Brazzers Love Graph

Because love triangles have really boring structure.


The Data

I scraped all web pages of individual scenes on Brazzers website (at slow rate and text only) and for each scene I archived whatever structured data I could find.

In my database, Scene is the main entity and it has a bunch of attributes such as id, title, URL, release date, duration, number of views, likes and dislikes etc.

A Scene can also have relationships with other database entities:

  • Tag (many-to-many relationship), with 7.72 tags per scene on average.
  • Niche (each scene can be linked with only one niche), which refers to a niche website in Brazzers network.
  • Model (many-to-many relationship), consists only of model/actor name and their gender. On average there are 2.25 models per scene.
  • Activity (many-to-many relationship), an activity that can happen in a scene. The connection of Scene and Activity has attribute duration (how long some activity lasted in some scene). Duration of all Activities in a Scene don’t quite add up to duration of the Scene, because for some reason Brazzers didn’t count non-sexual activities like intros, chit-chat, and various manual labor activities (such as fixing the broken pipe, cleaning the pool or pizza delivery). Anyways, there are 13.84 distinct activities per scene on average.

The database consists of:

  • 7670 Scenes
  • 1974 Models (1642 female and 332 male)
  • 574 Tags
  • 178 Activities
  • 33 Niches

Now for some saucy pics… of graphs!

The Big Picture

Let’s start with a huge graph of all model relationships within the porn network. Nodes represent models and edges represent models being in the same porn scene.

So edges basically show us who had sex with whom. If there are more than two models in the scene, every model will be connected to every other model in that scene. I’m an old-fashioned kind of guy, so if they are having sex in the same place and at the same time, I assume they did it together. Everyone present is involved by default.

This means there is going to be a lot of same-sex relationships whenever there are more than two models involved.

I could have fixed this by checking if there are tags such as “lesbian” in the scene and if not I wouldn’t make connections between female models in that scene. Since Brazzers’ main target audience are straight men, there are no sex scenes between male models, so I wouldn’t have connected any two male models. But I settled for a broader definition of sexual act instead – if they did a scene together, they had sex, “no homo”.

Complete Brazzers Love Graph


Popular models are at the graph’s center. Since there are less male models and content is mostly straight, they are represented by nodes with the largest numbers of edges.

A popular male model and its connections

Nodes of popular female models build a web among each other and are almost all connected to popular male models nodes.

A network formed around two popular and two somewhat popular female models
A network built around a very popular female model (upper central node) and a very popular male model (lower central node)

Less popular models are further from the center, but still have ties to it. Finally, some nodes are floating on the fringes of the graph because the models either did a small number of scenes or did scenes with small number of different models (this graph doesn’t say anything about how many scenes a model did. It only shows relationships between models).

Here’s a more informative graph in which female and male models’ nodes are colored differently, red and blue respectively, and node size is determined by the number of connections:



Network with visible gender and “popularity” of each model

By “popularity” of models, I mean how many connections they have in the graph. I didn’t scrape any info on numbers of likes and dislikes of models.




I hope that by seeing this Stasi-style graph you gained a new perspective on adult industry. Nothing new is found really, but just by looking at it from this point, interesting questions start to emerge. One of the obvious ones would be about gender ratio.

I don’t know about you, but the first thing that came to my mind when I rendered the graph of all those sexual relationships was ♪ risk management ♫.

It’s a network of unprotected sex. Apparently condoms are a huge turn off for audiences, so rubber is only acceptable in S&M and bondage context (in form of masks, boots and full body suits).

And, of course, sexual relations of these people don’t end with this network. Most models are performing for other studios too, some do escort work and probably most of them have personal relationships with regular people. Some of them party or at least have in the past partied with Charlie Sheen.

The more connected a node is, the more valuable and vulnerable it is. It means bigger risk of infection, especially considering the network tends to grow and most popular models are expected to “connect” even more.

I wouldn’t be surprised if their development team was calculating risk-to-value ratios of each model, which could help managers make decisions like approving scenes or choose different prevention strategies.

Maybe I just need to watch Cronenberg’s Shivers yet again and get it out of my system.

I’ll conclude with a warning: I barely scratched the surface of this data set, so expect more pr0n here.