Inside the Troubled Development of Star Citizen

By Julian Benson on at

For the past seven months, I’ve been talking to the people who have been making Star Citizen. This includes its directors, a number of anonymous sources who’ve worked on it, and the man who drives the whole project: Chris Roberts. From the outside, Star Citizen appears to have been wildly successful; to date, it has raised more than $124 million from passionate fans. The money has allowed its developer, Cloud Imperium Games, to open studios around the world and employ more than 325 talented developers.

Behind the closed doors of CIG’s studios, however, it’s been far from an easy ride, according to staff. They have all faced a unique challenge: how to nail down the scope of a game whose budget and ambition is always growing. Star Citizen has now been in development for five years, and over that time it has suffered through significant changes and unrest among its staff, huge delays and, 18 months ago, a radical restructuring of all its studios. CIG has released several discrete demos over this time, but there is still not even a date for the final game, which was originally planned for 2014.

Star Citizen’s development has been high-profile enough, expensive enough and, yes, troubled enough to spawn a whole ecosystem of theories as to what’s going on at Cloud Imperium Games, from theorising about the project’s technical challenges to wild accusations about what’s happening to the money. Various community scandals have added yet more fuel to the fire, turning Star Citizen into a lightning rod for controversy. The questions I wanted answers to were: what exactly has been happening over the past five years? What are the reasons behind Star Citizen’s various delays, and what specific development problems has it encountered? Have things been mismanaged? And, as many Star Citizen backers are now beginning to wonder, can it ever actually be finished?

Chasing this information has not been easy. There’s a reason that many of the sources in articles like this are usually anonymous: people fear both legal and professional repercussions for speaking out. In the course of contacting over 100 different people while researching Star Citizen’s development, I was told by multiple sources that they were worried about legal repercussions if they spoke to the press. Speaking out publicly about a previous employer carries professional peril, too; prospective future employers may see you as a risky hire. Nonetheless, over the course of the year we found that many of the people who had worked on Star Citizen were willing to talk about their experiences, which painted a picture of a development process riven by technical challenges, unrealistic expectations and internal strife.

starcitizen12

The other side to the story, of course, is that told by Cloud Imperium Games’ current staff: its director, Chris Roberts, its project leads, and the developers who have survived the upsets that drove others away. At the stage where CIG allowed us access to Roberts and other members of the Star Citizen team at its Manchester studio, we already had a pretty clear picture of the problems that have dogged the project thus far. Roberts and his team did not deny any of them (though they did contest the severity of the problems’ impacts). But despite everything, most of the staff we talked to still passionately believe in this unwieldy, ever-changing dream project. Many of its backers still believe, too, even as others have been demanding (and mostly getting) refunds.

Plenty of people have sermonised about Star Citizen’s future. We can’t pretend to know how it will work out in the end. But we can know how it got to where it is today.


 

From modest beginnings

At the beginning, Star Citizen was not a $100m+ idea. Inspired by the success of Minecraft in 2011, which first appeared on the scene as a rudimentary prototype but generated enough money from fans to fund its ongoing development, Chris Roberts envisioned a space game that he could put together as a prototype with a modest amount of money. Roberts was famed for the Wing Commander games back in the ‘90s, but had been distanced from game development since around 2003, after his subsequent game Freelancer had been badly compromised by development difficulties and a publisher buyout.

“I was looking at the Minecraft model and thinking, if I can raise enough money from myself and some private investors, I can get a game to the alpha stage. Then I could release it and people could pay for it, and then I could improve it over time,” he says. “I was thinking I could maybe do some things with this tech that I wished I could have done when I was doing, say Freelancer. What we're doing with Star Citizen is not that different from what I pitched for Freelancer in terms of the ambition, the scope, and the living universe. It just never got done [back then].”

To build this prototype, he would need people and a game engine. “I didn’t want to build a game engine from scratch,” Roberts said. “I did that for all my old games, but it just takes time. It can have all these cool features and maybe it's 5% better than the current state of the art engine, but then I've lost two years in doing it.”

At that point in 2011, Roberts had two appropriate choices: Epic Games’ Unreal Engine and Crytek’s CryEngine. The Unreal Engine is one of the oldest and most-used engines in the industry, and in 2011 Unreal Engine 4 was nearing release. Crytek, on the other hand, had released CryEngine 3 in 2009, and the studio didn’t plan on releasing a new version for a few more years. CryEngine has powered some beautiful first-person shooters from Crytek, with richly detailed, large open environments for players to explore, but hasn’t often been adapted to other genres.

Roberts got hold of both a version of CryEngine and an early build of the Unreal Engine 4. “I was judging both and playing with both of them and ultimately decided on CryEngine because Unreal 4 back then was very early. It had all sorts of power and flexibility, and it’s used a lot – but at that point, they were still refactoring even fundamental systems. It still had time to mature and the CryEngine was just a bit more mature.”

Choosing a ready-made engine instead of building one from scratch meant that he would theoretically be able to build his prototype the game much sooner. Roberts’ choice would have consequences later, though, because as the project grew, a huge amount of the engine had to be rewritten to support the specific needs of Star Citizen. But we’ll get to that later.

At the same time as choosing an engine to run his prototype, Roberts was building the team that would help him make it. With no idea of whether the prototype would be a success, it didn’t make sense to set up a studio and hire staff at the beginning. Instead, Roberts set up his studio virtually and, besides a few freelancers, delegated a lot of the grunt work to third-party contractors who already had established teams of developers. He contacted Sergio Rosas, Roberts’ art director back in the 1990s, who now ran an outsourcing company called CGBot in Austin, Texas. He also hired a studio called Behaviour to create assets for the prototype.

As this small team started to work on the prototype, the project attracted the attention of a couple of people working over at Crytek, the game engine’s developer. Sean Tracey, Paul Reindell and Hannes Appell were all fans of the Wing Commander games and lent a hand where they could. Appell, for instance, who is now director of cinematics on Star Citizen, created videos to show off the prototype to investors using the assets that were created by Behaviour and CGBot.

Even as a prototype, Star Citizen was a global game. Chris, working out of LA, was joined by developers in San Francisco, Austin, Montreal, and Mexico (with a little help from Crytek out in Frankfurt). This early remote setup “made me think well, maybe we don't need to have one centralised studio”, Roberts told me. “Maybe we can spread this out and we can collaborate across the different areas and go to where the talent is.”

This setup, however, would be the start of another set of issues during Star Citizen’s early development: a spread-out set of developers with oceans between them faced challenges that a more centralised studio wouldn’t have. From the very beginning, CIG was using third-party contractors, remote studios, and virtual collaboration. All those choices made sense when developing the prototype, but as the team, scope, and budget of Star Citizen grew, the studio structure would have to radically adapt.


Where did the money come from?

If one thing has driven the Star Citizen idea’s transformation from modest, manageable-sounding prototype to the be-all and end-all of space games, it is money. With expanding funds came expanding ambition – and with Cloud Imperium Games working independently of publishers or corporate investors, there has been nobody to rein it in.

[Investors] always go ‘This is great, but who wants to play it? I know you sold lots of copies back then, but that was back then. Does anyone care now?’ So I was like, if I can raise some money with crowdfunding, that will prove there's a demand. – Roberts

Roberts was building his prototype to show off to potential investors at GDC in Cologne in August, 2012. But in February, another industry-changing thing happened: the Double Fine Adventure Kickstarter. Kickstarter had been around for a while, but this was when it grabbed the games industry’s attention. Double Fine’s founder Tim Schafer asked for $400,000 to make a new adventure game and got there in 24 hours. After 30 days, the campaign had raised $3.3 million.

Soon a long line of developers, old and new, started pitching games on Kickstarter, often on the premise that they were the sorts of games that publishers will no longer support. Brian Fargo raised $2.9 million for Wasteland 2, Planetary Annihilation raised $2.2 million, and Shadowrun Returns raised $1.8 million.

“I'd originally planned to go out and show the demo to investors to raise a little money,” Roberts said. “But then the Double Fine Adventure had launched. I thought ‘Maybe if I can do some crowdfunding I can establish some demand.’ It doesn't matter if I go gushing to investors. They always go ‘This is great, but who wants to play it? I know you sold lots of copies back then, but that was back then. Does anyone care now?’ So I was like, if I can raise some money with crowdfunding, that will help me with the investors and prove there's a demand.”

This is when Roberts brought in a friend from his days in the film business, Ortwin Freyermuth, to co-found Cloud Imperium Games and provide legal advice. Joined by Roberts’ wife Sandi Gardiner, Ben Lesnick, and David Swofford, all of whom still hold key roles at Cloud Imperium, they began to work on the campaign. Gardiner was in charge of marketing; Lesnick, who used to run a Freelancer fan site, took charge of community development; and Swofford, whom Roberts had first met back in the Wing Commander days, took charge of media relations.

This core group planned, Roberts told me, to “do a campaign in the same way I'd prep a launch at E3.” A teaser website was made to drum up interest; Roberts travelled around the world showing off the Star Citizen prototype to the press. In the background, Roberts had a team building a crowdfunding site.

“At the beginning there was no Kickstarter plan,” Roberts told me. “The problem I've always seen with Kickstarter is you build this community that gets excited about this game and they're engaged for the 30 days of the campaign... then, 'Bam, it's finished'. I was going to have to build my own site anyway because Kickstarter isn't viable after the initial campaign… so why don't we just build a site from the beginning that can be the home for the community, where they can have forums and we can communicate with them and crowdfund?’”

Roberts spent $10,000 building the first Star Citizen website. “Being a typical old school software developer I was like ‘A website, that's fucking easy,’” Roberts laughed. Anyone who remembers the reveal of Star Citizen remembers what happened next.

On October 10th, 2012, while on stage at GDC, Chris Roberts announced Star Citizen. All the press who had seen the prototype published their previews and the press who hadn’t seen it wrote their own news posts. A wave of interested fans went to Star Citizen’s website looking to back the game and... “Poof. It fell over and crashed,” Roberts remembers.

Here you can see the moment Roberts is told, mid-conference, that the site has crashed due to high demand:

The campaign had clearly gotten people interested, but at first very few people could get through to actually pledge their money. After a week of fixes the site was stable, but a group of potential backers were calling for a separate campaign on Kickstarter, whose platform was more familiar and robust than CIG’s home-built solution.

The team capitulated. This is the pitch video, showing off what the game looked like in those early days:

When Roberts started he had a relatively (considering Star Citizen’s current budget) modest goal. “Our actual goal was $2 million overall,” he tells me. “But by the time we had the Kickstarter, we had already raised about $1 million on our site.” Based on that trajectory the team decided to aim for a smaller goal on Kickstarter: $500,000. It launched on October 18th and after 30 days, Kickstarter backers had pledged $2,134,374. On Roberts’ own site, they had pledged a further $4.1 million.

Those investors Roberts had been hoping to court with a small Kickstarter success and simple prototype? He didn’t need them any more.

“I think it’s for the betterment... not for them, but for the project,” Roberts told me. Without private investors, he explains, “there's no a 'I need my return on the money, I need you to get [the game] out so I can sell my stake' or 'We need you to sell to EA or someone'.”

Looking at it another way, however, Star Citizen has lots of private investors. Over 1,500,000 of them, at the time of writing. They might not be expecting to make money, but they are expecting a game.

After a year’s work Roberts had a core team, a game engine, a prototype, $6.2 million, an engaged community, a vision for a space game like he had always wanted to make, and the freedom to make it without corporate investors or publishers fronting the money. That was almost four years ago and while the Kickstarter campaign closed, the one on CIG’s site didn’t. Through continued stretch goals and regular interaction with its community, pledges to Star Citizen showed no sign of slowing. By November 2014 the developer had raised more than $65 million – an astonishing number.

"The stretch goals were really starting to get a bit scary. I was like 'When's this going to stop? We're just adding more and more and more and more and more" - Paul Jones

At that point, Roberts says he “didn’t want to keep doing stretch goals, because you're just making stuff up that isn't core to what you want.” It wasn’t just Roberts who wanted the stretch goals to stop, as Paul Jones (the art director at CIG’s UK studio) told me. “There was one point when the money was going up so fast and the company was continuing with its policy of stretch goals – and the stretch goals were really starting to get, from my point of view, a bit scary,” he says. “I was like 'When's this going to stop? We're just adding more and more and more and more and more.'

“Thankfully we got to the point where that stopped and we could think, yeah, that's enough now. We have to start working this out.”

In late 2014, CIG started selling concept ships to raise more funds. These ships didn’t yet actually exist, but when they were added to the game at a later time, players who preordered them with money would have them delivered to their hangar. The prices vary wildly; you can pick up a small fighter for £15, but in the past, ships have been sold for as much as £1,900. This has proven one of the most controversial topics among the Star Citizen community, many of whom question why CIG is spending time and charging money to develop these ships for a game that does not yet exist.

CIG has, however, released a string of modules that show off chunks of the game in action. There is the Hangar Module, which lets players walk around 3D models of their ships (at least, the ones that currently exist); Arena Commander, a multiplayer dogfighting module; and an early version of the persistent universe, in which you can fly multi-crew ships, take part in missions, and use in-game shops. These modules are not Star Citizen, however. They are slices of it; you can’t just stitch them together and say, hey, that’s the game. But each one has brought in more money.

kotaku_kickstarter

As of today, Star Citizen has raised $124 million. It’s a lot of cash – but when you learn the extent of Cloud Imperium Games’ expansion, it’s not too hard to see where it might have gone.


A tale of two studios… wait, no, three. Actually, five.

Following the huge success of all these crowdfunding efforts, CIG rapidly expanded upon its remote team of staff and contractors. Today it has its own studios in Austin, Santa Monica, West Hollywood, Manchester, and Frankfurt. While this has allowed CIG to hire talented staff around the world, it has also been the root of significant development challenges. Each separate studio had to go through the difficulties of developing its own culture, and as for getting them working together? That was an even bigger headache.

“Austin was the first studio,” a source explained. “But when Chris moved over to Santa Monica and opened the Santa Monica studio [in April, 2013] that became the flagship. The relations between the studios changed so rapidly and so much. First Austin was the backbone; they were the dev studio that was doing a lot of the heavy lifting while the LA studio did the [high-level] modelling, artistry, and production.”

The original plan, Roberts told me, was to have a larger team based in Austin. It was where he had made Wing Commander and founded his previous studio Digital Anvil, so he knew the area and the sort of talent that was centred there, and he believed his dollar would go further. The problem was that in the decade Roberts had spent away from game development, many major developers had established themselves in Austin: Bioware, Arkane, Blizzard, NCSoft and Sony Online Entertainment (now Daybreak). “People aren't that much cheaper in Austin than they are in LA,” Roberts explained. Costs were high.

Does it make sense to make Austin a major gameplay programming area when you could have had job adverts out for a year and still not found anyone with CryEngine experience? -Tony Zurovec

Roberts had also opted to use CryEngine as Star Citizen’s base engine, but it’s only in the past five years that anyone besides Crytek started using the engine to make games. So finding anyone savvy with the technology became a challenge. Tony Zurovec, who is CIG’s director of the persistent universe and is based out in the Austin studio, explained the implications: “Does it make sense to make Austin a major gameplay programming area when you could have had job adverts out for a year and still not found anyone with CryEngine experience?”

Because they couldn’t find the right people to hire, CIG had to lean more heavily on contracted studios. At one point, Zurovec told me, Behaviour (the Montreal studio that helped build the prototype) had 50 people working on Star Citizen. Roberts also contracted out work on the first-person portion of the game, named Star Marine, to a Denver-based company called Illfonic. Star Citizen’s AI was being handled by yet another company called Moon Collider.

starcitizen13

This contract labour was filling the gaps in CIG’s staff, but it wouldn’t do as a long term solution. “In general,” Zurovec explained, “it's good to get everyone facing in the exact same direction and that's much easier to do when you're all part of the same company. Any time you've got contract labour, it's harder to get those people on the same long-term vision as you've got.”

So, in 2013, Roberts had two studios, multiple contractors, and three third-party studios working on an already ambitious game that was only growing more complex, and he did not have the staff and expertise to develop it. By August, though, after the release of the Hangar module, there was enough cash in the pot that Roberts could start a new studio somewhere where there were more plentiful, cheaper staff. He settled on Wilmslow, just outside of Manchester. It sounds like an odd choice, but there was someone there whom Roberts had worked with throughout his gaming career: his brother, Erin.

Erin had got his start working with his brother at Origin Systems and moved to Digital Anvil with him in ‘96. Unlike Chris, Erin had stayed in the games industry and gone on to lead studios with hundreds of staff. In 2013 he was studio director of TT Fusion, one of the studios making the Lego games. When Chris returned from Gamescom in August, 2013, fresh off the release of the Hangar module and buoyed by new backer funds, he asked Erin to found and lead a UK studio focused on making Squadron 42, Star Citizen’s single-player campaign. Erin agreed, handed in his notice, and by December he and a few ex-members of his core team at TT Fusion set up a small office in Erin’s house. By January 2014 they had moved into offices in Wilmslow and rapidly started to grow.

“[Without investors] there's no a 'I need my return on the money, I need you to get [the game] out so I can sell my stake' or 'We need you to sell to EA’” - Roberts

It quickly became clear that England had advantages over Austin and LA. It was cheaper to hire staff who were no less qualified. Today the studio has more than 180 staff, but the first 30 or so largely came from Erin’s previous studio. This was a core group of developers who had all worked together before and had an established way of doing things.

This efficiency created its own problems, however.

“It started to become clear that deadlines weren't going to get hit in other studios,” a source at the Manchester studio told me. “For example, the multiplayer demo at PAX East in 2014. If you remember the live stream… any CIG livestreams were just horrible, but that first major livestream was to show off multiplayer and it just didn't work at all. That was because [the LA team] didn't hit the deadline for multiplayer. They were still patching the game on the show floor ten minutes before they were going to show it; of course it didn’t work.

“When it became clear certain deadlines weren't going to be met, [the other studios] would sort of borrow a couple of our employees for a few weeks. A couple of employees became a couple more for a few more weeks, and a few more, and it started to accrue.”

Then, on April 21st, 2014, something strange happened: Crytek didn’t pay its UK staff their full paycheck. The same thing went down in May, then June. The company had a cashflow problem that hit all of its studios and prompted staff to leave. For CIG, this was a windfall: suddenly, after 18 months of struggling to find CryEngine-trained staff, it was able to hire people with a decade’s experience with the engine, direct from Crytek UK.

However, this influx of qualified talent also meant the UK team started to pick up more of the work from the US studios. Soon, entire other projects were being handled by the supposedly single-player-focused studio.

“We had ten-year veterans of Crytek. They knew the engine inside and out,” a source told me. “And because we were doing single-player, we ended up working with the AI guys a bit more, which meant we started taking on AI. That meant that our studio was where we knew the AI and we knew the engine.” Vanduul Swarm – essentially a horde mode to show off Star Citizen’s dogfighting – had been under the remit of the LA studio, but like much else, it became a Foundry 42 project, increasing the UK studio’s workload.

Another source on the UK side told me that “a lot of stuff started coming through from the other studios which our art leads would look at and go 'That's not good enough'. Then they'd take it and reshape it and make it work and it would come out the other end looking phenomenal. That obviously meant the art department had less time to work on its own tasks, [because it had] to keep correcting stuff from other studios. It built up like that until the whole ship manufacturing pipeline, with the exception of a couple of little bits that someone in LA wouldn't let go of, ended up at Foundry because we were the ones who'd played with it and knew about it. That took lots of resources away from concentrating on single-player.”

If you ask people to keep working 60-, 70-, 80-hour weeks without an end date in sight, it's going to put a strain on their personal relationships and personal lives. - CIG source

The art teams started to feel the strain. “These guys were working really hard to achieve a goal that was basically unattainable,” elaborates a source. “If you ask people to keep working 60-, 70-, 80-hour weeks without an end date in sight, it's going to put a strain on their personal relationships and personal lives. That in turn is going to start increasing turnover.”

The UK character team, made up of a lead and two junior artists, ended up handing in their notice over three months, between March and June 2015. “They were all miserable,” another source told me.

“Honestly, I don't know anyone working that [number of hours] here,” said Chris Roberts in response. “I don't want crunch as a culture, we don't do it. [...] We rarely do any kind of crunch to overtime. It's mostly if we're going to Gamescom. Even then you could walk around and there would only be specific people that would be staying around if they had something to fix.” When I asked Roberts and art director Paul Jones about the art team’s departure, they gave other reasons for the resignations – ones more to do with personality conflicts than working hours.

“We can't allow for individualism,” Roberts asserts. “On a smaller team, maybe you can have that brilliant person that doesn't play ball with everyone else because they just do their thing and that's all they do. On a team of our scale there is no aspect of this project where you will be working just by yourself, you have to be able to play well with others… what you want is people who are positive, can do, and want to work as a team, not always trying to go 'This won't work' or 'It's about me'.”

kotaku_map_4

“It's tough,” Foundry 42’s art director, Paul Jones, tells me, explaining the challenge of growing a team that works together. “You've got to find your directors, you've got to find your leads. Once you've run through your contact list or people you can hit up you're bringing on people you don't know and you're working off their resumé or recommendations.[...] When you're growing a company that's one of your biggest challenges, finding the right people and finding the right managers who are able to grow a team. It's not just add water and instantly you've got a plant. You've got to be able to nurture and take the knocks.”

“It's a big project and a lot of personality conflicts happen lower down, not always with me,” says Roberts. “One of our big challenges is making sure that we’re trying to get the chemistry right and that everyone works together as a team. It's something you have to keep reinforcing. It's kind of like having children and telling them that they have to say 'Thank you' or 'Please'. Just to get them to behave like you want a nice person to behave. We've spent a lot of time trying to hone chemistry, which has resulted in some departures and friction, and we've also spent a lot of time putting a structure in place that enhances it and encourages it."

While these culture clashes and growing pains could not necessarily have been avoided, the delays that they caused also cannot be ignored. When the character team left, for instance, “it basically put characters on hold,” Jones explains. “people who weren't character artists kindly stepped in and helped progress certain areas so it didn't grind to a halt and helped develop the new shader tech so we could get the quality that Chris wanted. It pretty much was a snail's pace for quite a long time. For sure, it was a difficult period.”


The problem with contractors

Outside of Cloud Imperium Games’ own studios, there were also significant problems with the contracted studios whose work CIG had relied upon, stemming largely from a lack of communication and direction. This would mean that mistakes would be left unchecked and development could progress in the wrong direction for months, necessitating further months of time-consuming course correction. This was a very frustrating experience for some of the developers working at those external studios, who shared their experiences with me.

In 2013 CIG contracted Illfonic, a third-party studio based in Denver, to build the Star Marine module and the first-person systems needed for Star Citizen. Star Marine was to be a vertical slice (ie, a demo) of Star Citizen’s first-person multiplayer. Much like Arena Commander, it was essentially a multiplayer mini-game – something the backers could play with to get a sense of what CIG was building for the final game. For CIG, the module would act as a test bed for its ideas for first-person.

The first release of Star Marine was to focus around a single map, Gold Horizon. It was a deep-space pirate base that two teams would battle over.

The plan was to build ship and station interiors from environment kits. For this to work, a source explained, “you need to have the same slottable pieces for all different types of art styles. All standard doors, for example, whether they be for a moon base or a mars base, have to share the same dimensions. If you’re building a new environment and new art assets to go with it then you create them as standard, modular pieces so other environments in the same style can be built quickly without needing bespoke assets.”

starcitizen11

CIG wanted to use the environment assets Illfonic had created for its Gold Horizon space station level as an environment kit. But when CIG tried to fit the assets into their levels, they found that none of the assets worked with CIG’s kit system; they had all been built to the wrong scale. A source told me that after the studio had worked on the Gold Horizon map for more than a year, CIG asked Illfonic’s artists to remake the whole thing with new metrics to satisfy the Squadron 42 team. “It sucked for the artists,” my source told me.

“I'm always very perplexed by this,” Roberts responds, when I ask him how this deviation had happened. “We got everyone together and had a whole art summit in Austin in 2013. I thought we were all on the same page but I guess at some point we weren't, because I started to hear back from the environment guys that 'this thing doesn't fit with what we're doing.’ The communication wasn't good, but it was also a problem because there wasn't one person in charge of all of that.”

In short, no one person was in a position to spot deviation before it became too severe. This wasted months of work and necessitated months more to correct the problem. Illfonic worked on Star Marine for nearly two years, but production issues like the one above have meant that nothing it worked on has been released, and much of what it did create has been rewritten by CIG.

We can't allow for individualism... what you want is people who are positive, can do, and want to work as a team, not always trying to go 'This won't work' or 'It's about me'. - Roberts

 

According to Roberts, these problems are now largely behind him. With the opening of a Frankfurt studio in 2014, made up of a group of ex-Crytek developers who were fed up with not getting paid on time, CIG finally had the resources and the talent to bring all of Star Citizen’s development in-house. This was the final major Cloud Imperium Games expansion – for now.

“Around this time last year was when we got to the point when it felt like we had a decent amount of critical mass in terms of the team size,” Roberts says. “We were doing more outsourcing in the past to supplement all the stuff we were trying to do. [...] Once we had our studios and our production processes established and strong leaders here, the outsourcing became less and less effective. We shifted work from the outside to the inside as we started to have resources to do it.”

Trying to make an ambitious game while also establishing hundreds of developers in separate teams, most of whom have never worked together before, would be extremely difficult under any circumstances. Any new game studio will take time to find its rhythm and develop a shared culture. But CIG was doing so in the public eye, funded by more than a million people. Even if Star Citizen’s growing pains are all in the past, the impact hasn’t been erased. Fledgling studios were working at far less than peak efficiency and communication was, frankly, deficient, leading to development delays and frustration among CIG staff. Personnel, production and management problems affected developers’ lives at every one of CIG’s studios during this period.

And, of course, all of this sucked up a lot of time. In game development, time is money.


Stretched thin

The Star Citizen described in the 2012 Kickstarter campaign is rather different from the Star Citizen that is now in development. The initial goal was essentially to fund the development of a single-player campaign, with all the features of Star Citizen’s living universe set aside as stretch goals. But CIG decided not to complete the single-player first and add the massively-multiplayer components in later; instead it’s building both at the same time.

As well as working on both Squadron 42 (the proposed single-player mode) and the expansive persistent universe in tandem, CIG is also continuing to support all of the different modules that it’s released so far with bug fixes and new features. (You can read about each of Star Citizen’s released modules in greater depth on the game’s site, but they are are essentially demonstrations of some of Star Citizen’s key features: the Arena Commander module, for instance is a multiplayer mini-game that shows off Star Citizen’s dogfighting.)

Multiple sources have told me this style of development was a mistake: it spreads the team and resources thinly, it means CIG has to maintain multiple separate live releases as well as continuing work on the game.

“A wise person would say, you raised money for a spaceflight sim [Squadron 42] and you've got grand visions for a first person, third-person social component,” opined one source. “Let's use the crowd-sourced cash and build the flight component. Using this giant amazing team, focus on all these various attributes. Monetise that portion and then roll onto the next component.

starcitizen8

“You can solve those problems one at a time, but when everyone is trying to solve all the problems all at the same time, you're going to do nothing but waste money and come across huge technical limitations that you're going to need more hands on to solve. The problems were inherent from the beginning. It's just not a wise decision to attempt everything at once.”

The immediate effect of working on so many systems concurrently was that CIG’s engineers were spread thinly. Across the studio there were developers who couldn’t move forward on their slice of the game because the underlying technology they needed to implemented hadn’t been finished. Everyone needed time from the engineers, and there just weren’t enough of them to go round.

“We were demanding all of this tech to make single-player work, but two thirds of the engineers across the whole company were dedicated to getting the persistent universe online,” one source recalls. “Which in and of itself is a mammoth task. Getting the server side stuff up at the backend, getting all the support working, and getting a platform with which to support a game in the first place… there's a reason it took Blizzard as long as it did to get World of Warcraft on its feet and a reason it took another five or six years to figure out how World of Warcraft should really work.”

The problems were inherent from the beginning. It's just not a wise decision to attempt everything at once. - CIG source

This delay in development on the technology side threw up problems for the UK team working on Star Citizen’s single-player component, Squadron 42. “We started to hit a tech wall with the single-player,” a source explained. “If you start with an engine that has netcode and supports connections between multiplayer users, you can start to prototype a multiplayer game fairly quickly and easily with just some different coloured cubes. Single-player [is different], you've got the considerations of flow and pacing, and where you want things to happen, and how the plot is going to go in, and how you're going to do exposition, and how all that's going to come together. Without tech to support it, it becomes very difficult to prototype anything.” The design team “very much slammed into a tech wall and it wasn't going anywhere anytime soon.”

“I would say of all of our people, the most frustrated would be our designers because they're the ones that feel like they can't move until they have the tools from the engineer,” said Roberts, addressing the problem. “The engineers are working to provide the tools but, of course, engineering involves the most R&D and is the most unpredictable [department].” Roberts tells me that art and design are more reliable to schedule, engineering work can often be hit by surprise delays, such as bugs, that slow everything down. “So there are cases where the designers are moving forward, but they're waiting for a system for them to finish off.

starcitizen3

“It's also why engineering is one of the biggest things we've been ramping up and hiring. We've been building up, but you always hire more slowly than you plan and schedule. Not deliberately; it's just because it's hard to find good people who fit in.”

This shortage of engineers in 2013 and 2014 fed an unhelpful cycle. “What ended up happening was that a studio would start working on smaller components to verify it was something they wanted to go after,” a source explained. “Once that team was able to prove that a particular component was pretty neat, it became mandate and a team was built to support it, spreading the team elsewhere even thinner.” This fostered a culture where studios were “vying for power,” the source said. “They operated on the principle that 'If Chris is paying more attention to the studio, we'll have more power.'”

“A year ago our internal headcount was much less, we had much less capability, and so we absolutely were stretched,” Roberts admits. “We still are understaffed for what we need to do on the engineering side, which is why we have open positions. Back then it was worse. To have enough people to work on the online stuff or the Squadron stuff, you had a robbing Peter to pay Paul situation: you didn't have the full staff. Now [we] have a much bigger staff, so we're managing to make forward progress on the Star Citizen MMO side, as well as Squadron 42, as well as support the live modules. Last year that was much harder.”

The race isn't finished yet. Wait till the finish line is crossed and then I'll say people can make judgements about whether it was the right way or not. - Roberts

Roberts hints at another problem in the quote above: the added drain of supporting a live release (or, indeed, several) at the same time as trying to develop a coherent game.

“The demands of the live version of the game take some of the best people away to support [it],” Roberts told me. “If you add up the days it takes [engineers] to do the task they were scheduled for, they're not usually that far off it. The big uncertainty is how much they get sucked into trying to fix an issue that's blocking the next live release.”

There are benefits to the method CIG has chosen, say some of Star Citizen’s developers. All the work that CIG does in making the modules is done in a shared codebase, so any engineering done to create a feature for the modules is accessible to all developers working on the game, Erin Roberts explained to me. The team at one studio can draw upon work done for another, which means that no artist, designer or engineer should have to do anything twice.


Another way?

Critics of Star Citizen’s release structure often point to Elite Dangerous as an example of how the game should have been made. Searches of Star Citizen’s forums bring up a number of threads where backers are discussing which method CIG should have opted for, like this one started in January 2016.

Shortly before Star Citizen’s Kickstarter campaign wrapped up, David Braben, the founder of Frontier Developments and creator of the hugely influential space game Elite, crowdfunded a contemporary sequel to his 32 year-old game. Braben promised a vast space game filled with star systems, space stations, pirates, and traders, one which players could fly around in with other players. The comparisons to Star Citizen were immediate.

starcitizen6

The campaign was a success, but besides selling pre-orders to Elite Dangerous on its website, the budget, stretch goals, and core game were largely locked. Frontier released beta versions of the game for backers to test, before releasing the full game in December 2014. Since Elite Dangerous’ release, Frontier has continued to patch, balance, and add new features. It has also released major expansions; the Horizons expansion, for instance, added planetary landings. While Elite Dangerous wasn’t as feature-rich as Star Citizen plans to be at launch, it existed. It was playable, and enjoyable. And there was nothing stopping Frontier from improving the game over the next few years.

“It's just an approach,” Roberts said when I put this to him. In his opinion, while Frontier’s approach was “a totally viable way to go about it”, it launched with the “basic, bare minimum” of features (“you're just trading and earning some money and upgrading your ship”), the game was “the same game you had 20 years ago”, and Frontier is adding new things in and solving the problems of each new feature “one thing at a time.” CIG’s approach, he says, “is a whole other scale of difficulty”.

There was a period of time where our fan base was able to do more with what they had than we were. - CIG source

“It's better to have your engineering considerations taken care of now, rather than trying to retrofit as you go along,” Roberts argues. “It definitely is a harder approach to take, but I think long-term it will build a better foundation.” Roberts compares Star Citizen to MMOs like World of Warcraft and EVE Online, saying: “if it works it's hopefully going to stay alive for a long time, like 10-plus years.” For that to happen, “it needs to have a solid foundation that's scaleable and can be constantly maintained and built upon easily. If we don't do a lot of this stuff upfront then we're going to repay that technical debt later on.”

Roberts admits that working on solving all the technical challenges of Star Citizen’s feature set before launch “has caused a lot of the frustrations on the design and the art side, because they've been blocked on getting past this technical debt.” But he argues that “you will see much more rapid progress” in Star Citizen’s development with those technical hurdles behind them.

To his detractors, Roberts says: “The race isn't finished yet. Wait till the finish line is crossed and then I'll say people can make judgements about whether it was the right way or not. Instinctively, it was the way that felt right to me.”


Engine trouble

Let’s go back to a really early point in the Star Citizen story: the decision to use CryEngine. Multiple sources told me that adapting a game engine built for first-person shooters to run a massively-multiplayer universe has been a huge hindrance throughout Star Citizen’s development. Some of them say that building an engine from scratch would actually, at this point, have been more efficient.

“The obstacle CIG has had from day one is CryEngine,” one source said. “CryEngine is an excellent engine for making one kind of game: shooters. [To make Star Citizen] they've had to completely gut it.”

In many instances, rewriting sections of the engine makes total sense: the CryEngine’s network code was designed for small-scale multiplayer games, not potentially thousands of players; the AI code couldn’t natively support the complexity of tasks that Star Citizen’s living universe required; the rendering system was built to light environments on Earth, not in space.

Other decisions made much less sense to my sources. “They wanted a brand new first-person system so, having licensed CryEngine, an engine built from the ground up to create first-person games, they stripped it and started building a brand new first-person system inside an engine that already had one,” said a source at Foundry 42. “I just think that was a stupid, stupid decision.

starcitizen1

“I remember a lot of meetings where we were saying 'We've got these missions but we need functional first-person systems to do it.' We were seeing members of the public taking the assets from what had been released and putting them into CryEngine and using its native first-person system to block out levels. There was a period of time where our fan base was able to do more with what they had than we were. We were saying, 'Even if it's not the tech that we end up using, if we have what they have we can at least start work, putting stuff together.' It didn't materialise. That was bad.”

Stripping out the first-person systems that come supplied with the CryEngine seems like a bizarre decision, but Roberts says it was essential for his plans to unify first and third-person perspectives. “In most first-person shooters, the first-person is floating hands in front of a camera and the third person is an entirely different set of animations and assets,” he explains. “It works fine in first-person games, particularly single player first-person games. It's also the reason why, if you play a game like Crysis, the fidelity of the animations in the single player game aren't the same as you see in the multiplayer game. That's also true of Call of Duty.”

In other words, the animations you see as a player in the first-person view are different from those another player looking at you would see. You can see what Roberts means in this GIF released by Olly Moss, showing a Firewatch animation from a third-person perspective:

In Star Citizen, Roberts wanted to use one unified set of animations and character models no matter what perspective you were using. “That aspect is incredibly technically challenging,” he says. “Very few people do it, and certainly not to the fidelity we do. You just can't do this stuff out of the box with CryEngine SDK or Unreal.”

One big source of delays was a fundamental change to the CryEngine: moving it from 32-bit calculations to 64-bit. This would allow the engine to track an object's position in the game space to a significantly higher degree of precision, catering for vast open environments. When CIG tried creating these environments in the engine with only 32-bit tracking, they ran into problems. “We were having all kinds of things like the HUD tearing and ships jumping around all over the screen,” says Nick Elms, creative director at Foundry 42.

“Some of the technology debt that we’ve had to pay just in terms of moving from 32-bit to 64-bit precision is huge. It is a massive task under the hood of the engine,” Elms continues. But, he says, the shift has allowed the team to move from the “tiny little bubble that we tested space combat in [in Arena Commander] to these huge environments that you can now fly in. Not long ago that just wasn’t even feasible.”

It's one thing to be someone who goes to work and doesn't work because they don't want to be there. It's another thing to be someone who likes their job but is unable to do it - CIG source

Because so much of the engine was being replaced, and updates were continually being pushed out to the different studios, it meant a lot of work had to be redone. “Something gets changed and breaks what you've done, so you have to go back and do the work again and again,” a source told me. “This is why you make sure your tools have been written and finished before you start building the game.” The source admitted, however, that “saying that as a criticism of CIG [specifically] is kind of unfair because virtually every game studio in the world suffers from that problem… In this instance, simply because of the scope of the game and the amount of rework being done it was particularly bad.

“I tried my hardest to wrap this tech into something that was functional –  to create a proof of concept of what I was working on – and it was just intensely frustrating. It's one thing to be someone who goes to work and doesn't work because they don't want to be there. It's another thing to be someone who likes their job but is unable to do it.”

This was all a shock for many in the UK studio, particularly the ex-TT Fusion developers. “We just came from the Lego universe, which was very much getting a game out literally every four to six months,” Erin Roberts told me. “It’s a very different way of developing because [at TT] they’re on a twelfth-generation engine where you don’t even need to evolve the program route for the first three months of the project. Whereas, here, we’re innovating every day and expanding the capabilities of the engine in different areas to come up with what we need to deliver.

“You can have a day where you come in and you are blocked. For people who worked at TT that’s unforgivable really; it’s something that somebody would be dragged outside and shot for. Whereas here it’s one of those things that we have to accept – you get on with something else, you’ve got to move to something else and start doing some paper designs or something like that.”

starcitizen10

“One or two guys that came across and weren’t comfortable building technology while making a game, so they went back to make Lego stuff,” Elms says. “They enjoyed having an engine which is there and they know exactly what they have to do.”

While Chris Roberts understands the frustration for designers, he maintains that the work had to be done. “There isn't an engine that can do what we're doing,” he says. “If there was then we'd have licensed it. We had to refactor it to scale. [...] You can have millions of kilometres you can travel. Draw distances are hundreds of thousands of kilometres. You can't do that in a 32-bit engine. It doesn't matter if you're using Unreal or CryEngine or Unity. We would always have had to do that refactoring. Yes, you can get into CryEngine and do things simply and you can do the same in Unreal and Unity, but it won't work for what we need at all.

“I don't believe that if we'd picked Unreal 4 that we'd be in any better a position,” Roberts continues. “I can see how [adapting the engine from 32-bit to 64-bit] would be frustrating for some people, but I don't look back and say 'I wish I'd gone with Unreal'. We would have had this no matter what. People in the industry, even internally here, our designers and artists aren't technically [informed]... they just know their tools.”

As the budget and scope of the game grew, one source opined that CryEngine made less and less sense as the basis of the game. “[Star Citizen’s] original remit was $500,000 to create a nostalgic, spiritual successor to Wing Commander in CryEngine. That's a perfectly reasonable remit. Then more money started coming in and Chris Roberts started to make promises to the press that he'd not discussed with the development team. I don't know whether he had plans, but all we knew was that he was promising things in videos at PAX and in interviews that the development team could not deliver with the current state of affairs.

CryEngine was a fine pick when $500,000 was all they were looking for and they needed tech to build a game on. You can't build your own engine for $500,000. But you can with $100 million. - CIG source

“It was clear that CryEngine wasn't up to the task,” the source continued. “CryEngine was a fine pick when $500,000 was all they were looking for and they needed tech to build a game on. You can't build your own engine for $500,000. But you can with $100 million. In order to make Star Citizen work it needs proprietary tech. A lot of what was happening was to do with rewriting CryEngine in order to make it do what was needed. That obviously slowed everything down.”

starcitizen14

Another problem with CryEngine, a different source told me, “is that it's got an obscene number of lines of code that double up and old legacy code that sits there undocumented. What can we throw out from the old CryEngine and, if we do add our own stuff, what other components are going to negatively impact this? It's like going to a hardware store that's been hit by tornado: yeah, you can find a hammer at some point but you've got to dig through a bunch of crap to find it.”

There are still major components of the CryEngine that are being rewritten. The engine’s networking code, for instance, is currently in the process of being stripped out and redone. But not all my sources agreed that CryEngine was a mistake. One agreed with Roberts that Unity and Unreal, the other two serious options, would have thrown up roughly the same problems. It might have made more sense for CIG to build their own engine, but at the beginning of this project, the scale was so much smaller that it would have been impractical.

I think it's easy for people to scapegoat CryEngine, or any other engine choice. - Roberts

“I've built Engines before; while at Digital Anvil we built an engine from scratch, and I did the same while we were back at Origin,” challenges Roberts. “ It was at least two years before we had stuff we could show, really viable, visible stuff. When I got into this I said I don't really want to waste time, that two years...

“I think it's easy for people to scapegoat CryEngine, or any other engine choice. If we were doing a simple FPS shooter, we wouldn't have had the same issues or challenges, but we aren't doing that, and we didn't raise the money that we raised because we were doing something simple. We raised it because we wanted to push the boundaries.”


A heck of a production

As CIG took on more staff and began tackling more areas of Star Citizen’s development at once, it became clear there were problems with the game’s production – not only with how development was being planned out, but also how it was managed, and how all the work was delegated. Over the past two years, there has been a significant reorganisation of the entire company in an attempt to enable it to better build the game.

A number of different sources tell me that from the early days Star Citizen, the high-level roadmap goals were well-defined, but the way it was broken down into tasks was lacking. “With a milestone you typically break things down by some kind of project management structure where you can say, this is a priority one, this thing has to get done within these four weeks. This is a priority two, these are priorities four and five, these things can drop off and these are stretch goals, we can push them into the next milestone,” says a source.

My sources tell me that granular organisation just wasn't happening, meaning everything could become a priority. “If everything's a priority, then nothing is a priority. If nothing's a priority, then things just get done when they get done, and that made things like the milestones pretty useless.”

starcitizen13

A second problem: teams were overreaching with what they could realistically achieve, being overly ambitious with the goals they set themselves each month. Normally, those sorts of expectations would be reined in by the producers, but that’s where CIG had an issue. “A lot of the coordinators and producers didn't have a lot of experience in this type of development, so they didn't know what could actually be achieved,” a source told me.

“If an artist, an engineer, or a designer says 'it's going to take two days' but it actually takes a month, [the producer] wouldn't know how to assess the time because they don't understand what's being asked of them, or they're not factoring in the person's previous speed or ability to execute. They would simply say 'Cool, you think that's when you will get it done, let me write it down.'”

Making games, particularly games as large as Star Citizen, is a complex business, where no developer is ever really working completely in isolation from anyone else. This means that when a task takes longer than planned it causes delays not just for that one developer, but for a whole string of people. A single ship in Star Citizen is made by multiple people, from concept artists to 3D artists, to texture artists, technical designers and a whole host of others. Same goes for characters, props, environments. As a result of problems with the production pipeline, at one point, character models that were supposed to take three months were taking closer to six.

It was just a giant clusterfuck of documents and data, a lot of which conflicted with other bits. Sometimes documents were out of date and sometimes someone worked on something that someone else had already done. - CIG source

“There was a lot of documentation,” one source said. “Too much. It was just a giant clusterfuck of documents and data, a lot of which conflicted with other bits. Sometimes documents were out of date and sometimes someone worked on something that someone else had already done. Other times there were two people making the same thing simultaneously but unaware of each other, and so they’d end up making two different things designed to do the same job. This was across all of the studios.”

No game development starts with a perfect pipeline; they are established as the best practice is discovered. But as Jones puts it, for CIG it was “the perfect storm. [We were] growing the company, there's new technology, multiple studios. You're trying things out and that's difficult when half of one studio is responsible for one thing while the other part of the project is with another studio in a different time zone.”

As 2014 drew to a close, Roberts tells me that he got together with his brother, Erin, and said “this isn't going to work if we're going to continue [in this fashion].”

The brothers took a step back and saw that while there were production problems for the company as a whole, “rating it on who was delivering, the UK was being much more consistent,” Chris Roberts tells me. He took a closer look at how Erin was running the UK studio and realised how rigidly the team adhered to their production schedule, and how closely the production staff tracked the developers’ work. While Foundry 42 was a new studio, the core team of developers had all worked together for years making Lego games at TT Fusion. “They did so many of them [...] they had to be super organised,” Roberts explained. “Sometimes they had six months to do something.”

starcitizen4

Roberts mandated that all the CIG studios would adopt the UK method. It wasn’t a decision taken well by everyone: “The ship pipeline was run out of LA at first and then it just wasn't working and the UK was frustrated, so we moved it over to the UK, and then the folks in LA felt like they had been demoted, and they were upset with that,” says Roberts. “There was some fallout, some people left.” Erin Roberts became responsible for CIG production globally and CIG hired a lot more producers to cope with the finer detail scheduling Erin’s system required

I think you got a lot of stepping on each other's feet and that added to the frustration the different studios felt for each other. - Tony Zurovec

A management hierarchy was also introduced for each different pipeline. So, for example, Nathan Dearsley is in charge of ships, Francesco Roccucci is in charge of AI, Ian Leyland is in charge of environment art. “What you now get is clearly delineated responsibilities in terms of who is doing what,” Tony Zurovec tells me. “Once you have that, I think in general it minimises your problem of blame. [...] If you go back to the days of Arena Commander [the multiplayer dogfighting module], you had so many chefs in the kitchen, I think you got a lot of stepping on each other's feet and that added to the frustration the different studios felt for each other.”

Paul Jones, Foundry 42’s art director, says that the new structure has conferred advantages for him, personally. “[In the beginning] I was massively overloaded with just trying to deal with the whole of Squadron 42 and it's different art disciplines and the concept art,” he says. “I'd have conversations with Erin and let him know that some plates would stop spinning, that I couldn't spin them all. We realised we had to spread the load more, definitely in this studio.”


Firings

There were victims of the restructuring: in Austin, a significant portion of the studio’s developers were made redundant.

When the Austin studio was founded, it hired designers, artists, engineers, all the people it needed to start fleshing out Star Citizen. There were, as we’ve seen, a multitude of reasons why development got moved to different studios; today, Austin is focused on server infrastructure, customer support, developer operations.

“The parts of the Austin development house that made sense, like the main backend server engineering, the persistent universe stuff, and things like that, that's all still handled out of Austin,” Zurovec explains. When it came to the other parts of development, though, “better opportunities presented themselves – more people in different locations where they had tonnes of experience with the engine.”

starcitizen2

This restructuring was perceived by some on the outside as CIG closing down the Austin studio. “People were freaking out, saying 'Oh, you're shutting it down, it's a catastrophe,'” Zurovec recalls. “It's actually not a catastrophe: you’re trying to spend every dollar as effectively as you can. The irony is you're making difficult decisions to make it better and the public perception sometimes gets skewed because you don't have all that information out there. You've just got that one obvious piece: ‘They're laying off a dozen people here’. It's true, but what you miss is that Foundry 42 in Germany is now 50 people and the UK has been hiring constantly.”

Zurovec lays out the high level reasoning behind the restructure and why it was a good thing for CIG. But I’ve spoken to people at Austin, and those early mistakes that needed to be course-corrected had a human cost: people lost their jobs. That shouldn’t be glossed over.

“There were, I think, two purges at the time I was there that were anywhere from six to maybe fifteen [people],” a source told me. “Some I felt were deserved and some were absolutely not.

It's very easy when you have distance to blame it on the other guy, to say 'They broke the build. Those idiots in wherever did it.' - Roberts

“The way they did it the last time was horrible. Other places where I've been laid off, or been there during lay-offs, they usually have two meetings. Everyone gets an invite to one of the two meetings. You don't know if you're in the meeting of the people who are being laid off or the people who aren't. It's actually become a joke within the industry: it's like being led into a room and being shot in the back of the head – you never knew that it was going to happen. But, at CIG, they set up in one of the offices and they called people in over a period of about six hours.

“You had no idea if you were going to be called but everyone who got called in got shot in the back of the head. So there was this high amount of anxiety where your supervisor is coming over and assuring you that he's pretty sure that you're safe, but even he doesn't know. Then the guy next to you gets called into the meeting and you lock eyes with him and there's that look of 'Oh, man, I'm so sorry.' It's the worst way you could have handled lay-offs. They didn't do it to cause anxiety, but I was surprised. I'd never seen it done that way.”

When I talked to Chris about redundancies and restructuring, he maintains that it was necessary. “You've got to get the chemistry right,” he says. “It's not going to look good on their CV. But they may not fit. And sometimes you can hire someone and after a month or two realise it's not going to work out. That's the hardest thing, because you never hire someone because you want to fire them. You never do that. It's so much work to fire people.

“I'm actually bad at that,” he claims. “I'm a big softie and always giving people extra chances, but I'm trying to be tougher on our organisation because every time we're in a situation I think we can turn around, but the person doesn't turn it around, they get a little more bitter and pissed off, and then the exit is more noisy.”


The blame game

The sense that things weren’t working as they should and that it was someone else’s fault came out in a lot of my discussions with CIG staff, past and present. It was described as a culture of blame, where when a job was delayed or couldn’t be completed, developers would shift responsibility onto other people, particularly those working at other studios. The animosity led to a hostile work environment; one source even told me that they’d suffered panic attacks as a result.

Many developers I spoke to said that they felt their own studio was pulling weight for the others. Several sources I spoke to on the UK side mentioned that UK artists ended up having to re-do a lot of the US team’s artwork – but according to Chris Roberts, similar accusations were lobbed back in the other direction.

“You have that from a UK guy here who's complaining,” he says. “I had exactly the same stuff from the US guys complaining about the UK guys. It's very easy when you have distance to blame it on the other guy, to say 'They broke the build. Those idiots in wherever did it.' We've actually been working really hard to stamp that out. We try to make sure that key members of the team spend time with the people they're working with in the other studios, because that's a constant challenge, keeping the communication going, fostering cooperation, and trying to tear down any barriers that are erected by distance.”

starcitizen7

The culture of blame left many of the staff and former staff I spoke to demoralised. “It was a nightmare at times,” one source told me. “It was a fear for your job. You know the adage ‘shit rolls downhill’? If anything went wrong, there were a few people whose problem it never was. The only people who ended up with responsibility for a mess-up or a problem were people with no power to put it on anybody else.”

In many of my interviews (though not all), Chris Roberts himself was singled out as a domineering and, sometimes, disruptive presence.

“It would usually be around the time of a presentation,” another source told me. “We'd be near a deadline. Chris would insist on something being done a certain way, you would try to advise 'Well, actually no. That might make this happen...' and he’d say 'No, no, no, you need to do it. Do it this way.' You'd do it that way and then it would break and he'd be unhappy with it.”

“That was an absolutely common occurrence,” the source continued. “It creates this incredibly stressful environment where you have no real control over your work, and you're going to be held responsible. So everyone is trying to pass the blame to someone else.”

“I've worked with other directors in the past who had huge ego problems, but they weren't bottlenecks,” a source told me. “They were instruments to resolve problems, not create them. I think in Chris' case he creates a lot of the problems by the way he chooses to position himself within the company. He is the lead, the director, the project director, and the CEO. You're going to destroy the manager's ability to manage the teams and the directors' ability to manage the teams when you've got that level of interruption from a person that they all report to.”

"I really do listen to everybody but then I make a decision and I expect my decision to be enacted" - Roberts

“I really do listen to everybody but then I make a decision and I expect my decision to be enacted,” Roberts said in response to the claims above. “When I really lose it, it's because people passive-aggressively don't [do what they’ve been instructed], and instead try to push their agenda, coming up with reasons why it needs to be this other way. That really, really annoys me because it just creates friction all the time. I like to have a lot of really good creative people around and I like them to contribute all their ideas but when I say we're going left instead of right, everyone needs to go left. It's not an ego thing – it's about the project.

“If you don’t have one singular drive or vision that you're working towards then it's going to become muddled. That's kind of why I like the setup of movies. You may disagree with what the director is doing, how he is shooting a scene, how he is blocking it, but it doesn't matter: you still make it happen for that director because it's going to be on his shoulders. If the game doesn't work, it's on me, not on a junior designer or something. So it's my call whether it's right or wrong. So, please say 'This is what I think should happen'; I will listen and in quite a few cases I'll be like 'That's pretty good, let's try that'. But when I've made the choice [...] I expect people to go that way. I really don't like passive aggressive behaviour. It just really drives me crazy.”

starcitizen15

Another source told me that Roberts would sometimes give leads a dressing down in public. “The leads had a Skype call once a week and every week someone would be read the riot act, publicly, for all the leads to see,” another source told me. “It was a public shaming. It's something that happened more often than it needed to, and there's no justification... there's never justification. [...] Chris' involvement absolutely created that atmosphere of fear and tension and overall agitation.”

“I do lose it,” Roberts told me. “There were definitely times in the early days where I felt like people kept giving me the same answers, and I would let them have it. I’d say 'You just told me this last time. You've got to have a better answer. Why isn't this thing working? Give me a reason. You can't just say it was their fault'. I'm generally quite nice until I feel there is something not happening, or something that I need to call out. But that's the way I've always been. Sometimes I think it can be hard for some people.”

I generally give people a pass a few times but at some point I will say 'You've got to get your act together. A bunch of people are depending on this thing to happen, so what are we doing? What's the plan? -Roberts

“What really irritates me is if someone says they were going to do something and then they come back and say they didn't, but they don't have a solution or a reason,” Roberts elaborated, a little later in our interview . “I generally give people a pass a few times but at some point I will say 'You've got to get your act together. A bunch of people are depending on this thing to happen, so what are we doing? What's the plan?' People can consider it being tough or public shaming, but I'm not going 'You're a fucking idiot' – that's not how I operate. But I will call people out if they've said one thing and then come back a week later and don't follow up properly, or haven't had attention to detail on it.”

Various sources detailed instances where Roberts had pushed for something in Star Citizen and members of the development team pushed back. Once, a source says, Chris came to work after playing The Order: 1886. Impressed by the highly detailed art, he asked CIG’s character artists to match that standard. The team, my sources told me, saw this as impossible. “That's fine for a single-player game where you're able to control stuff and stream things in a certain way,” one source explained. “You do not expect that for any kind of MMO or open world. But that's common knowledge for anyone that's worked in games.”

The second example I was given was from a time when Roberts came back from seeing another in-development CryEngine game, Kingdom Come: Deliverance. He had been deeply impressed by the character inventory and outfits system, which involved multiple layers of clothing where each item has different properties, from its material to its weight and shape, that affect how it animates. Chris wanted it for Star Citizen.

starcitizen5

“All of the developers who had worked on inventory systems said ‘OK, well, that's why this works in their game and why it won't work in this game’,” a source told me. “We spent four months having to prove ourselves right because that's the way it works with Chris Roberts. He will never believe you if you say you can't do something. You have to actually spend the time to make the entire thing and then show him when you push the button it doesn't work. That was a persistent issue. The team lost four months on that, a lot of manpower and hours, proving that, yes, in fact that doesn't work.”

Another source added “There are no compromises with Roberts or reasoning with him once he has made up his mind about something.”

The final example: Star Citizen’s camera system. Roberts has pushed for players to be able to switch between first and third-person perspectives at any time in play. A source explained that “[Developers on the team] who had worked on first-person shooters in the past said 'You can craft a first-person/third-person experience, but it will be a really shitty experience.'

“You certainly can't create a first-person shooter and a third-person shooter and imagine that the sight lines and the speed that a character can travel are going to work out for both styles of gameplay,” the source continued. “I don't want to say there hasn't been one, but there's a reason why no one has really tried attempting it since the '90s. I think that because technology had progressed so far beyond what Chris had done previously, his 'I want what I want' was in direct conflict with the quality that he was expecting.”

The thing about these three examples is that Roberts says his team has managed to make them work. The high-fidelity character models, the complex inventory system, the first and third-person camera: they all appear to be on show in the alpha. In the Social Module you can see the quality of the character models; dress your character, layering the clothes over one another; and make use of the first and third-person camera. We’ve still not seen whether the game can handle all these complex and graphically demanding systems on a live, well-populated server, but the people who told me it couldn’t be done haven’t been proven right yet, either.

starcitizen16

“So, that irritates me,” Roberts said when I asked him about these particular events. (He concedes that they all took place.) “That is someone who has their preconceived notions and will settle for something that isn't good enough. The first and the third-person, there were a lot of people on the team who said that you cannot do it. They said it wasn't possible. Illfonic [the contractor building Star Marine] said the same thing. The same with the character stuff: I said we need to do it. And the inventory… it happens right now in [version] 2.4. You can put jackets and trousers on, and caps. There's layering with the armour. So everything that that person told you that couldn't happen, it's all in the game now and it's all at the quality [I asked for]. We've got as good or better than what 1886 has.

“That is an example of people saying 'No, I can't do it' and fighting that corner. They are the people I don't get on with and they are the people who end up not being at the company… If I was going to be clichéd, if it was an American person I'd say I want an Ameri-can, not an Ameri-can't. I want someone who tries [...] because we absolutely can do it. We’re trying to do something that has the fidelity that you see in The Order, or has the fidelity you see in a first-person shooter but has multiplayer online and this huge universe and I absolutely, to the very fibre of my being, know it can happen.

“It's not easy. I think the difference between a great game and an okay game is people that try to do the hard things. It's the JFK speech about going to the moon: We do it because it's hard, not because it's easy. That's what I expect from the team and if the pushback is 'Well, I didn't do this when I was working at this other company', then, well, here we're trying to push it a bit more and if you can't get on board then you're probably not right for the team. I'm sure that I'm not the only game creator who is in that situation occasionally.”

When you have a company structure like CIG, you make the directors irrelevant. You make it so that the employees start going directly to Chris Roberts. Why would you go to your director? His opinion doesn't matter. - CIG source

The final criticism that my sources directed at Roberts was his habit of directly contacting artists, designers, and programmers about their work, no matter how far down the chain, side-stepping his directors. “Imagine that with every piece of content that was being put in software [...] he is at some point in time going to have a comment,” a source said. “If you can imagine being on Facebook and every time you post something, Mark Zuckerberg would respond, you're going to have a very different reaction to what you post.”

This reportedly caused friction between junior staff and their leads and directors. “If you're a young artist or a young designer, a young engineer, anything that Chris commented on, if he said 'Hey, that's nice,' the developer would treat that as 'That's done, Chris likes it.' Well, that developer’s director would be like 'We've got a long way to go. Chris' quality bar is lower than mine.' There were a lot of conflicts. It happened across the board, in design, art, and engineering”.

starcitizen7

Another source said that this made the directors “irrelevant.”

“When you have a company structure like CIG where you have the person at the the top who at any moment will go down to the lowest level employee and sidetrack them completely from their task, and also undermine the director... in that case, you make the director irrelevant. You make it so that the employees start going directly to Chris Roberts. Why would you go to your director? His opinion doesn't matter.”

“That was probably the most ridiculous project I've ever worked on,” a source said. “I've got little tolerance for poor leadership and management. I think they're two of the pillars that holds a team up. It doesn't matter what industry you're looking at. If you've poor leadership anywhere you will find unhappy people. I think that at CIG you're going to take your cues from the person who is running the ship and if the ship is not doing very well [...] it creates problems from the top down.

“I think he got too much money too fast, that he wasn't held accountable for his words, actions, or things he was supposed to deliver, and he turned the project into a runaway idea. I think that if there were people around him to take the CEO duty off his shoulders, or he had more trust in his directors and didn't micromanage them as much, that he would be seen as a much easier person to work with.”

starcitizen6

In response to these accusations of micromanagement, Roberts happily admitted that he checks a lot of individual art that goes into Star Citizen. “There's a certain OCD about checking everything because I want everything to look good. It's also a matter of getting [the message] to the team, what I like and what my taste is, and then you build up trust.

“I'm pretty visual, I have a background in art, so I like to see stuff in development and then I feel like I can give feedback to my directors. Most the time I will say that something looks awesome… [but] it’s really my call if I want a redesign of the ships. It's not an art director's. [Likewise] If I feel like that an asset doesn't need a redesign then it shouldn't have a redesign.

“All the art directors I have right now I'm very happy with, and I don't feel like I'm getting on their case to make sure it's happening the right way. I'm not doing massive micromanaging at all. Now, in the process we have, Paul [Jones, art director] lays it out and says 'Here's four shapes, for a ship shape form. Which one do you like the most?’ Then we discuss it and move on from there.”

There's a lot of Star Citizen that is incredibly impractical, while not a lot of it is impossible. With enough time and money and clever people, anything can be made, right? - CIG source

I asked Jones, Foundry 42’s art director, how this was working for him. “Having the right people and key hires, that's definitely made the machine start working,” he said. “That's always difficult with art, art is always subjective. As art directors, our role is to [...] get into Chris' head [and] provide Chris with what he wants. It's not all about ourselves. We still inject a lot of ourselves into it: we do a lot of the heavy lifting in terms of how [a ship] can land, where the hatch should be, [coming up with] a cool mechanism for the landing gear. Stuff that he shouldn't really need to worry about. We have all those conversations and if we're doing that job right, we present our idea and then he should just say 'Yeah, cool' or ‘Tweak this' and then he doesn't have to worry about it.

“We do regular check-ins with him on pretty much everything and once we have a good idea of what he's after we can start pushing it through the pipeline. That's the way it should be. He should be able to trust us to deliver what he needs.”

To his detractors, Roberts makes no apologies: “I'm pretty single-minded but I'm telling you that's what made Wing Commander the game it was. That's the only way I know how to do it. I'm not willing to compromise on some stuff and I'm pretty stubborn about it. Sometimes, for someone else that can be tough to deal with or frustrating but I don't really want to change. I'm not doing this to always make friends with everyone, I generally like to feel I'm affable but really I'm doing this to make something I can see in my head. I want other people to play it and I want to play it.”


Can Star Citizen be made?

I asked each of my sources if they thought Star Citizen could actually be made, knowing what they do, and there was no clear consensus. But there was clear agreement on ‘overscope’.

One source, for instance, said that “there's a lot of Star Citizen that is incredibly impractical, while not a lot of it is impossible. With enough time and money and clever people, anything can be made, right? I think it suffers from the same problem that has dogged all video game development since the beginning: overscope. There's not a video game ever made that's not had stuff cut from it, or dropped, or been redesigned because it turns out it was too big.

Star Citizen started from this small development targeted at doing this one specific thing with a specific set of technology, which was absolutely fine – and then it grew and grew. Rather than adapting to new technologies and approaches for the new scope they stayed with what they had, which slowed everything down. In the end, what they should have done was decide a figure after the Kickstarter and gone 'Right, we're going to $25 million', and if they had hit that, they should have gone 'We're done. This is the game and with $25 million we can make it in this time.'

“But instead they just let it grow and grow and grow. I know they’ve said 'we're not adding features anymore', but the feature set they've already got is so vast and unwieldy and huge and the tech they're trying to adapt is not supporting it. If it had infinite time and infinite money and everyone working on it had infinite patience then, yeah, at the end you'd probably see something and it would be pretty cool.”

If it happened then I would believe in God. -CIG source

Another source flat-out believed that Star Citizen could not be made. “Not what they've promised, absolutely not. If it happened then I would believe in God.” They added: “the biggest obstacle to finishing, if you were going to take out money and time, is the scope. They could make a smaller game. They could make a really nice dogfighting shooter in space and it would probably do really well. That's the general attitude of most people in the industry.”

Another source said that if the team had focused on a single module first then it “could have been done a year ago”, and “then CIG could have built on top of it and rolled out updates.”

“Stop trying to give your audience every part of the game and calling that beta complete,” the source continued. “Say 'Hey, listen, we're going to give you a tiny bit of this thing here and a tiny thing of this here. We're going to polish these things up to a beta state, roll those out, monetise it, and now we're not relying on fake items to keep the game afloat.' We'd actually be keeping the game afloat with real assets.”

To Star Citizen’s detractors, Todd Papy, design director at the Frankfurt studio, had this to say: “I'm a firm believer of 'You tell me I can't do that and I will prove you wrong'. That to me is the mindset you need to have at this company. We have the guys here that believe in the game; they wouldn't be here if they didn't. They understand the challenges ahead of them and we have our ups and downs.”

starcitizen20

"Where we're at now, I'm much happier," said Chris Roberts, posed the same question. "I feel more in control of the beast. At the beginning it's like you're on the back of a bucking bronco and you're trying to stay on while getting it wrangled in the direction you want it to. Now I feel good. I know there's loads of work to be done but I feel like we have an organisation that can deliver it; maybe it won't deliver it as fast as everyone wants but that's just the nature of it."

Despite everything, everyone I spoke to for this article praised the talent of their colleagues. My sources were agreed that they were working with some of the best developers they’ve come across in the industry.

One person in particular put it best: “Chris Roberts is surrounded by some incredibly talented people whom he pushes to the absolute limit, then demands that they stay there. There were some absolutely amazing things coming out of Star Citizen. Enough to fuel several AAA games. If Star Citizen does end up hitting the goals that Roberts has planned (by beating his horses to death to get to the finish line), there is no doubt in my mind it will be amazing.

“If Star Citizen does end up hitting the goals that Roberts has planned (by beating his horses to death to get to the finish line), there is no doubt in my mind it will be amazing.” - CIG source

“Will it be fun to play? Not sure. Will it be an amazing tech and beautiful art demo? Absolutely.”

Another thing was clear from conversations with Star Citizen’s current staff, people who have either survived these tumultuous years or were not there to experience them: they still believe that Star Citizen is possible, and they are working hard on it. But no game can be made on belief alone. As of right now, there is no Star Citizen. There are several different live demos, but they are just that: demos. There is no game.

Star Citizen is the biggest crowdfunding success in history. The story of its development so far, rocky as it has been, is fascinating – and, as we’ve discovered, much more complex than it looks from outside. But sooner or later, Roberts and his team have to deliver a game, or all of this will be just a prelude to the most expensive failed development of all time.


Illustration by Mark Wynne. Research assistance from Rich Wordsworth and Christopher Bateman. This article is part of an ongoing series on Star Citizen from Kotaku UK, compiled from 7 months of reporting and investigation.

If you want to learn more about Star Citizen, you can read the rest of our in-depth series here.

Correction: The original article featured artwork by Santobell and Chmee that had not been properly credited. Those images have been removed.