Guerrilla Games on Designing ‘Horizon: Zero Dawn’

"It’s tough to read ‘your game is boring’ in capital letters."

About two years ago, less than one year before shipping Horizon Zero Dawn, Guerrilla Games held their first big playtest session with twenty players. They had been working on their new open world adventure for a few years, had all kinds of surveys prepared, and were ready to analyze the experiences and feedback the players gave them. The development team thought it was going to be a huge hit.

“It was absolutely brutal,” said Guerrilla Games Eric Boltjes in a crowded hall at the Game Developers Conference in San Francisco. “I truly felt panic, it’s tough to read ‘your game is boring’ in capital letters. Especially since we brought this idea from scratch with so little to go off of.”

Eric Boltjes has worked on nearly every game that Guerrilla Games has put out in his 15 years at the company including most of the Killzone series, he was also a lead designer on Horizon Zero Dawn. He spoke about how tough the feedback process can be, “If you’re lucky you get to make a prototype for one of your ideas, but feedback is going to be mostly negative no matter what,” he said. “It’s only 10 percent of what’s going to ship in the full game but it leads you to second guessing yourself constantly. That continues through the whole process of development, even after shipping.”

Back in 2001, Guerrilla Games asked everyone in company to come up with ideas they could develop into a game. Company members were given time at work to create a pitch, they ranged from a few sentences to full blown presentations with complex illustrations. They sifted through the pile and narrowed it down to what would eventually become Horizon Zero Dawn.

The pitch included 70 slides with four primary game design goals. Those included creating a majestic wilderness in a post apocalyptic world, inhabiting it with awe-inspiring machines, including exotic tribes and cultures, and having it all work with the freedom of an open world. None of these concepts were familiar to the eight person team that set out to prototype these different features. “Up until then we had only really made Killzone,” Boltjes said. “Which is beautiful in it's own way, but with a dystopian twist.”

Boltjes and his team were going from designing a linear shooter to a massive open world action game, which is an incredibly daunting challenge.

The core of Guerrilla’s design philosophy was starting with a wide scope, they decided to create prototypes for the different elements brought up in the pitch. In doing that they could actually play the ideas instead of leaving them on paper. If they worked it would be easier to expand on them and if they failed they could be scrapped.

One prototype they designed focused on what the combat would be like with giant robots, it included a giant low-poly dinosaur with blocks for armor that looked like legos. If you shot at it’s armor chunks of it would fall off, similar to how armor worked in the game. Another focused on the world itself, featuring an early version of Aloy riding a real horse through a city, showcasing what kind of environments they wanted to design. It was the first thing that started to flesh out what the robots purpose was, having them be part of daily human life.

“It takes a lot of time and effort to do this approach and a lot of the prototypes got thrown away after we invested so much time in them,” Boltjes said. “And in the end it didn’t feel like a cohesive experience, we had all these prototypes and people were wondering where the game was.”

Once the team had a good amount of the gameplay features fleshed out they started to create a proof of concept. Something that would be the adhesive to tie all the mechanics together. They ended up with a core loop. The player would explore the world, interact with it by talking to NPCs or fighting robots, gather parts of robots they killed or rewards from a completed quest, and then progress further into the game.

A little while later the team created a short trailer-like video that established the main character, an idea of what the world would look like, and provided context for all the systems and features they built. “It boosted the teams confidence,” Boltjes said. “Instead of having islands we had an actual game. And then we wondered why we didn’t do this earlier.”

Boltjes and the other lead designers did the same thing with the narrative. They formed a team of dedicated writers to sit down and go through the bits and pieces they had. All the ideas they had developed were scattered and eventually had to be thrown away in favor of a cohesive narrative built from the ground up.

Once context was established they had a unique world alongside gameplay elements they had been developing since the project started, but they had struck a rare balance between creating one concept and designing several individual ideas. “If you add context too soon you might prevent cool ideas from coming to life,” Boltjes added.

At this point the project was starting to feel whole. “We had most of the core mechanics, we had the narrative, and we had the world to put it all in,” Boltjes said. “This is where most of the big problems came in.” The team's lack of experience with the open world genre started to show and their roots in the linear design philosophy behind games like Killzone opened them up to more issues. They realized they had spent too much time on the beginning of the game, the tutorials they designed there did not reflect the majority of the gameplay that took place later in the story.

“Systems that we had emphasized ended up being less important since gameplay was shifted over time,” Boltjes said. “Throwing rocks became less important, but we spent an hour in the beginning teaching players how to do it.”

Other problems arose with things including the complicated nature of the games economy, the simplicity of human combat, and the lack of freedom in choosing how to fight robots. They were able to fix many of the issues by adding cover to human combat making it deeper and by creating a focus system that made the options available when fighting robots more clear.

But they were already too deep into production and couldn’t fix every issue, several playtest sessions, like that brutal first one, helped identify areas of improvement as they grew closer to launch in early in 2017. They were able to add some tools to help mitigate the issues they weren’t able to completely fix, but it ended up being a major learning experience.

“Issues that we couldn’t identify earlier because we couldn’t play the game started to show up,” Boltjes said. “At the end of production we found a lot of answers to the questions we were asking when we started but we realized that we didn’t ask all the right questions.”