In 2012, as work on Mass Effect 3 came to a close, a small group of top BioWare employees huddled to talk about the next entry in their epic sci-fi franchise. Their goal, they decided, was to make a game about exploration—one that would dig into the untapped potential of the first three games. Instead of visiting just a few planets, they said, what if you could explore hundreds?
Five years later, it’s hard to find anyone who’s ecstatic with the results. Mass Effect: Andromeda, released in March 2017, disappointed even the biggest fans of BioWare’s longrunning series. Although some people enjoyed the game, it was widely pilloried, with critics slamming its uneven writing, frequent bugs, and meme-worthy animations (our own review was just lukewarm). The PS4 version of Andromeda has a 70% on Metacritic, lower than any BioWare game to date, including the ill-advised Sonic Chronicles.
Almost immediately, fans asked how this happened. Why was Andromeda so much worse than its predecessors? How could the revered RPG studio release such an underwhelming game? And, even if the problems were a little exaggerated by the internet’s strange passion for hating BioWare, how could Andromeda ship with so many animation issues?
I’ve spent the past three months investigating the answers to those questions. From conversations with nearly a dozen people who worked on Mass Effect: Andromeda, all of whom spoke under condition of anonymity because they weren’t authorized to talk about the game, a consistent picture has emerged. The development of Andromeda was turbulent and troubled, marred by a director change, multiple major re-scopes, an understaffed animation team, technological challenges, communication issues, office politics, a compressed timeline, and brutal crunch.
Many games share some of these problems, but to those who worked on it, Andromeda felt unusually difficult. This was a game with ambitious goals but limited resources, and in some ways, it’s miraculous that BioWare shipped it at all. (EA and BioWare declined to comment for this article.)
Mass Effect: Andromeda was in development for five years, but by most accounts, BioWare built the bulk of the game in less than 18 months. This is the story of what happened.
The first Mass Effect, released in 2007, was critically acclaimed but hardly perfect. Fans and critics praised its character design and storytelling, yet many people hated the Mako, a clunky land rover that the player could drive to traverse planets. So BioWare doubled down on what worked—the story, the dialogue, the combat—and ditched the exploration, axing the Mako for subsequent games in the trilogy, Mass Effect 2 (2010) and Mass Effect 3 (2012). Outside of that whole ending kerfuffle, both sequels were widely loved.
For the fourth Mass Effect, BioWare wanted a fresh start. Rather than develop a Mass Effect 4 at the studio’s main headquarters in Edmonton, which had made the first three games, BioWare decided to put its Montreal studio in charge.
Casey Hudson, executive producer on the main trilogy, would start a new team at BioWare Edmonton to work on a brand new intellectual property, which they gave the code-name Dylan. (That new IP’s code-name, a source said, came because Hudson and team wanted to make the Bob Dylan of video games—one that would be referenced for years to come.)
Meanwhile, BioWare Montreal, which was founded in 2009 to develop downloadable content like Mass Effect 3’s Omega expansion, would lead production on the next Mass Effect.
In early conversations throughout 2012, a team of directors in Montreal brainstormed ways in which to make the next Mass Effect feel distinct. This group, which included several veteran BioWare employees as well as Hudson, who wanted to help guide the project through its infancy, had lots of fresh ideas for a new Mass Effect. There’d be no Reaper threat, no Commander Shepard. They could pick a brand new area of space and start over. “The goal was to go back to what Mass Effect 1 promised but failed to deliver, which was a game about exploration,” said one person who worked on the game. “Lots of people were like, ‘Hey, we never fully tapped the potential of the first Mass Effect. We figured out the combat, which is awesome. We figured out the narrative. Let’s focus on bringing back exploration.’”
One early idea was to develop a prequel to Mass Effect, set during the First Contact Wars of the series’ lore, when the humans of Mass Effect’s galaxy had interacted with aliens for the first time. In late 2012, Hudson asked fans if they’d prefer to see a game before or after the original trilogy. The answers were resounding: most people wanted a sequel, not a prequel.. “The feedback from the community, focus groups and the team working on the project was the same,” said one person who worked on the game. “We wanted to do a game set after the trilogy, not during or before.”
So BioWare changed course, ditching the prequel idea. The word “Contact” stuck, though, and that became the codename for the fourth Mass Effect. (For clarity, we’ll refer to it as Andromeda for the rest of this story.)
By 2013, Mass Effect: Andromeda had entered pre-production, the phase of development in which the team behind a game figures out the scope, workflow, and outline of what needs to get done. BioWare also hired a new director, Gérard Lehiany, who had previously directed the Spiderman games at the Activision-owned studio Beenox.
Lehiany, who wanted to lead the game’s narrative team, came up with several ambitious ideas. One of those ideas became the core concept of Andromeda: during the events of the Mass Effect trilogy, the galaxy’s ruling Citadel council had sent a group of colonists out to a new galaxy to find habitable planets, as a contingency plan in case Commander Shepard and crew couldn’t thwart the devastating Reaper attack. You, the player, would take on the role of Pathfinder, leading the quest to rebuild civilization in the Andromeda galaxy.
Another of Lehiany’s ideas was that there should be hundreds of explorable planets. BioWare would use algorithms to procedurally generate each world in the game, allowing for near-infinite possibilities, No Man’s Sky style. (No Man’s Sky had not yet been announced—BioWare came up with this concept separately.)
Procedural generation is a process that leaves the creation of some of a game’s content to algorithms rather than requiring that each item of the game is hand-crafted. Such a technique could dramatically increase the scope of a space exploration game. It was an ambitious idea that excited many people on the Mass Effect: Andromeda team. “The concept sounds awesome,” said a person who worked on the game. “No Man’s Sky with BioWare graphics and story, that sounds amazing.”
Throughout 2013 and 2014, Andromeda’s developers played with all sorts of ideas that today sound distinctly No Man’s Sky-ish. They built prototypes in which you would pilot a spaceship around the galaxy, then use it to land on planets. From there, you could hop into your Nomad space rover and explore each new world, hunting for habitable terrain. Then you could go back into space and fly around some more. “It was a very ambitious project,” said one source. “We wanted to give the feeling of really exploring space.”
There were some hiccups, however. One lingering question for the Andromeda team was how they could possibly implement a BioWare-caliber story in a game with procedurally generated planets. Some teams felt perpetually understaffed, and there were technological difficulties. BioWare’s level designers used a tool called WorldMachine that could simulate erosion and build realistic mountains on each planet, but other teams had trouble figuring out how to generate high-quality worlds without getting in and doing it by hand. “Unfortunately that was the only team that was able to figure out how to do stuff more procedurally,” said a person who worked on the game. “No one else had the resources.”
The Mass Effect: Andromeda team knew they were going to run into major technical barriers with or without procedural generation. Over the past few years, one of BioWare’s biggest obstacles has also become one of EA’s favorite buzzwords: Frostbite, a video game engine. An engine is a collection of software that can be reused and recycled to make games, often consisting of common features: a physics system, a graphics renderer, a save system, and so on. In the video game industry, Frostbite is known as one of the most powerful engines out there—and one of the hardest to use.
Developed by the EA-owned studio DICE, Frostbite is capable of rendering gorgeous graphics and visual effects, but when BioWare first started using it, in 2011, it had never been used to make role-playing games. DICE made first-person shooters like Battlefield, and the Frostbite engine was designed solely to develop those games. When BioWare first got its hands on Frostbite, the engine wasn’t capable of performing the basic functions you’d expect from a role-playing game, like managing party members or keeping track of a player’s inventory. BioWare’s coders had to build almost everything from scratch.
(Over the past few months I’ve heard a great deal about Frostbite’s challenges. In August of last year, I went to BioWare Edmonton’s studio and interviewed many of the leads on Dragon Age: Inquisition for my book, which tells the full story of that game. In short, they had a very, very hard time.)
By the time BioWare entered pre-production on Mass Effect: Andromeda, the Dragon Age: Inquisition team had built some of the tools that they’d need to make an RPG, but not all of them. Engineers on Andromeda had to design many of their own features from scratch, including their animation rig. “Frostbite is wonderful for rendering and lots of things,” said a person who worked on the game. “But one of the key things that makes it really difficult to use is anything related to animation. Because out of the box, it doesn’t have an animation system.” (Frostbite was later attached to an animation system called ANT, that source said, but it was full of “duct-taped issues.”)
While describing Frostbite, one top developer on Mass Effect: Andromeda used the analogy of an automobile. Epic’s Unreal Engine, that developer said, is like an SUV, capable of doing lots of things but unable to go at crazy high speeds. The Unity Engine would be a compact car: small, weak, and easy to fit anyplace you’d like. “Frostbite,” the developer said, “is a sports car. Not even a sports car, a Formula 1. When it does something well, it does it extremely well. When it doesn’t do something, it really doesn’t do something.”
“Whenever you’re trying to do something that fits the engine—vehicles, for example—Frostbite handles that extremely well,” the developer said. “But when you’re building something that the engine is not made for, this is where it becomes difficult.” Designing the large maps of Andromeda’s planets became a struggle on Frostbite, where the maximum size of a map was initially 100 by 100 kilometers. The Andromeda team needed their maps to be way bigger than that. Other struggles included the streaming system, the save system, and various action-RPG mechanics that Andromeda needed in order to work.
“It’s been painful,” said a developer. “The pain started with Dragon Age: Inquisition and continued on with Andromeda as well.”
Pre-production on Mass Effect: Andromeda was a tale of two cities. Several people from the team described 2013 as one of the best years of their professional lives and 2014 as one of the worst. Whereas 2013 was full of possibilities for the developers of Andromeda, 2014 was full of politics. Conflicts emerged between BioWare staffers at the company’s two main studios, in Edmonton and Montreal. Developers in Edmonton said they thought the game was floundering in pre-production and didn’t have a strong enough vision, while developers in Montreal thought that Edmonton was trying to sabotage them, taking ideas and staff from Montreal for its own projects, Dragon Age: Inquisition and Dylan. By the end of 2014 at least a dozen people had left BioWare Montreal for other studios, and it wasn’t clear to the remaining staff whether those positions would be replaced. The animation team in particular was understaffed, sources said, and when people left, their positions sometimes weren’t refilled.
In August of 2014, Casey Hudson left BioWare. Not long afterwards, Gérard Lehiany also departed, and BioWare brought in longtime Mass Effect writer Mac Walters, who was based in Edmonton, to serve as Andromeda’s new creative director. Different people point to different reasons for these personnel shifts, but the directorial change had a massive impact on production of the game, as Lehiany had been leading the story team up until that point. When Walters took over, he brought a new vision to the game.
The Mass Effect: Andromeda team was also having trouble executing the ideas they’d found so exciting just a year ago. Combat was shaping up nicely, as were the prototypes BioWare had developed for the Nomad ground vehicle, which already felt way better to drive than Mass Effect 1’s crusty old Mako. But spaceflight and procedurally generated planets were causing some problems. “They were creating planets and they were able to drive around it, and the mechanics of it were there,” said a person who worked on the game. “I think what they were struggling with was that it was never fun. They were never able to do it in a way that’s compelling, where like, ‘OK, now imagine doing this a hundred more times or a thousand more times.’”
One person who joined Mass Effect: Andromeda just as it was leaving pre-production recalled that at that point, BioWare Montreal’s tools and pipelines weren’t quite where they should have been. Typically, a game developer will spend pre-production building a “vertical slice”—a chunk of the game that serves a proof of concept for what the team can accomplish—and then enter production equipped to build more levels. On Andromeda, that hadn’t quite happened, two sources said. The team had a backup plan if the procedurally generated planets didn’t work out—they’d just go back to the galaxy map used in previous Mass Effect games and fill it with more hand-made planets—but as they exited pre-production, they still hadn’t made a final decision. “In an ideal world you’d have one of those [planets] proven out so the process is repeatable,” said a person who worked on the game. “But we were still answering those questions of if we could do that type of thing.”
On top of that, sources said, some teams didn’t have the proper tools and pipelines to create the parts of the game they needed to create. “I think it’s universal for the game, that while they said they were done with pre-production and moving into production, a lot of teams really didn’t have what they needed to do their jobs,” said one developer.
“We started to realize by summer 2015 that we had great technological prototypes, but we had doubts they would make it into the game,” said another person who worked on the game. The Andromeda team had gotten systems like spaceflight up and running, two people said, but they couldn’t figure out how to make those systems fun to play. “I think production reality hit hard and they had to make some really strong cuts.”
By the end of 2015, Mass Effect: Andromeda’s leads realized that the procedural system wasn’t working out. Flying through space and landing on randomly generated planets still seemed like a cool concept—and by then, many people at BioWare were looking with great interest at No Man’s Sky—but they couldn’t make it work. So they decided to rescope.
First, word came down that they were moving from hundreds of procedurally generated planets to 30. Some of their terrain would still be generated by WorldMachine and the other technology they’d built, but the content would all be crafted by hand. Some time later, that number shifted again, from 30 to seven, according to several sources. For some teams—design, writing, cinematics—this move led to many questions. “There was that time of, ‘what does this mean for us on the development team?’” said one Andromeda developer. “There was a waiting period of, ‘Well, we’re doing it, so ‘what is getting cut, what’s staying, who’s gonna work on which parts of those?’”
Almost every video game goes through some sort of scope change during development, as designers and artists start sacrificing their babies in order to ship the game on time, but Andromeda was unusual. Typically, a massive rescope like this would have happened during pre-production, so the developers would have time to pivot and plan things out before starting to build the game. “If there’s one thing that should’ve happened in hindsight, the cuts that were made should have happened earlier,” said one person who worked on the game. “So there would’ve been less of them. I think in general the team tried too hard to execute a game that was not doable.”
Many of the gameplay mechanics—the combat, the multiplayer, the driving—were on time and proceeding smoothly. Those weren’t touched by the rescope. But the content—the story, the levels, the cinematics—was way behind schedule after Andromeda’s messy pre-production cycle. That was a problem.
Since 2009, the company called BioWare has actually been comprised of three studios, in Edmonton, Montreal, and Austin. Both Dragon Age: Inquisition and Mass Effect: Andromeda were developed by people across all three of those studios, an initiative that the company refers to as “One BioWare.”
In some ways, BioWare employees say, this initiative is great. Collaborating across studios can lead to some solid creative mind-melding, and it’s far easier to recruit in Montreal and Austin than it is in Edmonton, a city best known for its obscene temperatures and giant shopping mall.
But trying to collaborate across timezones can lead to some unique and frustrating challenges. People who worked on Mass Effect: Andromeda say that even the most basic tasks would take valuable time away from production of the game. When two people were in Austin, four were in Montreal, and three were in Edmonton, it could take an extra hour just to get everyone connected during a meeting, as anyone who’s tried to use video-conferencing software can attest. Add outsourcing studios in South Africa and Vancouver and you’ve got a logistical nightmare. “Game development when you’re all in the same building is hard enough,” said an Andromeda developer. “When you’re spread across the globe, it’s incredibly challenging.” (Companies like Ubisoft, known for having studios across several countries, rely upon an army of producers to organize and coordinate their work. BioWare wasn’t as equipped or experienced with that sort of large-scale production.)
To finish Mass Effect: Andromeda on time, however, BioWare didn’t have a choice. As the game ramped up production and the leads realized just how far behind schedule they were, BioWare put as many people as possible on Andromeda, although some teams remained blocked for a long time. The story in particular wasn’t where it needed to be, four sources said, because the creative team had spent so much time talking about high-level ideas during pre-production. Many of those core ideas had persisted, like the main character, Ryder; SAM, the artificial intelligence that melds with Ryder; and the final sequence on Meridian. But most of the quests and dialogue needed to be designed and written. “What you see [in the final game] is writing that has been done in the past two years rather than the full five years of writing,” said a developer on the game. “The writing team—writing the characters and everything—was unleashed too late, just because of too many discussions about the high-level direction.”
It wasn’t just the writing. Almost every Andromeda developer who spoke to me for this story said the bulk of the game was developed during that final stretch, from the end of 2015 to March 2017. Most of Mass Effect: Andromeda was made in just a year and a half, by those accounts. “It really wasn’t until Mac Walters came on board—and that was very much a reaction to the state of the critical path—he was really brought on board to give it direction and get it into shape,” said one person who worked on the game. “Before that it was quite rudderless.” Some at Montreal saw the directorial shift as Edmonton trying to take over their game, while some at Edmonton saw it as them needing to come in and rescue it.
One developer close to the project disputed the characterization that most of the game was developed in the final 18 months, saying that most of the ideas in the final game closely resembled those early visions.
A result of this backloaded development was a whole lot of nights and weekends at the office. Several people told me that the last year and a half of Andromeda was the worst crunch they’ve ever faced. “Working on that so late was difficult,” said a developer, “and harder than it should’ve been.”
Some said what made the crunch worse than previous Mass Effect games was what the developers called regression. On a typical video game project, the last few months are devoted to “polish,” a phase of the game in which the developers can fix bugs, fine-tune mechanics, and improve existing content to make everything feel as smooth as possible. In the final months of many games that turn out to be good, developers say that the game gets markedly better in that short final stretch. On Andromeda, however, everything just kept regressing. “We’ll put something together, and it’s been bug tested and signed off and approved,” said a developer, “We’d say, ‘OK, we can now move on from that to the next thing.’ And while our backs are turned, what we’d just put together falls apart.”
The root causes of this regression could be tough to identify. Sometimes they were as complicated as an engineer updating a core system; other times, they were as simple as a character modeler changing the look of an NPC. It was draining, developers said, because they felt like they were constantly redoing parts of the game that they thought they’d already finished. “The types of problems that arise from the entire project coming together in the last year and a half really compound at the end,” said one developer.
This video, comparing leaked footage of Andromeda from April 2016 to a final build of the game, is one example of what this regression could have looked like. The downgrade in quality could have been caused by any number of factors ranging from the lighting changing to the animations taking up too much memory, a source said, but those were the sort of problems they faced. While we’ve seen other games get graphical downgrades from their initial trailers to final products (Watch Dogs, The Witcher 3), it was unusual to feel like they were regressing this much so late in development, that source said.
The “downstream” teams—effects, cinematics, audio—were hit the hardest during the final months of development. “Because the entire project was behind schedule and teams were wrapping up later than ideally they should have, that just compounded even more on the downstream teams,” said one person who worked on the game.
There was a lingering feeling of hopelessness during those final months, developers said, and some who worked on Mass Effect: Andromeda say it was the most challenging project they’d ever faced.
“For the last few months of the game, we spent most of our effort just trying to keep it together rather than polishing,” said an Andromeda developer. “Just trying to stay ahead of how quickly it was falling apart.”
In game development, it’s common for the directors of a team to put together lists of the biggest risks or challenges they think they’ll face. On Mass Effect: Andromeda, the leads flagged several of these major risks. There was the story. There were the procedurally generated planets. And then there was the animation.
Mass Effect: Andromeda has been criticized for several things, but the images that will be remembered most are the memes—GIFs of gorilla walks, silly faces, and the main character’s eyes darting from side to side like she’s watching a high-speed game of tennis. It’s been subject to a great deal of speculation over the past few months, with theories ranging from naive (EA bought BioWare and now they’re all lazy) to deranged (BioWare made all the characters ugly because they’re SJWs).
In reality, as is often the case in game development, there is no single explanation. Mass Effect: Andromeda’s animation issues were, to put it simply, the result of a game with a turbulent development cycle.
During pre-production in 2013 and 2014, as the engineers and technical animators tried to figure out how they would handle Andromeda’s animation, they ran into several obstacles. They’d hired an Egyptian company called Snappers that could create beautiful facial animations, but there were lingering questions about how to implement those animations into the engine and how to scale across the whole game. And there were constant arguments over which technology to use. Some on the team wanted to use a program called FaceWare, used by EA’s Capture Labs studio in Vancouver, but others argued that it wasn’t good enough. Most of the lipsync in Mass Effect: Andromeda—like other Mass Effect games—was handled by a common piece of software called FaceFX that can interpret sounds and automatically move characters’ lips accordingly.
One source pointed to this Twitter thread by former BioWare animator Jonathan Cooper (who did not work on Andromeda but did work on previous Mass Effect games) as the most accurate description of the process. “Because time denotes not every scene is equally possible, dialogues are separated into tiered quality levels based on importance/likelihood,” Cooper wrote. “The lowest quality scenes may not even be touched by hand. To cover this, an algorithm is used to generate a baseline quality sequence.” Another big factor, sources said, is that Andromeda lets you create your own character. Fans have compared Andromeda’s facial animations to the likes of The Witcher 3 and Horizon: Zero Dawn, but those games have predefined main characters, which makes it far easier for animators to predict exactly what their faces will look like during any given cut-scene or line of dialogue.
Three people who worked on the game point to one critical moment early in pre-production, when BioWare decided to switch animation programs from 3D Studio Max to Maya, as a move that cost technical animators a great deal of work. Although it was seen as a necessary overhaul by BioWare’s leadership—Autodesk, the company behind both 3DS Max and Maya, was recommending that game developers shift gears—the animators were upset that it happened during pre-production rather than before they’d started working at all. “All that technology was invalid, simply because we’d used a red pen instead of a blue pen,” said one developer, lamenting the months of progress they lost. (Later, some of Andromeda’s animators wound up using 3DS Max for a large chunk of their work anyway.)
“That was hell for all the animation team,” said one lead developer. “That, mixed with the facial animation tech that came late, you had all the conditions to just have everything go really badly from an animation point of view.”
So as Mass Effect: Andromeda entered production, animation became a big red flag. “You have to lock all your tech in pre-production to move into production,” said a developer. “One that was not locked was facial animation. It was flagged in 2014 as one of the major risks of the project.”
One critical issue, said two people who worked on the game, was that the animation team remained understaffed throughout development. “The biggest problem with the animations was manpower,” said one of those people. “You use motion capture for in-game stuff, but the actual work of bringing that animation into an engine so it responds to your controls, that’s not something you can motion capture. It takes people and it takes time. They had very few people. They’re actually quite talented people, but if they had ridiculous schedules, then it just makes me sad to see the end result.”
Because the animation team remained relatively small, BioWare, like most AAA studios, outsourced a fair amount of the cinematic work in Andromeda, using studios in Russia, China, India, and other countries. This work may have been negatively affected by the game’s delayed production cycle. “If the story is locked and writing is locked 18 months out, then at that point you know what the scenes are gonna be, you can do up your storyboards, send that to outsourcer and they’ll send you back a finished scene,” said one developer. “But with the [writing and design] teams still working until very late in the process, that foundation shifts so much that it makes it very difficult to rely on outsourcing.”
The animation issues we saw in Mass Effect: Andromeda were the result of all of these factors, said people who worked on the game. Not just one.
Not long before Mass Effect: Andromeda launched, BioWare sent early builds of the game to mock reviewers, as nearly every AAA game developer does in the months before their game comes out. A mock reviewer will typically offer a private, early assessment of a game, a report on its strengths and weaknesses, and a predicted Metacritic range. Companies frequently make major strategic decisions based on Metacritic scores, so it’s that number that gets the most attention.
When the mock reviews came in for Mass Effect: Andromeda, BioWare’s leads were relieved—the Metacritic was expected to be in the low-to-mid-80s, according to two sources. Although Andromeda’s developers knew the game wasn’t perfect, they were fine with a score like that. If they hit somewhere between 80 and 85, they could use what they’d built for Andromeda to make the sequel way better, much like Casey Hudson and his team had done from Mass Effect 1 to Mass Effect 2.
Then the GIFs started. EA put Mass Effect: Andromeda out early for EA Access users on March 16, five days before the game came out, which led to a weekend full of memes, anger, and nasty harassment as players shared images and gifs of the game’s many glitches. Combine that with the fact that three of the other games released in March 2017 turned out to be all-time classics (Zelda: Breath of the Wild, Horizon: Zero Dawn, and Nier: Automata) and you’ve got a recipe for low review scores. When the Metacritic score finally settled, Mass Effect: Andromeda wound up with a 70 (on PS4, where it has the most reviews), far lower than those who had seen the mock reviews expected.
The results were catastrophic for BioWare Montreal. Even as the team kept plugging away on patches to fix bugs, add more romance options, and polish animations, their management informed the Montreal studio that it would be scaled down and that Mass Effect was going to be shelved for a while. All hopes for an Andromeda sequel were immediately dashed. EA moved many of BioWare Montreal’s developers to EA Motive, putting others on support roles for BioWare’s other games, including Dylan and the next Dragon Age (internally referred to as Dragon Age 4).
It’s a tragic ending for a studio that set out to do something ambitious but couldn’t quite deliver.
“At a very high level, even though the game was in development for five years,” said one person who worked on the game, “Mass Effect: Andromeda was just trying to do too much with too few resources.” BioWare aimed for the stars—hundreds and hundreds of procedurally generated stars—but just couldn’t get there. Like Ryder setting off for Andromeda, the studio’s journey ended in nothing but disappointment.