The Big Sort

In 1991, a film called JFK put more than a hundred and fifty actors on screen. Kevin Costner was billed first. Kevin Bacon was billed fourth. Far down the list, past more than a hundred other names, four actors appeared almost back to back: billing 117, then 123, 124, and 125. Six people, one movie, one shared credit that would follow every one of them for the rest of their lives.

Now run all six through a model of the entire Hollywood collaboration network, and watch them scatter. Costner and Bacon land together in the same small elite. The four deep-billed actors land in four different tiers, none of them elite, spread clear across the map. Same film. Same set. Six different places in the structure of Hollywood. The shared credit, it turns out, decides almost nothing. What decides is everything that happened to each of them outside that one movie.

Long exposure of a film set: one actor stands sharp in the center while everyone in motion blurs into ghosts — **The thesis, a decade early.** A long exposure I shot on a film set: Hold still and the frame keeps you. Keep moving and you blur out of the record. Set photography by the author.

Everyone argues about the top

Everyone argues about who sits on top of Hollywood. Almost nobody talks about the shape of the whole thing.

The previous essay was about the top. It took the old Six Degrees of Kevin Bacon game, rebuilt it as a weighted network where a lead role counts for more than a face in the crowd, and asked who is really at the center. The answer was not Bacon. Across the three measures we used, he ranked 141st, 181st, and 202nd. Samuel L. Jackson ranked first on all three.

That essay found the center. This one is about the rest: All 259,489 actors in the network, and the structure that sorts them. A map that only marks the capital city tells you nothing about the country.

Three ways to matter

A quick recap, because the rest depends on it. We built the network from 20,236 U.S.-linked feature films released between 1950 and 2025. Two actors are linked when they share a movie, and each link is weighted by how prominent their roles were, using billing order as the stand-in: The top-billed lead counts for far more than the last name in the crowd.

Then we measured each actor three ways. Reach: How few steps it takes to get from them to everyone else. Bridging: How often they sit on the shortest path between two other people. And prominent influence: A weighted score that rewards repeated, high-billed work alongside other well-connected actors. Three different ideas of what it means to matter.

Sorting a quarter-million careers

Ranking people one measure at a time gives you a leaderboard. It does not give you a structure. To see the structure, you have to stop sorting by any single number and start looking at whole profiles.

The method is called Gaussian mixture modeling. The intuition is simpler than the name. Give a computer every actor’s full profile, all three scores at once, and ask it to find the natural groupings: The clumps where careers look like each other. Not one column sorted high to low, but the whole shape of a career held up against everyone else’s. The model even hands back a probability for each actor, so we can see who sits comfortably inside a group and who straddles the line between two.

We asked it to find anywhere from two to twenty groups and let the data pick. It settled on seven. Seven natural realms of a Hollywood career.

The shape that falls out

The seven groups stack into three broad tiers.

At the top is the core. The model calls it the Hollywood Elite: 3,965 actors, about 1.5% of the network, with a median of 25 films and 31-year careers. These are the people who work constantly, in prominent roles, with other people who do the same. They are not central because they are famous. They are central because they keep showing up, together, for decades. Jackson is here. So are Robert De Niro, Morgan Freeman, Willem Dafoe, Bruce Willis, and Gary Oldman. So, it turns out, is Kevin Bacon: Not first, but firmly inside the core.

FIG 1 · PLACEHOLDERthe seven clusters projected onto two dimensions (PCA), the small Hollywood Elite pulling up and away from a dense cloud of peripheral clusters near the origin, a handful of names labeled.

Below the core sits the working middle. The group that reaches closest to the top is one the model calls the Aspiring Elite: 17,047 actors, about 6.5%, a median of 8 films and 18-year careers. They look like the core in miniature: Real filmographies, real connections, but not yet the concentrated run of repeated high-prominence work that defines the top. Below them are larger groups of shorter careers, a median of two films each, that fall between the extremes. Tally everyone with more than a single credit and it comes to only about 31% of the network.

Which leaves the other 69%.

FIG 2 · PLACEHOLDERrole-tier composition by cluster (lead, supporting, minor), the elite core dominating lead and supporting parts while the peripheral clusters are almost entirely minor roles.

The periphery

The largest fact in this whole dataset is also the quietest one: 69% of the actors in Hollywood’s network appear in exactly one film. Ever. Nearly 179,000 people, one credit each.

The model does not lump them together. It splits them into three groups by the kind of film that one credit came from: A big studio production or a small independent one, a slightly higher spot in the billing or a slightly lower one. But they share the defining trait. Their median bridging score is zero. Not low. Zero. You cannot be a bridge between two places if you only ever stood in one. A single film gives the network nothing to connect you to.

It would be easy to read that as failure. It is not. Every one of those nearly 179,000 people got cast, showed up, and did the work. They are in the network, permanently, their names sitting in the same graph as Costner and Bacon. The model sorts them low on connectivity. That is a fact about the shape of an industry, not a verdict on a person.

I knew that 69% before I could count it. My first movie job was set photography on a no-budget zombie feature in Atlanta called Ace the Zombie. The director had sold his car to fund it, everyone worked for free, and the whole thing ran on his gift for gathering people. Many of the actors I photographed on sets like that live in this periphery: One credit, everything they had, and a name that now sits in the same network as Costner and Bacon. And here is what the graph cannot tell you about them: They shined. Actors love a camera the way most people fear one. The structure files them at the edge. Through a viewfinder, nobody looked peripheral.

On a film set, an actress pulls a face beside the slate while the crew films — **Cause, then effect.** Two frames, seconds apart, on the set of a short film. She pulls the face; the whole set breaks. Set photography by the author.

One beat later, the whole set is laughing — **Cause, then effect.** Two frames, seconds apart, on the set of a short film. She pulls the face; the whole set breaks. Set photography by the author.

The Ace the Zombie slate held up in a used bookstore, cast and crew crowded between the shelves — **Ace the Zombie, roll 1, scene 14a, take 1.** Writer, director, and cinematographer crammed into a used bookstore. Everyone worked free. Set photography by the author.

FIG 3 · PLACEHOLDERcluster sizes drawn to scale (bar or treemap), the one-film groups that make up 69% of the network dwarfing the tiny elite core.

Back to the set

Now we can go back to JFK and finish the story.

Costner was billed first, the lead, a prominence score of 1.000. Across his career: 58 films over 43 years. Bacon was billed fourth, a prominence of 0.981, with 64 films over 47 years. The model puts both of them in the core, the Hollywood Elite. Nothing surprising there.

The surprise is the other four, billed 117, 123, 124, and 125. All four fall within eight billing slots of each other, buried in the same deep stretch of the same cast list. If a shared film made shared standing, they would land together. They do not. They land in four different groups:

Caroline Crosthwaite-Eyre (billed 117): JFK was her only film. The model files her with one group of single-credit actors.
Nathan Scott (billed 123): JFK was his only film too, a nearly identical career to Crosthwaite-Eyre’s. But a hair lower in the billing tips him across a boundary into a different single-credit group, the largest in all of Hollywood: 41.6% of the network on its own.
Jorge Fernandez (billed 124): Two films, not one. That second credit lifts him out of the single-film tiers entirely, into a short-career group.
Doug Jackson (billed 125): Billed dead last of the four, and yet the strongest career of them outside JFK. Three films across a 25-year span, his latest in 2013. The model places him highest of the four, up in the working middle with the Aspiring Elite.

FIG 4 · PLACEHOLDERthe six JFK cast members plotted onto the cluster map, Costner and Bacon together in the core, the four deep-billed actors scattered across four separate groups.

Read the billing order and these four look identical: Deep credits, same film, same afternoon. Read their whole careers and they could not be more different. The film they share tells you almost nothing about where any of them ends up. What they did everywhere else tells you everything.

That is the entire finding in one movie. A shared credit is not shared standing. Footprint decides.

Step back and the picture is not a ladder with Jackson on top and everyone queued below him. It is closer to a pyramid with almost nothing at the peak and almost everything at the base. A 1.5% core that works together for decades. A thin middle reaching for it. And a vast floor of people who passed through once.

Two honest limits. This is one snapshot of U.S.-linked feature films from 1950 to 2025, and billing order is only a proxy for how prominent a role really was. The model sees what those choices let it see, and no more.

A dashboard where you can look up any actor’s tier and scores is on the bench. The concept page is up now, and the full searchable version is what it grows into.

One thing this snapshot cannot show is motion. It is a single frame. A later essay will rebuild the network year by year, to watch actors climb, drift, and fall between these tiers over time.

For now, the sort stands. Six people walked onto one set in 1991. Thirty-five years later, the structure of Hollywood remembers them in six different places.