How DICE Does It

Karl Magnus Troedsson, general manager at DICE in Stockholm, has been with the company for 11 years now. He began as a producer on the RalliSport Challenge racing series, which was work-for-hire, published by Microsoft prior to the studio's acquisition by EA. His first Battlefield title was the original Bad Company.

In this extensive interview, Troedsson speaks about what the team has learned over the years of creating the successful Battlefield franchise and how that then is fed from the creative core of the studio out to other developers. He also speaks about how the company views its staff -- both from the standpoint of how to best utilize them on game projects to the culture of the company and why that's so crucial.

Troedsson tackles the topic of innovation, saying that "sometimes, I think we get too much crap for not being innovative." He explains how the studio fosters innovation, where new ideas come from, and how best to integrate them into an overarching, popular franchise like Battlefield.

In your GDC Europe talk, you talked about the "everything and the kitchen sink" approach that Battlefield originally had. How did you make a determination, as a studio, to move to focusing directly on what was important for the series?

Karl Magnus Troedsson: Well, one thing is it comes out of having a long-lasting franchise like Battlefield. Once you've done the first push to get everything into the kitchen sink, you actually have it in the sink, and then you can continue building on it. So the first iteration was actually the hardest.

And then, if you're smart about it, you always take what you have and then you add something new, you polish some, and some of it you actually just keep the way it is. So you always evolve the actual game from the one to the next one.

So from this point on, I would just say that it's come from the fact that we've realized that quality isn't always the same as just pushing everything we have into the game. And quality is not the same as just having super innovative features; quality needs to come from both.

You need to have innovation in the form of new features, or something you have changed that is really innovative -- but also just spending time on polishing it or, like, owning it. Because otherwise, you can have the best feature that is new or cool or whatever, but if it's buggy, or it doesn't work as intended or whatever, then it doesn't matter.

So it's a balance of everything, in the end.

KMT: Yes.

When you say, "spend time polishing it", how do you plan for that in the production cycle?

KMT: Well, the old production cycles like, "oh, we hit alpha", and then you start polishing -- that doesn't work anymore. Because first of all, alpha is a very fluid state, I would say. Every piece of content should be in the game, and you should actually be able to ship it; that's how some people define it.

But what we do inside the studio is that we set hard deadlines before the alpha -- sometimes we call it a "pre-alpha", and that is when the whole game is complete with everything in it, where we say, "every piece of content, every feature," etcetera.

And that's part of the development phase that we just call "polish". And it's hard, because usually you just want more in there all the time. So at that time you just go, "Oh, we'll push it a little further." We cram that polishing phase, and make it smaller and smaller, but that's when you shoot yourself in the foot. And if you keep a serious amount of time where you just polish the game, that's when you can get it up to quality.

Does it vary from project to project, or is there a rule of thumb in terms of how long you think you need?

KMT: It's a bit varying, but at the end -- where all pipelines, all engines, all technology, everything is working perfect, you have the entire game -- you can look at it, you can play it, you can feel it, you can sense it, listen to it, look at it. Then you can do a lot of progress in just a week.

But we're talking months, here. On a big project like Battlefield 3 or something, you want like maybe one to two months to really polish the game, and that's before you start fixing the bugs. So then you hit some kind of an alpha, or whatever, and then you start crunching all the bugs as well.

Are we talking about things like modifying like timings of, say, shots or damage? What are we talking about?

KMT: Well, it depends quite a bit, actually, when it comes to comparing multiplayer and single player. Because multiplayer, which is in our DNA mostly, it's been an iterative process all the time. From day one, we are playtesting the game. Every day there's a new playtest, a new mode, or a new map or something. People give feedback, the multiplayer team goes back in, and they change and update. And that could be, like you said, updating the weapons or ranges -- the small nitty-gritty details, these kinds of things. So that happens all the time, just like an iterative feedback loop, almost.

But another example is on the single player side, where it's more like, build the whole thing and then you take a serious look at, "Okay, so I'm on this field right now, I can see troops running down there -- how can we make this more spectacular?" or "Is it actually too much going on right now? Should we just remove the troops from there, so you have a quiet moment, so you can actually assess the situation and be a bit more tactical? And then we throw everything we have at you when you get down the hill," or something like that.

So you actually move things around -- you change the scripting of the mission, you change where you place your AI, and where you want like, "Oh, the cutscene shouldn't kick in here, let's put it further away," or something. Or even down to, a lot on the polishing on the art side is like, "Oh, we want the sun over there instead, so you get the flare through the trees," or whatever. It might be, "We want more bushes here, more foliage," etcetera.

Is it mostly emotional, creative decisions, or is it also data-driven? Or does it vary depending on what part of the game you're talking about?

KMT: Single player is very emotional, I would say. The team at home, they have their... almost like a powwow, they just lock themselves in a room and they look at the mission, and listen to other kinds of feedback coming from the outside, as well. It's very emotional, I would say, like trying to get the experience to a higher level.

But in multiplayer, it's much more data-driven. After the playtest, all the multiplayer designers, they can get heat maps, and they can see where did people get shot on the map, how many kills? Is it the same team that always wins on this map, and why is that so? Is it because they have a better home base or something like that, better location, or etcetera?

So multiplayer is much more data-driven than single player, I would argue. However, single player is definitely getting there as well. Similar telemetry data systems that we have on multiplayer, we are applying on single player as well, so you can see a playthrough, you can see all kinds of people get to Mission 7, and then there's a huge difficulty spike here. "Okay, we need to do something about that as well."

Do you have formalized playtests? Does everyone in the studio play the games, or is it more just who has time?

KMT: Who has time. It's sent out to the entire studio. But it's definitely a challenge, because we know that playtest means quality if you do it the right way, with iteration and feedback, etcetera. But especially now, when we're getting more and more players into matches as well, and it starts to get harder and harder... We have test locations within EA and Europe, of course, that fill up the servers if we don't have enough people.

Is there too much data? Can you balance a game into blandness using that kind of data?

KMT: I think you can. This is one of the reasons why I think that people still like Battlefield, is because we have a recipe that we stick to pretty closely. We change things all the time, both in a new title, but also post-launch, where we feel something is wrong.

But the one thing we have learned is that when you do these kinds of changes, you do it in smaller steps. In the beginning, when we didn't have that much experience, we went from 1 to 10, and then from 10 to 3, and from 3 to 5, and then, okay, maybe it was four and a half we should've been at. But over the years, we're getting more and more experience with it.

So I think it's important for those people to find their niche of the kind of games, and then stay true to that. For instance, to your point, if you take all the weapons and you balance them so it doesn't really matter which weapon you use, then you'd probably reach some kind of a blandness. At the same time, you have the challenge -- at least for us on Battlefield -- that we don't want any super weapon; we don't want any weapon to be substantially better than the other ones.

So we look a lot at, like sniper rifles, they should have about this range, about this damage, etcetera, and then we work with attributes. So this is a bolt action, it has a slow reload; this is a semi-auto, then you need lower damage, because you can get more shots fired quickly, etcetera. And you look at those features around it -- the weapon, the gadget, or the vehicle.

There must be tremendous amounts of institutional knowledge in your design team, at the point where you can jump off with a reasonable idea. You're not starting from zero.

KMT: No, it's incredible. It's one of the things that really fascinates me when, [creative director] Lars Gustavsson -- I usually call him the granddad of Battlefield, I don't know if he likes that or not -- but he's amazing. He's been with us since Codename Eagle. He was part of doing that as well. And we've actually formalized this over the years, now. So we have a Battlefield core group where all these designers and a lot of that knowledge actually sits, but a lot of it is in our DNA, in the walls in the studio.

How do people in that core group transfer that knowledge to the rest of the company?

KMT: Well, we have seen this first and foremost when we put new teams on making a PDLC pack or something like that. And a lot comes from participating in the playtest, and then just sitting down and have discussions saying, "Okay, what you do in here, perhaps, with the shotguns, maybe you shouldn't do that. Look at the core product, or look at the product we did before. Shotguns should have this kind of range, not what you're suggesting here." A lot of it is very much verbal, and playing the game together, and putting people physically to sit next to each other is also very important; physical space is very important when you develop games, actually.

How do you organize your physical space?

KMT: Well, we care quite a bit about our office, and I don't know if you've been there...

No, I haven't.

KMT: You'll really love our office, and we're spending a lot of time taking care of it. Both in keeping people happy in the studio, but also because of what we'll be talking about here. We have semi-open areas. We have a room where maybe you can have 12, or another room where you can have 24 people, and then we have smaller rooms. And depending on what kind of pod we have -- because we do organize the teams into like, you can have a narrative pod, or maybe a bigger single player pod, and then you can have a multiplayer group, and these kinds of things.

And then we try to put cross-functionality people in there, so you have programmers, and designers, and artists, everyone sitting together. It doesn't work having all the artists in one corner, and all the programmers in one corner -- that's a recipe for disaster.

Something that I think interested our audience is when we spoke to Patrick Soderlund about how Battlefield 1943 came out of the labs project in the studio. Have you been involved in that?

KMT: Yeah, I was the EP at the time, for the franchise.

So what do you think of that? First of all, in terms of both the intelligent use of resources within the studio, but also in terms of keeping people satisfied, interested, engaged?

KMT: Well, the coolest part for me about Battlefield 1943 was that the whole idea started with, at the time, we had the engine -- we had Frostbite 1. We were making Bad Company, and looking at those products.

And then we just thought... There was me and a couple of others sitting in a room, and over a couple of weeks, the ideas were tossed around like, "Oh, you remember 1942? You remember World War II? Wouldn't it be cool if in this brand new engine -- and the capabilities that we now have -- to have like a spitfire just skimming the surfaces of the tree foliage, and just dropping bombs on a Tiger tank?" Or something like that. It was like going back to old memory lane.

And we just felt like, "this is so good that we need to do something about it." And we have those ideas -- a lot of those, all the time -- but making these big games as we are, it can be quite hard to actually realize them, because so many people are working on the products that it's usually better to take those people and put them into the big project as well, to make that a little better.

At this point in time, like I mentioned, we were just between projects, so we had a lot of free people in the studio. So we just started. "Are you free? Do you want to do something?" "Yeah, sure!" "Okay, go talk to that producer over there, talk to Patrick Liu," who was the producer that ran it. "He needs people."

So people came and went -- there were very few people that were actually on the product all the way from the beginning to the end. From the development standpoint of the process, etcetera, it was an innovative way for us to work. But it was also a challenge, because not having that continuity from people in the team made it much harder for him to actually get the product together.

What did you learn from doing this experiment?

KMT: That we should probably experiment more. That's probably the most important one.

Have you thought about doing more experiments?

KMT: We have since then, but none of them have really seen the light of day.

But I mean I'm assuming you still learn from them...

KMT: Oh yeah. It's like I mentioned in my GDC talk, that we really take things -- like the movement, or part of the movement, from Mirror's Edge, and put into BF3, for instance. That's a typical example. And there are those examples that have gone into big products that people don't really know about, but they actually came from smaller, other, more innovative test experiments inside of DICE.

If you talk about it that way, it's obvious the benefit it could have, but you still need management buy-in to make it.

KMT: Oh, yeah. In the background of everything, we run a business, of course. But with the success we have had, there are opportunities. It's quite easy for us to explain to upper management that, "We want to do this. We want to try this out."

And EA is actually very prone to try out new things, and I have to say that sometimes I think we get too much crap for not being innovative. We do release new IPs, and we do take care of IPs that have been out there. And perhaps not so successful, but we try to get new things out there as well.

Now, that might sound strange coming from me, working on Battlefield. We've been around for 10 years, and we just keep doing more of the same, but being innovative in that space. But EA is quite easy to convince -- for us, at least -- to do more experiments, etcetera.

It actually comes down more to our own discipline, like I mentioned: "Okay, we have five guys that are free over here. Should we do an experiment with them or, you know what, they actually need more help over there, to figure out that new feature that they're working on." What do we do? We usually put them on the project. So it takes discipline, as well.

It also takes, when you look at the project, you need to be mentally prepared to have some slack in your production -- the amount of people who are actually working on something.

Okay, so some flexibility in the staffing.

KMT: Yeah, exactly.

When you talk about innovation at a studio that works on these kinds of games, how do you define that?

KMT: It's tricky. Innovation is something that actually changes something for the better, renews something -- and I think we do that all the time. But some of them are big and some of them are small, and some people disagree, saying, "That's not an innovation!" and it's like, "No, maybe not for you, but for the people playing the game it's a big thing."

But it comes down to, the longer a product has been running, the more you need to challenge yourself to actually do something innovative. At the same time, as I mentioned before, as well, it needs to be the right level of innovation. You need a couple of them and then you feel like, "This is really, really good", and, "This is going to be interesting", perhaps, for both our core audience, but also hopefully attract new players into being interested in playing the franchise.

How do you identify those targets?

KMT: That's, again, quite emotional actually, and it goes back to knowledge of our products and then feeling like, "Is this the right thing for us to move forward?" That's one big source of it. And then we seek validation -- we could test it with a marketing group, or something like that. But we also keep very close tabs on what people say in the "interwebs" -- what people actually are talking about out there.

We spend a lot of time just reading up on the forums and the comments, etcetera, and that can be pretty harsh; it can be a bit of like beating yourself on the back, going out there. But even though there's a lot of negativity sometimes, every now and then you'll just find gems like, "That's so good. What he just said right there is so good; we need to do something", and either fix what he's talking about, or add what she brings up, or whatever it might be.

What is your most important source of feedback?

KMT: First and foremost, it's ourselves. And this might sound like an ego boost, but it is, because we've been with the franchise for so long. And also actually it comes down to the fact that we still build Battlefield because it's the game that we want to play ourselves.

That's where 1942 came from, and Lars Gustavsson, and Patrick Soderlund, and these people that were in and running Refraction Games at the time, which was purchased by DICE. You know, from the stories I've heard, they built the game that they wanted to play themselves, and that is still very obvious.

A lot of people, myself included, I still play our games; I'm still grinding away in BF3 because I really love it. And I'm not saying that because I'm shining light on my own product here -- I truly love playing the multiplayer.

So first and foremost we look to ourselves and say, "Okay, what is it that we want to do? What do we aspire to make better?" And in combination with that, we're quite competitive, and we always drive ourselves pretty hard. And as I mentioned, we very rarely sit back and go, "Oh, it's perfect; we don't have to do anything!" We always want to strive to make things better, and that drive -- together with the knowledge of the product -- drives us forward to be like, "Okay, we should do this the next time, and probably do this the next time as well."

Having worked on both work for hire projects and also, what you're describing as, to an extent, a passion project, obviously you're going to feel more strongly about the one you feel the most passionate about. But can you talk about that contrast?

KMT: Well, it was interesting because this man [Troedsson gestures to indicate the nearby] Patrick Bach, he was my creative producer when we did RalliSport -- I was the producer, and he was basically the lead creative kind of guy. We were very passionate about that product as well.

I mean, it wasn't our idea from the beginning, and we were only part of it the last month of the first one, but we still felt very passionate about the second one because... you make it your own, right? You take it on, you start talking to, in this case, the producer in Microsoft. We had a very good cooperation there, I would say. They had some very good people on their end. So it was quite easy to be passionate about that, still.

I think it comes down to that you have to ask yourself why you're working on the project -- even if it's a work for hire. Is it just because you need to make money for the studio? Okay, then you probably have a problem. But I mean, even back then, DICE didn't really do that -- we were really passionate about making racing games. DICE had done several racing games before the rally games. So we were pretty passionate about that in the studio, as well.

But every game developer needs to soul-search inside for why they're doing something, and if it starts to feel wrong, and actually that passion goes away -- at least for us, I can only speak for us, but it wouldn't work for us. We could not work on something we're not passionate about.

At the same time, it's not always funny, like I also brought up; sometimes life can be pretty hard. Sometimes you need to help out another team, or you need to do something else that you might consider being a bit of a distraction. But you know what? You need to do that as well, because it's going to make all of us better, and make everything move forward.

It sounds like part of what is important to DICE is the internal culture.

KMT: It's super strong. Culture within DICE is something we, and the management group, talk about a lot, and we try to spend as much time as possible to nurture it and take care of it. And it's not overly complicated -- it's not like you hire a branding expert and you have them tell you what to do. I mean, sure, you can do that as well, but for us it's much more... As I said, we're still nerds. Sure I have a shirt [Ed. note: here, Troedsson is referring to the fact that he's wearing a fashionable, collared shirt to our interview] but I'm still a nerd; I still play games. That's why I'm in the business.

But having a great party, for instance, a good release party, everyone can like, "Yeah, cheers!" "Nice work!" Everyone slaps their back, etcetera. That's very big for us, when it comes to just meeting, and talking, and communicating, and having a good time together.

And then that's one example, but then you have a daily, how we approach each other in the office, etcetera. Culture goes into everything we do -- we're competitive as hell, but we're also very eager to make sure that people are happy, and not just from the management and down. But people are overall very friendly, and it's a pretty nice place to work, from that perspective as well -- as long as you are passionate and eager to make better.

You wouldn't be satisfied, and you wouldn't be happy at DICE, if you came there and were like, "Oh, this is a 9-to-5 office; I'll just do my thing." That would not work. Then people will start [being] like, "Woah, you're misaligned with the culture."

It's obvious that it's about the right people as much as it is about the right environment.

KMT: Absolutely. I would say this: the right people with the right mindset.

Those are the two ingredients.

KMT: Yeah.

How many people are at the studio now?

KMT: Well, it depends a bit on how you slice the different business units, but I would say we're about 270, 280 at the moment. We're growing rapidly right now.

Is it difficult to grow rapidly and also protect the culture?

KMT: Well, it's not difficult, but you need to be aware of it. You need to be aware that when you get a lot of new people into the studio, that's going to mix up culture, values, these kinds of things. And you need to allow things to change a bit and mold, because you can't just say, "This is the way it is -- conform. Be ready to be assimilated."

It needs to be more like, things change when more new people come in, but at the same time you shouldn't change the core values of the company, or something, unless people actually, as a full unit, say that "This doesn't work for us."

You need to say that "This is what we stand for as a company, this is how we do things here at DICE," and hopefully if you're the appropriate person we're talking about, if the hire was a good one, they are just going to say, "That's awesome; I want to be part of that."