Atlas Reactor: How an MMO studio set out to create the next great competitive game

It started with a pitch jotted down in stream-of-consciousness on a piece of paper. It wasn't a very good pitch.

"It was absolutely horrible," says Atlas Reactor executive producer Peter Ju. "I'm terrible at writing."

William Cook, the game's lead designer, is not much kinder. "It wasn't a pitch. It wasn't presented clearly as a thing. Pete even wrote it after the fact. It was a conversation he was having with another engineer and a designer, and he wrote that down."

You can also listen to this feature below, thanks to Polygon Longform. Links to subscribe in iTunes, in your podcast player of choice or to download the episode as an MP3 are also available below.

While the two lead developers on Trion Worlds' upcoming turn-based competitive game sound hard on each other, they've got smiles on their faces and laughter on their lips as they recall Atlas Reactor's humble origins. Nobody could have guessed, least of all these two, that the lackluster slip of paper full of excited but unfocused ideas held the key to Trion's next big game and a major shift in perspective for the company.

The creation of Trion Labs

In November 2012, Trion Worlds had just released its first expansion pack for its first game, the well-received massively multiplayer role-playing game Rift. With a year and a half of live service and a single expansion under its belt, the studio was beginning to think about what it should do next. It had no intention of not continuing to support Rift, but Scott Hartsman — then the company's general manager — wanted to figure out how Trion could continue to grow.

"We made a very intentional decision to pursue stuff that's outside of what people traditionally associate with Trion," Hartsman says. Today, after a short time away from Trion in 2013, Hartsman is the company's CEO. But even in 2012, he was pushing for the company to do things differently.

Atlas Reactor

"All of our projects previous to that were super high-concepted and then fundraised around, and then teams were built around those concepts," he says. "We very intentionally wanted to turn that whole thing on its head this time."

The idea that Hartsman and his team came up with is something they called Trion Labs. Rather than higher-ups at the company creating a big budget idea and looking for funding, they wanted employees at all levels to pitch ideas.

The first game to come out of this process was Trove, an MMO with a Minecraft-esque voxel-based visual style. While that title launched this past summer, Hartsman stresses that the process of going through Trion Labs was heavily iterative for it.

"Trove was a number of games before it was the game you see now," he says. "It was originally super sandboxy. After that, it was actually a voxel shooter. I'm not kidding. It wasn't until they stepped back and tried an MMO adventure in a voxel world that we were like, 'Holy shit, this is fun.'"

Trove was also unique, because a couple of Trion employees had already been crafting it in their free time by the time the idea of Trion Labs came into being. It skipped the pitch process entirely. Pitching was an important part of how Hartsman envisioned Trion Labs, though.

"The fewer words you use to communicate your vision for a game, the clearer it's going to be," he says. "The more accurate you can make that small set of words, the more similar everybody's own internal map is for what that project looks like and what our real priorities are."

Hartsman is heavily critical of "the old days," when studios drafted up 100-page design documents. He calls them "documents that people write once and nobody reads" and contrasts it with one- or two-page pitches that can be read in minutes and tweaked if they aren't working.

If Trove was Trion Labs' first success story, Atlas Reactor would be the project to demonstrate Hartsman's ideal for how that process of pitching and editing should work.

Atlas Reactor

Problem solvers

Peter Ju is an engineer by trade, and his approach to game development is one of a problem solver. He looks for ideas that present interesting conundrums, concepts he can puzzle over until he cracks them. Atlas Reactor was born from exactly one of these problems.

"I had this idea," Ju says. "I like a lot of competitive games. For a lot of them, I felt like the ramp-up part takes a long time. It takes a long time to get to what people say is the fun part of the game."

As Ju toyed with this idea, he was simultaneously thinking about turn-based games and why they so rarely went into the realm of player-versus-player competitive modes.

"Look at board games," Cook says. "There's a lot of competitive board games that are all turn-based. But you don't see that as successful in video games."

Atlas Reactor

Ju saw an intriguing dilemma in this. People at Trion, including himself and Cook and Hartsman, loved turn-based games, but he knew there had to be a reason that the genre didn't translate to PvP. He wanted to decipher that reason and then figure out a way past it. He calls it his "greedy, personal reason for pursuing this game."

So, after a conversation discussing all of this with a fellow engineer, Ju wrote his initial pitch — the one that he admits to being bad. Previously, Ju had been working on a technology exploration. He had created what Hartsman describes as "an action-physics 3D platformer with fighter mechanics." That hadn't turned into anything meaningful, but Ju proposed taking that technology and turning it into his weird new idea for a turn-based competitive multiplayer game.

"The earliest pitches for Atlas Reactor didn't really get many people excited," Hartsman admits. But the team shares the pitches in such a way that anyone in the studio can read them and provide feedback. Elsewhere in the company, Cook was paying attention to Ju's pitch and saw a flicker of potential in it.

"Pete didn't have anything to present to other people, but he had this conversation piece that was very interesting," Cook says. "Pete's idea was really compelling to me, because we're all big MOBA fanatics. XCOM had just made a splash. I saw those two things within that idea, so I basically went to him and said, ‘Hey, Pete, I like your idea. I'm not glomming onto it or anything, but I'm going to write you a new pitch.'"

Cook wrote a new one-page pitch that merged the big concepts Ju was wrestling with into two easier-to-digest directives for the project: visceral, XCOM-inspired turn-based combat, and team-based MOBA gameplay. "They were two ideas that people could immediately latch onto at Trion, because they were playing and enjoying those games," Cook says. "And I think it was getting at the same things Pete wanted to experiment with."

"There's a lot of competitive board games that are all turn-based. But you don't see that as successful in video games."

As an engineer, and as someone increasingly excited about the ideas floating around in this game pitch, Ju didn't want to wait to start experimenting. He created a playable demo to show the company four days after Cook's new pitch was turned in.

"I think about games in terms of how I can build them efficiently," Ju says. "I knew I could prototype this quickly."

The earliest demo for Atlas Reactor played on a blank grid. Cylinders represented characters, and they had no real abilities beyond shooting at one another and moving across the grid. But even in this earliest form, people in the studio were beginning to see promise.

"I played that prototype," Hartsman says. "I saw the fun immediately. There was something there. It was fun enough to where even in its ugly stage, people were asking when the next playtest was."

For Hartsman, seeing that people wanted to play more of it was the first sign that this might be the next project worth pursuing out of Trion Labs.

Atlas Reactor

Survival of the fittest

From the start of the Trion Labs idea, Hartsman knew that not everyone would be happy.

"There will invariably be some amount of hurt feelings when somebody's puppy doesn't make it," he says. A dog owner himself, Hartsman chooses that analogy carefully, as something that recognizes the weight of those hurt feelings. "But at the end of the day, we all understand that we all want the greatest success. When the company succeeds, everybody wins."

For Hartsman, the greatest key to avoiding bruised egos any more than necessary is to be transparent and honest to a fault. From the moment the company opened up pitches to the studio at large, he made it clear that very few would make it. He expected pitches in the double digits, and obviously, as a company that had made a single game in five years, it would not be expanding to that many projects at once.

Atlas Reactor

Trion employees ended up producing around 30 pitches. Many of them were culled immediately for being too ambitious in scale to make sense for the company.

"As a studio who has built big MMOs, a lot of those ideas sounded good," says Cook. "But we were about to launch Rift's Storm Legion expansion and support that product. The scale these pitches were asking for would take another studio of that size to create them. It was stuff that would require an MMO-sized team."

Other ideas were intriguing on paper but too weird to make it much farther. Ju describes a pitch from one developer that was a fishing-based MOBA. Senior producer James Karras came up with an idea that blended the arcade action game Smash TV with the hardcore action RPG Dark Souls.

"It kinda made sense in context," Karras promises, laughing.

These pitches lived or died not based off the decisions of Hartsman or other higher-ups at Trion, but based on reactions and feedback from the studio at large. While soliciting pitches from everyone at the studio, Hartsman also wanted to give them the power to help determine which projects made it to the next step.

The studio shares every pitch, prototype and playtest at Trion with the team at large via opt-in email groups. Any employees who want to participate and make their voices heard are encouraged to join those groups and give feedback freely. It was within those email groups that Hartsman first noticed a groundswell of support for the Atlas Reactor prototype, and an increasing number of employees emailing just to ask, "When's the next playtest?"

From there, Hartsman conferred with Trion's general manager to determine that Atlas Reactor was ready to move onto the next step: having a tiny team of two to three employees devoted to working on it and iterating at fast speeds.

"Think about it like a representative democracy," Hartsman says of the process of choosing which pitches actually move forward. "The general manager and I are acting like their representatives, but there's obviously some practical concerns that affect it also, like whether it's a game that fits into budgetary and scope and timeline that we have. Do we see it fitting in that container? Can we get enough people excited about working on this game? What do we think its chances are of succeeding and being completed at the right quality?"

For Hartsman, Atlas Reactor checked a lot of boxes and made sense to move forward with in large part because of the power duo of Ju and Cook behind it. He describes Ju as "one of the smartest human beings I've worked with in my entire life." Cook, he says, is "equally smart but in the opposite direction" — a philosophy major focused on game design.

"Literally diametric opposites!" Hartsman says with a grin. "These two guys just need time. They need to work. So we all decided to give them the space."

Atlas Reactor

Big problems and impossible solutions

Ju and Cook would certainly need that space to work out solutions to the many problems Atlas Reactor was immediately presenting them with. The first issue to solve: How could they make a turn-based game fast and exciting?

"I wanted the decision time to be in the order of seconds," Ju says. "Twitch-reaction games are sub-seconds, and chess and other turn-based games are infinite time. I wanted something where you had to make decisions relatively quickly. We wanted to make it snappy."

The first attempt at speeding things up came with the decision to have every player on the same team lock in their choices at the same time, rather than one by one. One team would make choices, those choices would play out, and then the next team would make choices, and so on. It was already moving faster than the average turn-based game, but Cook was bothered that players were still waiting more than they were playing.

Atlas Reactor

Another problem the developers encountered early in development was the idea of focus fire.

"The first person to stray too far forward allowed the other team to spend all of their actions just shooting at them," Cook explains. "That's how you lost the game."

Cook and Ju experimented with a number of ways to fix this. They tried adding in overwatch, a popular feature from XCOM where players could be set to automatically shoot the moment an enemy came into their cone of vision, but this encouraged players to not move at all. They toyed with giving players two "action points" per turn, so they could choose to shoot then move, or if they preferred, to just shoot twice. As it turned out, everyone always chose to just shoot twice.

It was at some point in this early testing phase that Cook first made a light-hearted suggestion.

"You know, one thing that could solve this is if everybody just went at the same time," he recalls telling Ju one afternoon.

The response was one of exasperation — that idea seemed too absurd, too impossible to figure out the logistics of. Ju says he found the idea "confusing as hell" and refused to even consider it.

"It was a joke," Cook says sheepishly. "I didn't even believe it was a good idea when I first said it."

As they continued wrestling with the problem of speeding up the game, Ju and Cook decided that the best way to give players incentive to not just focus fire was to provide them with interesting and unique abilities to use. At this point in Atlas Reactor's life, most of the abilities were tab-targeted — that is, players chose which character they wanted to attack, and the ability homed in on that player. But then Cook came up with the idea of a laser ability.

"The laser was set up so that if two characters are in a line, I could shoot both of them," Cook explains. "That was our first template. But then we realized there was no way to get out of the way of the laser. So we made a dodge ability. But dodge didn't matter if the other team could just tab-target or got two actions in a row. You had to dodge during the same turn they were attacking. So how do you build that?"

"I didn't even believe it was a good idea when I first said it."

They had made a conscious chose to avoid "reactive dodging" mechanics. While inspired by XCOM, Cook and Ju wanted to remove that game's randomness, where abilities could miss because they only had a percentage chance to hit, or because a character had a percentage chance to dodge. They wanted it to be based in skill, not luck.

Cook recalls Atlas Reactor's next big milestone in stark detail, with an equal mix of relief and dread. It was now a couple of months into 2013. They had been wrestling with the game and trying to pin it down for two or three months. It was getting late in the day; the parking lot outside the conference room where Cook was working with another designer was pitch black.

Ju popped his head into the door of the conference room and dropped a bomb: "You know what? We should do simultaneous turns."

Then he closed the door and left the building. Though he had joked about exactly such a change previously, Cook was floored.

"This changes the entire structure of the game," he says. "Suddenly we have to figure out how all these abilities we'd already started designing work with simultaneous turns."

It was an imposing amount of work ahead of the still-tiny team. It was also, it would turn out, exactly what Atlas Reactor needed.

Atlas Reactor

Sharing the load

It was around this point that the size of Atlas Reactor's team was beginning to bother Cook.

"Peter was gone for a while," he says, remembering a point where Ju had to be pulled off of Atlas Reactor to help with another, more pressing project. "It was just two of us, me and another designer, for a number of weeks. Our animators were all part-time. They were helping out with other projects constantly."

Karras estimates that the Atlas Reactor crew was down to just two full-time employees for a span of around six months. Though frustrating to Cook, this was actually by design.

As Trion has shifted from a single-product developer into a studio that houses a number of ongoing games with different levels of needed support and investment, Hartsman has worked to transform the studio's culture accordingly.

"When we're interviewing people, we tend to screen for people who genuinely have a desire to be helpful," Hartsman says. "That's who I am at my core. All of us just want to be useful. When you get good at doing a thing, it turns out that sometimes another game needs that thing."

Atlas Reactor

Hartsman says that this idea of people in the company shifting around to different projects as needed even extends to his role as CEO. When the company launched ArcheAge last year, he joined that team's customer service group in writing reports. He had experience working with data and wanted to pitch in.

Hartsman admits that some people are very specialized and brought in to work in specific roles on specific projects. But that's becoming increasingly rare. He estimates that around 50 percent of the studio has contributed to at least two games. The studio also provides freedom for employees to check out other projects that they're not assigned to in their free time.

"What we find is that the best, most creative people always want to be building something," Hartsman says. "Exercising a different set of mental muscles is a great way to keep yourself fresh and learn more. We have people who've launched indie games on iOS, because they're not competing with our products, so that's cool. Do your thing. We also have people who've done cool internal projects, like adding VR to our games. Everybody is allowed to learn everything about everybody else's games. Take all the source code. Take all the assets. Screw around with it. Remix things. Learn stuff. It's fun."

It's this liberal attitude toward letting employees find and contribute to the projects they're passionate for that led to the existence of Trion Labs as a concept, but according to Hartsman it goes deeper than that. For him, it's about fighting a negative cycle plaguing the game industry as a whole.

"I have friends that have worked on games that sold five million units, and then they're told, ‘Sorry, your game failed,'" Hartsman says. "What the fuck kind of a world do we live in where five million sold of anything is a failure? When I came back to Trion as CEO, a lot of what I was thinking about was how to re-architect everything about the company — its finances, its targets, its hiring, its physical structure — such that we can cut that bullshit out."

Hartsman has been in the game industry since the ‘80s. His experience lends his tone a legitimate aura of frustration as he discusses this topic.

"We're churning really great people out of an industry," he says. "We're doing it through layoffs, and we're doing it through crunch. Both of those are just killers."

Atlas Reactor

With multiple projects in varying stages of development at any one time at Trion, Hartsman's hope is to evolve into a company where they don't need to staff up to hit milestones and then staff back down afterward. Instead, people can naturally shift from project to project based on where they're needed. He's also proud of the fact that Trion doesn’t have mandatory crunch time.

"Does that mean Trion is a 40-hour-a-week utopia?" he asks. "No, it's not. Everybody understands that fires happen, and sometimes you've got to fight the fire. But it's as close as I've ever gotten, and that's very intentional and very by design."

Hartsman's biggest secret to pushing Trion in this direction is just being realistic about the company's goals.

"There are some companies where every game they make has a billion dollar target," he says. "We are not that company. Some of our games will hit that; some of our games won't. At the end of the day, what we care about is whether every game is a sustainable business that can support its teams in a way where they can grow at a reasonable rate."

With Atlas Reactor, you can see how Trion's unique structure affects games on an individual level. Looking back, Cook sees the wisdom in how slowly the team built up, but he couldn't help but be frustrated at certain points along the way.

"A bunch of times during this process, we thought we were ready to go," says Cook. "We thought we had the best product. But we hadn't had the time to just stare at things."

Cook believes that by keeping the team as slim as it was for as long as it was, it allowed them to be flexible. Decisions like the switch to simultaneous turns couldn't have happened if the team had already staffed up and spent thousands of man-hours building a game designed completely differently.

"We were wandering in the woods looking for the right signs," Cook admits. "We didn't need a railroad track to the end at that point; we didn't even know where we were driving."

Atlas Reactor trailer

The inaugural class

Once Cook and Ju made the decision to shift Atlas Reactor over to simultaneous turns for all players, other pieces started to fall into place. Suddenly dodging abilities made a lot more sense. Suddenly tab-targeting, and the focus fire problem that was attached to it, didn't fit into the game. Naturally, for each problem solved, another popped up in its place.

"We had 10 phases," Cook says, now able to laugh at what was one of Atlas Reactor's biggest problems at one point. In its current form, the game splits abilities into four phases: preparation, dash, blast and movement. But initially in the transition to simultaneous turns, that number had ballooned.

Atlas Reactor

First, the game would play out buffs for allies. Then it would play out debuffs for enemies. Then there was an evasion phase for dodging. Then there was a charge phase for early damage. Then a phase for regular damage. Then knockbacks. Movement was still at the end of it all.

Cook says the team looked at games like Pokémon, which has "thousands of hidden phases," or Magic: the Gathering, whose core rules have introduced new phases repeatedly over the course of 20-plus years.

"10 phases? Well, we're less than those guys," Cook recalls reasoning to himself. "People play those games. They'll get our game. It was another problem where the game was too fun to play, so we didn't want to look at that problem."

Cook realized it was a serious issue when the team finally introduced a working user interface into Atlas Reactor. Fitting an easily readable explanation for 10 phases on the screen was not possible. The team began the hard work of rolling those phases into each other, making the game more readable and more watchable in the process. And, again, this may not have been possible with a more traditionally staffed-up team.

"One advantageous thing about people hopping off of our team is that we were going through all these iterative points," Karras says. "If we had put together a big team and just went with it — if we had a bunch of designers making abilities and a bunch of artists making stuff and the UI set in stone, we couldn't make those jumps. It's important to the process that it was a small team until we finally hit the point that we could expand."

At this point, the Atlas Reactor team has expanded. While Karras finds it hard to pin down a number — the group is still sharing some employees with other projects at Trion — he estimates that there are 30 core, full-time developers working on the game right now. And the team intends to continue ramping up.

But even now, as Atlas Reactor continues into its alpha and builds toward being publicly playable for the first time, Ju and Cook aren't done wrestling with the game's many interesting problems.

"It's important to the process that it was a small team until we finally hit the point that we could expand."

While they had decided on the game being free-to-play from the start, they're still figuring out how and in what ways it will be monetized. They're still exploring what features they need to add to make Atlas Reactor "eSports-compatible" — they don't believe they can force it to be an eSport, but they want to make sure all the tools are available for the game to succeed if players choose to shift it in that direction. They're still nailing down the process for designing and iterating on new characters, and trying to decide how many playable characters will be available when the game launches.

There are still enough problems to solve that Ju is smiling, happy to be part of a project that's challenging him every day.

As for Trion Labs, Hartsman says he isn't sure how frequently the studio will put out a call for new pitches, but he knows the time is approaching when it will again. He views Trove and Atlas Reactor — what he calls "Trion Labs' inaugural class of projects" — as opposites that prove the breadth of interesting content Trion has the skill to explore.

"We've already started getting more pitches," Hartsman says. "I got one last night."

While Hartsman has plans for a more organized call for pitches internally in the near future, he's also encouraged employees not to wait if they have a great idea now. Trove and Atlas Reactor have shown him the future shape of the company, and he plans on embracing it. Babykayak

For more on the game and those behind it, check out our in-depth discussion about Atlas Reactor in this episode of Polygon Minimap, our daily news podcast.