October 15th, 2013 Huge Tech Event: Taming Galactus


On Tuesday, October 15, 2013, OLC attended Huge Tech Event: Taming Galactus, Marvel’s Tech Team on Their Fictional Universe featuring Peter Olson, VP of Web at Marvel.


The talk focused on technical innovation, innovation being mostly apps. Olson talked about how Marvel’s content and IP [intellectual property] can be represented in technology. The use of technology allows Marvel to represent their IP in new and innovative ways.

First, Marvel tells stories with characters. They are the No. 1 comic publishers in the world, No. 1 and No. 2 for all-time box office hits, they have some of the highest-rated animated and live action TV prospectives, and they have the No. 1 entertainment brand for boys consumer products.

Technology at Marvel is both old and new: they use a traditional LAMP stack, but the tech team at Marvel have been branching out to node.JS, Go, and other document databases. “We’re in Hadoop for big data,” Olson said.

As it stands today, Marvel has produced content for over 70 years. Where they stand now, Marvel has published 30,000 issues, 8,000 characters with 5,000 creators, 32 movies, 30+ TV shows, and over 100+ games. “In the Marvel universe, insane things happen—there are gods, mutants, monsters and awesome hair, but this all creates a number of challenges,” Olson said. “How can we create semantic data to work with in tech?” he asked. “We’re not using petabytes of data, but it’s still extremely challenging to work with.”

Olson outlined the first problem: the entity problem. “We have static entities. Let’s take a look at Hawkeye, for example. He has blonde hair, purple costume, and he’s an archer. We would take his attributes and compare them to other characters in the Marvel universe. This works fine until the entity starts to change or assume new identities, die and come back to life, go through new versions, join or leave teams, and even cross over media platforms,” he said. The old model used by Marvel didn’t work anymore. Instead, Marvel found out that the characters’ timelines all look something similar to a human timeline. “Now we take a lifestream like Hawkeye’s and apply it to 8,000 characters,” Olson said.

The second challenge is comic bibliography. “Comics are basically periodicals arranged into series,” Olson began. “In most periodicals, bibliographical structure is created to organize. In comics, they are a marketing and storytelling tool. Writers and editors play with bibliography to help tell a story.” He went on to describe that stories are structured sequentially and that readers experience them in a specific way, which creates a lot of problems:

  • How can we present granularity to some fans, and generic to others?
  • Lower barrier of entry?
  • Show character changes over time?

Olson stated that they began to look more into flexible and dynamic structures instead of static and fixed—their old model.

“Graphs are useful when they relationship between things is as or more important than the structure of things. It’s also useful for modeling multiple domains in a single space and graph databases are optimized to traverse these spaces,” Olson said. He also presented three terminologies that are used within graph models: nodes (thing), edges (relationship between two things), and hyperedge (relationship between more than two things).

“Teams have been a challenge to Marvel,” Olson conceded. “From Avengers, X-Men, and ohters, they are very fluid. But they are easy to model with a graph. We applied graph to aggregation and connected to variations of other characters. We can tier it into universes and concepts, locations and even costumes.”

“Putting bibliography in graphs show which precedes series and stories. We can create sequences of publication and story. No new domain models are needed. This makes the universe easier to understand and it provides constant, yet flexible modeling tools. Best of all, it’s most powerful when coupled with user behavior. Data is part of the comic experience!”