NOT
CENTRALIZED,
BUT
DISTRIBUTED

A visual journey through decentralized systems — where every node matters and no single point controls the whole.

scroll to explore

THE PROBLEM
WITH ONE

A centralized system relies on a single authority — one server, one database, one point of failure. When it goes down, everything goes dark. It's efficient, but brittle.

Imagine a star topology: every spoke connects to a hub. Remove the hub, and every connection shatters simultaneously.

MANY VOICES,
ONE TRUTH

In a distributed system, nodes communicate to reach consensus — agreement without authority. No single node dictates the state. Instead, protocols like Raft, Paxos, and Byzantine fault tolerance allow the network to converge on truth.

Each blob-node here pulses independently, yet together they synchronize — a visual metaphor for eventual consistency.

GOSSIP
PROTOCOL

Data spreads through a distributed network like rumors in a crowd. Each node shares what it knows with random neighbors. Over time, every node learns every fact — no coordinator required.

Watch how a single piece of data (the amber pulse) propagates across the mesh, hopping from blob to blob through teal tendrils until the entire network is informed.

FAULT
TOLERANT

The beauty of distribution: remove any node, and the network adapts. Connections reroute. Data persists across replicas. The system breathes through failure.

Click any blob-node to "kill" it. Watch the network heal — tendrils reroute, remaining nodes absorb responsibility. This is resilience by design.

// no single point of failure
nodes.filter(n => n.alive).length >= quorum