message storm," leading to latency spikes and unstable confirmation times.

This is one of the reasons I am optimistic about Dusk. It doesn't treat the network layer as an add-on but directly incorporates Kadcast as a core component, alongside consensus and execution environment, as the foundational infrastructure of DuskDS's P2P network. The official documentation is very honest: Kadcast uses a structured overlay network to guide message paths, abandoning the broadcast-style forwarding strategy of gossip. The result? Significantly reduced bandwidth usage, much better latency stability—precisely the network quality a financial system should have.