Thursday, September 24, 2015

Destiny one year on: Bungie's 12 months at the final frontier of gaming

On 9 September 2014, in a converted multiplex cinema in Bellevue Washington, a team of over 200 people launched a brand new kind of video game. They had spent four years preparing for this moment; among their ranks were some of the most experienced software engineers in the industry. But they had no idea what would happen next.

Months later, David “Deej” Dague, the community manager at Bungie Software looks puzzled when asked about that night. “I’m not sure I have any memories of the first two weeks after launch,” he says, sitting in the darkened entrance hall to the company’s vast office, surrounded by cabinets hosting dozens of awards. “All I know is, the game was pretty stable.”

That game, of course, was Destiny, an online sci-fi adventure set in Earth’s distant future. Combining the fast-paced action of a first-person shooter (FPS) with the in-depth progression systems of a massively multiplayer role-playing game (MMORPG), the title effectively sought to unite two very different audiences: gamers who just wanted to shoot at stuff and gamers who liked to build characters, explore worlds and level up to access increasingly potent equipment.

Certainly, these two genres have been borrowing ideas from each other for years. Shooters such as Call of Duty and Borderlands have taken the exhaustive character progression and loot-collecting concepts of the role-playing sector, while RPGs like Mass Effect and Fallout have in turn become more dynamic and action-orientated. But with Destiny, Bungie wanted to bring all of this into a seamless experience where players could easily migrate from hit-and-run “strike” missions to demanding co-operative raids and arena-based competitive shoot-outs. And it wanted all this to happen online, without any sort of server partitioning between different play types.

The preparations were exhaustive. Years before launch, Bungie built a large in-house user testing laboratory where it hosted dozens of volunteer gamers, watching how they played, even monitoring their eye movements to help design the onscreen displays. “We knew Destiny would have a much more complex UI than any game we’d ever made before,” says head of user research John Hopson. “A lot of shooter players have never experienced mechanics like gear and talent points. If they weren’t using those elements during play-testing we needed to know why: was it because they’d seen them and didn’t care, or was it that they hadn’t noticed them at all. By analysing where they looked on screen we could say, OK, yes they read the help window and closed it – it just wasn’t telling them what they needed to know.”

On the technical side, there was a simple aim: avoid the sort of total infrastructure collapse that had plagued other big online releases like SimCity and Battlefield 4. For Destiny, a game selling itself as a new type of connected experience, that would be a disaster. Bungie built a state-of-the-art data operations centre (DOC), a control room filled with screens showing stats and metrics from the vast global server infrastructure that would run the game. The company employed a 24-hour hit squad of systems engineers, capable of reacting to server shutdowns and overloads whenever and wherever they happened.Reportedly, some of the staff had previously worked on the Large Hadron Collider project at Cern.

According to Bungie, over a million players hit the game on day one. The traffic was managed through a server bunker in Las Vegas, as well as rented data centres all over the world. It held up. However, what the studio quickly realised was, this was just the beginning. “The DOC did incredibly well to respond to that demand,” says Dague. “But the real learning happened afterwards, in the months that followed, when we were able to identify how players felt about the story arc, the narrative of their own character, the end game content, and about approaching the rank cap. We had to prove to players that we understand our own game, that we played it, that we knew who they were and what was important to them.”

Peter Parson, Bungie’s chief operating officer, concurs. “Certainly, on the positive side, people were playing the game; and the way they were playing – the core investment loop – was really smooth. But there were questions about the story ...”

Indeed, many players felt the story was fragmented and unclear. What was the vast Traveller satellite that was protecting humanity? What was the dark enemy that had followed it across the cosmos? Why were alien races attacking Earth? There were rumours a lot of content had been cut in the run up to launch so that the team could concentrate on other technical challenges. “We’d tried to do something different,” says Parsons. “The story wasn’t about a fictional character you’d created, it was about you and your journey as a player, your own personal narrative. But we definitely didn’t get that right.”

Another problem was the unclear character progression at the end of the story missions. Suddenly, players had to start collecting exotic armour that added a value named Light – but it wasn’t clear how to find it, or how this related to levelling up. “What we really did not do a good job of was sending players off to the end game,” says Parsons. “It wasn’t just the raids, it was starting to use the bounties, doing the high-level strikes – that’s something we learned a lot about. The community really helped, not just by providing feedback, but by assisting other players across that bridge.”


Bungie knew it had to react – but a video-game studio with over 200 staff is like an oil tanker: it doesn’t turn fast

Bungie knew it had to react – but a video-game studio with over 200 staff, in a huge variety of disciplines, is like an oil tanker: it doesn’t turn fast. To cope with the constant influx of feedback from players, the company totally restructured its development team. While small dedicated groups were tasked with designing each of the incoming expansion packs, a large new “live team” was formed – a group of 20 or so senior staff with a specific focus on listening to and implementing player feedback.

“They’re the operations team,” explains Parsons. “They manage and monitor the game around the clock to ensure we’re always up and running, and making the experience better. Although the group is usually about 20 people, they flex much larger when we’re getting ready to ship new content or need the support of extra engineers. We also allow the group to tap in to other specialists across the studio, providing everything from bug fixes to new content. We’re continually pulling people into that vortex, then pushing them back out. The fun part of that is even if someone’s working on something that’s far down the road, they can feel part of the process along the way.”

Alongside the Live team, Bungie has also organised a handful of its most senior team leaders into a special triage unit. This is effectively a sort of development tribunal: when bugs or gameplay problems are encountered, they decide what gets fixed first. “At any given point in time there are thousands of things in the world of Destiny that we’d like to make better but there are only so many resources,” says Parsons. “That team is constantly deciding what are we going to fix and what are we not going to fix – it has it’s fingers on everything in the game world. It’s all about managing development priorities, and they’re constantly changing. They’re free to bring in other team leaders to bring them up to speed. It’s not uncommon to get an invitation from triage to come up and explain what it is your trying to fix and why it shouldn’t be pushed to a later date.”

Hopson and his user research team soon became heavily involved in this process, analysing the live data and using it to guide decisions. “Mostly what we do is prioritise,” he says. “Very few things that fans suggest are bad ideas - it’s usually, that’s awesome we should do that, but is it more important than the 50 other things we wish we could get into the game? If three million people are having a problem then yes we should fix that now, but if it’s 500 people that need the fix? The design team have to set priorities.”

There have been occasions, for example, where a particular weapon has appeared over-powered, and the immediate goal is to work out how many people are being affected. One example involved the Vex Mythoclast, an exotic fusion rifle. “In its heyday, people were posting videos on youtube just completely dominating the crucible,” says Hopson. “So we asked, okay, how bad is this? How many people are actually running around with this thing? And we found that there were only 600 people with this gun. We didn’t have to panic and do a next day patch - we could leave it a little bit.”

“We also look at which weapons people actually use. There are some we thought would be really popular but weren’t. I mean, I don’t think anyone thought that the No Land Beyond sniper rifle would be super popular but we had a role for that weapon and people didn’t pick up on it in the way we planned. A lot of it is analysing the live data. We’re saying, okay, how are people actually playing? Destiny is a really complex game there are a lot of different ways to do it. Players have certainly surprised us in a bunch of ways - the loot cave being the most prominent example. We never thought people would just sit there doing that.”

But Bungie isn’t totally indulging players with however they want to play: it has a plan for them. As the game was being developed, the company worked out that player retention was going to depend on getting as many newcomers as possible to try all the different experiences from co-op strikes to single-player patrols to competitive multiplayer death matches. In his GDC talk on user research earlier this year Hopson, who has a PhD in behavioural psychology, talked about the buffet effect – a phenomenon in which people will eat more if they are presented with greater variety. Destiny, with its array of game types and events, is the digital equivalent. For the game to remain successful, goes the thinking, it needs ‘omnivores’ – players who’ll try a bit of everything.

Right now, Destiny does allow people to have preferences for co-op gameplay (strikes, raids, etc) or competitive (The crucible) – but to unlock the best new kit, they have to occasionally switch over. To make sure no one was being alienated by this structure, Hopson formulated 15 different types of Destiny participant, based on phases of the player lifecycle and different types of preferred engagement. The team then began to compare proposed new features against those profiles to work out how they’d be affected.

“You do a bunch of analysis,” says Hopson. “You say, okay, how does each player type engage with the game and what can we do for them? We’ve actually been amazingly successful in creating omnivores – pretty much anyone who gets to level 20 or beyond is playing like an omnivore. Even when we talk about someone as being a raider – raiding still ends up being only 40% of their time so they’re spending a lot of time in patrol or PvsP - even when we talk about PvP people, they’re spending a lot of time in raids to get the gear they need to do well in the Crucible. So the game design is working amazingly well in terms of encouraging omnivores.”

According to Parsons, Bungie also looked to seduce shooter fanatics by making Destiny’s more in-depth, grind-heavy gameplay work within their usual timeframes. The team knew that people who like to play Call of Duty online for an hour, level up once or twice and then log off, may be intimidated by an RPG that was going to demand hours of their time in order to complete epic missions and progress. So the team designed the short Strike and Patrol missions, which both provide plenty of loot potential, so that the game could be played for an hour in an evening. As Parsons explains, “If you provide people with multiple ways to engage with the universe – even if the core mechanic is just building and growing your guardian - they can have a rewarding experience. People say that for a console game Destiny is very grindy, but you can get a reward in less than an hour by knocking off a couple of bounties.”

Destiny also makes clever use of what psychologists call variable-ratio reinforcement schedules. Whenever a quest is completed, there’s a chance that a boss will drop valuable exotic armour or weapons, but it’s never certain. Indeed, in the current form of Destiny, every single piece of gear you’ve earned by the time you run your first raid will have come though a double stack of randomised rewards: first, you have repeatedly killed high-level bosses waiting for them to drop a legendary engram (random chance), then you will have repeatedly decrypted that engram waiting for an item of a high-enough level to take on the raid; or you will have taken a low-level legendary item, and then used high-level rare items (which are also dropped randomly) to “infuse” it and boost its level.

However complex the system, the hope is always that the next drop will be the big one – it’s this variable schedule that compels players. In this sense, Destiny has been darkly compared to a slot machine, and the analogy is pretty accurate: both tap into that “the next one is the big one” compulsion loop. At the same time Bungie has learned from the ‘appointment gaming’ phenomenon we saw in the social space. Facebook titles like Farmville succeeded because they were designed to allow players to keep dropping in and finding new stuff to do. Destiny mimics this with its daily and weekly challenges, as well as having vendor characters who’ll often arrive unexpectedly in the world with rare goods to sell.

“We planned these calendar rituals,” says Parsons. “If you type ‘where is’ into a Google search, [the name of a vendor] Xur comes up as the third or fourth search. These rituals of ‘I can’t wait for the next Iron Banner tournament’ or ‘what will Xur have when he comes back?’, even the more regular public events – people look forward to the surprises.”

What the team seems very keen to stress however, is that everyone at Bungie is a player of Destiny, not just a developer. “I see people playing this game every single night,” says Dague. “They are as deeply invested as the people who bought the product – that’s a wonderful thing. We have our own email debates that rage over our server about what Destiny is or what it should be or what we need to do to fully realise the potential. Everybody at Bungie has an opinion about Destiny – if they don’t they shouldn’t be here.”

According to Parsons, the studio has its own elite team of Destiny players, called the Tiger Corps, who will dive straight into new updates and expansions as soon as they’re on the company intranet, and start testing them. “They’re on the frontline because they act and react as an even more sensitive and opinionated version of our larger community,” he says. “We actively challenge them to provide feedback and, man, they do not hold back.”


A year after launch, the latest expansion, The Taken King, is out and getting good reviews. People are saying that this is what Destiny should always have been. The progression system has been ironed out, it’s clearer, it’s more logical, and it’s easier to get into co-op gaming sessions. But the thing is, it’s taken a year of experience to get here. “I don’t think I can stress how much of a shock to the system this has been to Bungie,” says Hopson. “I mean, we used to take pride in the fact that our games didn’t require updates – it was lesser studios that did that. We were shipping these perfect shiny things that were perfectly balanced.

“Destiny is just not that kind of game and we really didn’t know what the reception was going to be. We’d worked really hard, we were proud of it, but last summer there was a lot of tension, a lot of worry. Is anyone going to play? The answer, clearly, was yes.”
Destiny forsaken: the three stages that Bungie lost players

Head of research John Hopson has spent the year analysing player data. Here he explains the points where people tended to leave the game during the first 12 months, and why.

Almost immediately: “The first place we lose people is right away. There’s a certain number of people who rent the game or borrow it from a friend - they play a few levels and it just wasn’t the experience they were looking for, that’s cool – not every game is for every player.

Level 18-21: “There are people who finish the story and never make that jump to the mid-game. They never quite realise that they need to be running strikes and upgrading their gear by collecting Light, they just don’t step into that section of the game.”

“We knew that was going to be a problem – we tried to make the transition as smooth as we could, but we did lose a few people there. It was a bad problem by our standards, it was a rough patch, but we were only losing something like 4% of the players – by the general standards of video games, that’s not a whole lot. We learned that we were too abrupt in that transition.”

No comments:

Post a Comment

GOOGLE