Here is a fact that sounds wrong: Humphrey Bogart and Margot Robbie are three steps apart. Bogart starred opposite Lauren Bacall in To Have and Have Not (1944). Bacall shared billing with Nicole Kidman in Dogville (2003). Kidman starred alongside Robbie in Bombshell (2019). Three films, and seventy-five years of movie history folds up like a road map.
That isn't a fluke of two well-chosen stars. In Flickle's co-star graph — built from public film datasets covering nearly 1.5 million actors — roughly nine in ten famous actor pairs are connected within three steps, and half sit at exactly three. Hollywood really is a small world. The interesting part is why the math all but guarantees it.
Strip cinema down to its skeleton and you get a network: every actor is a node, and an edge connects two actors whenever they were both billed in the top cast of the same film. Bogart–Bacall is an edge. So is Kidman–Robbie. A step is one hop along an edge, and the distance between two actors is the shortest chain of hops that joins them.
Network scientists study everything this way — power grids, neurons, friendships, the internet. Cinema happens to be one of the best-documented social networks in existence: more than a century of casts, meticulously recorded, with dates attached. It is the lab rat of network theory, and has been since the field's founding papers.
Why should paths be short at all? Start with the crudest possible model, the random graph studied by Paul Erdős and Alfréd Rényi in the late 1950s: take a million nodes and wire pairs together at random. If each actor has, say, 40 co-stars, then two hops reach roughly 40 × 40 = 1,600 actors, three hops reach 64,000, and four hops reach over 2.5 million — more than the entire network. Every step multiplies your reach, and exponential growth devours a million-node graph in four or five bites. In a purely random world, short paths aren't a curiosity; they're a mathematical inevitability.
But Hollywood isn't random, and the difference should matter. When a film bills ten actors together, it doesn't create one connection — it creates forty-five, every possible pairing at once. Network scientists call this clustering: your co-stars tend to be each other's co-stars. Casts are cliques, in the literal mathematical sense.
Clustering ought to be terrible for short paths. If your connections all know each other, each new hop mostly revisits people you could already reach. A world of tight, closed circles — the French New Wave regulars, the Bollywood ensemble system, the Judd Apatow repertory company — should produce long, meandering chains from one circle to another.
In 1998, Duncan Watts and Steven Strogatz published a short paper in Nature that resolved the tension — using, as one of their three test cases, the film actor network itself. Their insight: take a clustered, clique-heavy network and randomly rewire just a handful of edges into long-range shortcuts. Clustering barely drops. Path lengths collapse. You don't need many shortcuts, because a single edge leaping between distant neighborhoods shortens millions of paths at once. That combination — clumpy like a small town, traversable like a random graph — is what they named a small-world network.
Cinema manufactures those shortcuts constantly. An actor who works across eras — Christopher Plummer, billed alongside classic-Hollywood veterans and superhero-adjacent casts over a sixty-year career — is a shortcut. So is anyone who crosses industries: Gérard Depardieu bridging French and American cinema, or the stars who move between Bollywood, Hong Kong, and Hollywood. Each one is a wormhole stitched through the graph, and a few thousand of them are enough to make everywhere close to everywhere.
The other accelerant is inequality. Connectivity in real networks is never spread evenly; a small fraction of nodes hoards a huge share of the edges — the familiar 80/20 pattern. In Flickle's co-star graph, the prodigiously prolific Indian star Mithun Chakraborty tops the table with 344 famous co-stars; Michael Madsen has 291 and Depardieu 251, while a typical actor has a tiny fraction of that. Measured by average distance to every other famous actor, John Goodman is the best-connected star in our graph, at just 2.23 steps to anyone.
Whole films act as hubs too. An ensemble like Avengers: Endgame, The Grand Budapest Hotel, or The Prince of Egypt puts ten famous names in one top-billed cast — forty-five edges minted in a single production. Route your chain through a hub actor or a hub film and you rarely need more than a couple of moves.
Theory is nice; here are the measurements. Across famous actor pairs in Flickle's co-star graph:
"Six degrees of separation" turns out to be generous. For famous actors, three degrees is the norm, and finding a genuine six-degree pair is nearly impossible even when you're trying.
Film fans didn't discover this alone. Mathematicians have long traded Erdős numbers — your distance, through co-authored papers, from Paul Erdős himself, who wrote around 1,500 papers with more than 500 collaborators. Cinema's version arrived in 1994, when three Pennsylvania college students argued that every actor could be linked to Kevin Bacon in a few films, and the Bacon number was born. Both games work for the same reason: papers and casts are cliques, and prolific hubs plus cross-field shortcuts shrink the world. (A lucky few, like Natalie Portman, have a finite Erdős–Bacon number — a path in both graphs at once.)
Once you know the structure, you can feel it. Chains that look impossible — silent-era comedians to current A-listers — almost always resolve in three or four moves, usually through a hub you'd never have guessed. That intuition is exactly what our daily games exercise: Flickle asks you to connect two stars by naming the films between them, and Pathle hands you the chain and dares you to reconstruct it. The math says a short path is always there. Finding it is the fun part.
Published 12 July 2026 · Play the daily movie games