Algorithm & Network Theory

The naive approach to finding a friend path is to fan out from one user until you bump into the other. That's painfully slow. Here's how searching from both ends at once makes the problem ~thousand times faster.

In 1929, a Hungarian writer proposed that any two people on Earth are connected by six or fewer social links. A century later, we put it to the test on Roblox's 70-million-user friend graph.

A handful of high-degree accounts shrink the entire Roblox graph to ~4-hop average distance. Network theory applied to Roblox, with examples and academic sources.

Patterns from real FriendPath searches: why almost everyone is ~4 hops from the CEO, why new accounts have shorter paths than you'd expect, and how UK-account hiding shapes the visible graph.