Crunch is a cause of great harm to the passionate people working in game development. Be it the strain on mental and physical health, the steady onset of burnout, or a general lack of support, it has become increasingly clear that a reliance on crunch has damaged studio cultures across the industry.
Failbetter Games -- the UK developer behind the gothic horror RPG Sunless Skies, and one of our Best Places to Work last year -- invited us to its office in London to talk to about ethical game development. Sitting down with CEO Adam Myers and communications director Hannah Flynn, we discussed how Failbetter has tried to eradicate crunch -- which some believe to be an inevitable part of creating a game -- in its entirety.
"Making games is complex, so it's not unusual for there to be a thing that will tempt a responsible employee to stay late one evening, or doing weekend work" says Myers. "When that happens, or when someone is thinking in their mind that they want to do it, we try quite hard to ask why and look for the underlying causes to see if there's anything in our process that produces that outcome. Then we try to figure out how we can iterate our production processes to try and change that."
"We had contingencies so we could cope with things overrunning"
According to Myers, avoiding crunch altogether means never putting staff in predicaments that require overtime in the first place. At Failbetter, tasks are scheduled ahead of time in such a way that it is consistently releasing content without the need for drastic amounts of overtime.
"We had a roadmap for updates, and we kept a fairly close eye on whether the work was taking the amount of time that it was expected to take," Myers says. "We had contingencies so we could cope with things overrunning, and they more commonly overrun than underrun in game development.
"In every area of our work -- like design or programming -- we would have a block of time of some number of days that were just a contingency reserved in the schedule. Sometimes that would all get used because somebody was sick or tasks overrun, and sometimes we would be able to do cool optional things, or bring forward work to get ahead for a future update."
This attitude towards planning eased a lot of the pressure that emerges during a looming deadline, which in turn allowed for a realistic pace of work.
"A lot of it was managing workload, stress, and making sure people were getting accurate estimates," Myers continues. "There's a sort of cognitive bias when you're thinking of your own work, where you create an estimate based on the best case scenario. It's called the 'planning fallacy,' and guarding against that so even when people give an estimate which theoretically means they can finish the work in a sprint, they've had the opportunity to say how long they really think it's going to take."
This process didn't become perfect overnight -- and isn't perfect even now, after the release of Sunless Skies. Rather, it was the product of critical self-iteration. Myers gave an example from early in Sunless Skies' development, in which an oversight in the schedule resulted in added pressure for QA.
"Delaying was what we chose rather than asking people to do overtime"
"We had this QA process for quite a significant portion of the Early Access development for Sunless Skies where we worked in two-week sprints. And the last sprint before an update would be a QA sprint, where we would have a bunch of focused time on bug fixing. We found that it was fairly good for imposing discipline on us, making sure we didn't slip in new features too close to the release day that might introduce bugs we wouldn't catch and make us delay it. But it put a fair amount of pressure on our QA tester because their workload was very focused on that sprint.
"The one time we had to delay a release was because a crucial bug was found late in the QA sprint due to its discovery being impossible until other bugs had been fixed. Delaying it was what we chose rather than asking people to do overtime, because it was a tricky one to diagnose and fix."
Following this, Failbetter took another look at its QA process, making changes that allowed for its tester to search for bugs earlier. "This has a much lower risk, and is far less stressful for our QA tester, who is very, very diligent and quite hard to stop from working late if she's worried she might otherwise miss a bug," Myers says.
This acceptance of delays and setbacks as a natural part of making games is best exemplified by Sunless Skies' full release being pushed back from September 2018 to January 2019.
"As we were getting closer to a complete game it became clear to us that there were four fairly substantial chunks of work that we really wanted to do, that would make the game better but weren't within the scope of the original plan," Myers says. "At no point did we look at that and say, 'Oh, maybe we'll do that by working some weekends.' That was never even in our minds. It's just a question of [deciding whether] we release the game with the planned scope in September, or do we delay it long enough to do all the extra stuff that would make it much better than it otherwise would be."
When a member of staff at Failbetter wants to do some overtime, lengths are taken to find out if there's an alternative.
"If you go up to someone and say, 'Hey, don't do overtime,' they actually feel more stressed"
"I found that if you go up to someone and say, 'Hey, don't do overtime,' they actually feel more stressed. Because then they're worried about how they can somehow achieve the things they want to do in their limited working hours, or are feeling unhappy about the quality of work that's going out.
"What I try to do instead with the people I line manage is to communicate with them, where they tell me as soon as they think they might be at risk of wanting to do overtime on something. Then we have a conversation to figure out why and try to come up with alternatives.
"Some examples of things we've done in the past [include]... switching someone's working hours, because it turns out it's not that they need to work more hours, but they are blocked by something. So they can work a Saturday and take the Monday off while still only working their 35 weekly hours. Or we can delay something, or we can cut something. There are often ways that, after talking about it, everyone agrees are better than doing overtime."
The major factor that allows Myers to conduct development in this manner is the self sufficiency of Failbetter. Flynn notes: "We are fully indie and set all of our own deadlines. There is a very rare occasion when we would be working to an externally set deadline, or a deadline where we didn't feel we had some sort of control over. We don't work with a publisher, and we don't work on demos for E3 -- yet."
Myers adds: "We are still majority owned by employees. What they care about most is generally having a really good environment to collaborate in and make games that they are excited about. So there is no pressure from corporate overlords."
It begs the question of whether the type of development fostered at Failbetter Games would be feasible at a AAA studio. While Failbetter is making the most of the freedom self-publishing provides, pushing back a release date becomes more complicated with investors and publishers.
"I don't think there's anything about the production constraints involved in making a game with 300 people that means you can't do this," Myers says. "I think one challenge they have is that there are certain expectations for things like visual fidelity and size of the game world for a AAA game. Also, there's games-as-a-service, where people keep playing for months and months after the release, which just means there is loads and loads of work they need to do.
"To make [AAA games] in ways that are financially competitive with other studios that are exploiting their workers, I can imagine that it must be very tempting to take advantage of things like the lax US labour laws. But I truly believe that, aside from those ideological considerations, games can be done within the working weeks even with that level of project complexity."