Six Degrees of Separation: Does It Apply to Roblox?

The original idea

Karinthy's story Chain-Links includes a famous bet: pick anyone on the planet, and the narrator claims he can connect them to himself through no more than five acquaintances. It was a thought experiment, not science. But the intuition turned out to be roughly correct, and the phrase "six degrees of separation" stuck around long enough to become a play, a movie, and a Wikipedia rabbit hole.

The Milgram experiment (1967)

The first real test came from Harvard psychologist Stanley Milgram. He mailed packets to ~300 random people in Nebraska and Kansas, asking them to forward the packet to a target person in Boston using only personal contacts — each recipient could pass it to one acquaintance who they thought might be socially closer to the target. Of the packets that completed the chain, the average number of hops was 5.2. Six degrees of separation, give or take.

Milgram's study had real methodological flaws (most packets never arrived) but it became the canonical reference. And it provided the first hard number to bolt onto the theory.

The Facebook experiment (2011, then 2016)

The modern equivalent is the analysis Facebook ran in 2011 (and re-ran in 2016) across all ~1.6 billion users. They computed the actual graph distance between every pair of users with at least one Facebook friend. The result: average distance 4.57, dropping to 3.57 by the 2016 re-analysis as the network densified. Karinthy's "six" turned out to be a slight overestimate for the online world. Facebook's writeup is here.

Why is it always such a small number?

The math is surprisingly intuitive. Imagine the average Roblox player has 50 public friends. Each of those friends also has ~50 friends. Two hops out, you've potentially reached 50 × 50 = 2,500 people. Three hops: 125,000. Four hops: 6.25 million. After five hops you've covered more than the entire active Roblox userbase.

Of course you don't actually reach everyone — there's massive overlap, because your friend's friends overlap heavily with your own friends (this is the property network scientists call clustering). But even with heavy clustering, the network shrinks fast. This is called the small-world phenomenon, and it appears in almost any social graph above a few thousand nodes.

Now: does Roblox actually fit the model?

FriendPath has handled tens of thousands of searches since it launched. From the data we can observe (we don't store paths long-term, but we keep aggregate distance stats), here's what shakes out for Roblox specifically:

• The median distance between any two active Roblox accounts is around 4 hops
• The mean sits a bit higher (~5) because of long tails — accounts that only friended real-life classmates and never expanded out
• Searches that fail to find any path are almost always because one of the accounts is private, banned, UK-restricted, or has zero friends — not because no path actually exists

In other words: Roblox is firmly inside small-world territory. The famous six degrees number is roughly correct as an upper bound for a typical pair of active players. Pairs of creators are closer than that, often 2–3 hops, because creators tend to friend a lot of other creators.

The David Baszucki effect

One thing that's special about Roblox compared to, say, Twitter or Facebook: the platform has a small number of "hub" accounts whose friend lists are extremely dense with active community members. @david.baszucki is the most extreme example — he's friended a huge chunk of the Roblox developer community, and through them, he's a 2-hop neighbor of basically every YouTuber, big creator, and major studio.

As a result, paths between two random Roblox players almost always end up routing through a hub account at some point. David Baszucki, KreekCraft, John Shedletsky, and a handful of others act as natural bridges between otherwise-disconnected friend clusters. This is why your typical "distance to David Baszucki" number ends up around 3–5 even though Roblox has 70+ million accounts.

You can check it yourself

If you want to see the small-world effect in action, FriendPath is the easiest way. Type your username, type someone famous, and watch the search expand outward from both ends. The algorithm runs a bidirectional breadth-first search so the answer comes back in seconds even on a network this size.