"Play is universal, even more universal than humanity"

Pioneer Raph Koster reflects on metrics, VR and the human need to play

Sitting in a meeting room with Raph Koster at GDC, ready to begin our interview, I'm worried we won't have anything left to talk about. We've just finished recording an extended podcast with Justin Ma of FTL developer Subset Games, and Raph has been holding forth with vigour, so I'm a little concerned about keeping the pace of conversation brisk for another 30 minutes.

I needn't have worried. Koster buzzes with enthusiasm, engaged and animated. I try to keep up.

We begin, somewhat inevitably, by talking about his seminal design textbook, 2004's A Theory of Fun. Lighthearted and self-illustrated, Koster's manual is still considered a vital work, despite the incredible changes which have been wrought upon the industry in the last 13 years. A ten year anniversary edition of the book proved just how relevant Theory of Fun still is, so I ask Koster if he thinks there are immutable aspects to the craft of game design - concepts which are so innate to human enjoyment and learning that they'll always be relevant. He answers with characteristic certainty.

"I think it's unquestionable. We work on rapidly shifting technology platforms, and those continue to change in a whole bunch of ways, but I actually think that game design moves at a glacial pace. There's a much greater awareness than there used to be of the fact that the kind of game design we do in digital games and in tabletop games, and what happens in sports, are really deeply interconnected.

"It was really nice to see a day of tabletop talks this year at GDC after a couple of years of tabletop exhibits. That's great - that feels like game design becoming aware of itself as a discipline, independent of the technology platform. I think the tech is going to come and go, and it's going to rewind. The games industry has thrived on tech rewinds. Mobile was an obvious one, Facebook was an obvious one, Flash games was an obvious one.

"Each of those basically forced people to throw away their tech blinders and refocus on gameplay. With those we get explosions - new markets, renaissances. We get new entrants, new voices, new companies. Whole new things get built. Then the tech gets better gradually, and it turns conservative and locks up again. It's going to happen again. I don't know what it's going to be this time. VR isn't a reset - it's more like an extrapolation forwards, VR is like a PlayStation 8 if you like: it's a line forward of evolution.

"That was actually something which was said to me by someone who was working on Facebook games: 'we're done, we've solved it. The metrics will tell us what to do, we don't need a designer any more.'"

"IoT platforms in general, and I think toys to life is a specialised segment of that, I think that's one potential place that causes a tech reset. I think AR is another potential. Pokemon Go is a great example of that rewind. The tech on that...Obviously the scalability is an incredible achievement, but the rest of it isn't really pushing tech, so I think there are several things on the horizon which could be that - I don't think any of them are quite at fruition yet. These are the times when the industry gets exciting - blue oceans open up and people explore new territory."

I sense there's more to discuss on VR, both from Koster's previous public musings on it and some of his phrasing in our podcast, but I'm a little surprised to hear him describe it as a 'forward line of evolution', having presumed he was sceptical about its future. I press him on where he sees the HMD timeline heading - does VR really offer the sort of unique opportunities which it promises?

"I think it comes down to the fact that there's basically a stacked set of constraints. You're absolutely right, the first is that it has to be something that you can't get in any other way, even in a reduced fashion. For example, there's an awful lot of VR chat stuff. The fact of the matter is that the most successful chat systems, on earth, are plain text. We've had 'better' tech for a really long time now and it doesn't matter, and that's because it doesn't really affect what people want to do, which is to communicate in as quick and unencumbered a fashion as possible.

"It's not sufficiently better. It doesn't give you a radically fresh new perspective. The term from economics is 'satisficing'. A shopper will settle for the thing which satisfies the need they have well enough, they don't necessarily need the best possible thing. So that's a constraint on VR - it has to offer a thing you can't get in any other way. The next constraint is - is that thing desirable enough that there is a market for it?

"Kinect might be a useful example here. There were unique things you could do with Kinect. Absolutely - it opened doors to things which could not be done in any other way. I can't think of a game which was radically improved by Kinect. Some of the things, like dance games, let you do things which you couldn't do in any other ways, but it essentially traded hands for feet. And in the end dance fans stuck with feet, which in hindsight seems pretty obvious...

"There were unique things you could do with Kinect. Absolutely - it opened doors to things which could not be done in any other way. I can't think of a game which was radically improved by Kinect."

"You look at it in hindsight and you realise that yes, it offered something new, but there wasn't actually a market for it. There are going to be things which VR can do which you can't get in any other way, but is there going to be an audience for that? I don't know.

"The third thing is going to be plain old boring, early adopter, cost of entry blah blah blah. VR is in the interesting situation where the versions of VR which offer something truly unique and maybe have something which many people would really like, are the ones which cost the most, require a 12x12 empty room and no pets to run between your legs and kill you. The ones which are easily available and cheap and so on have not found something unique which they can offer in a better way than anything else. So if VR cannot land on something like that, it'll be Kinect or the Wii Balance Board or the light gun. I think that's an entirely possible fate for it, but that doesn't mean that there can't be other incredible awesome stuff. It's not a knock on it, it just doesn't yet tick all the boxes to hang around for ever."

Koster is far from dismissing VR as a technology completely. Pointing out that games have driven large chunks of technological advancement in other areas like networking and graphics, he says that there are clear avenues of development where VR can offer exactly the kind of unique and useful propositions he's just been discussing. They just might be in other industries.

"The thing is, it's so ridiculously easy to see a concrete, unquestionable application that just isn't games," he says. "That's fine. That's our role in the world - to play with things. We're playing with VR and it may turn out that it isn't a great plaything in the long run, but through play we will have found practical applications. That's great, that's our job."

Our talk shifts to broader issues, to the culture of play, of play as an inherent human value. I ask Koster if he sees much of a cultural divide between games designed for different markets, and whether that's a result of social differences. Acknowledging that different territories can exhibit some different preferences, he nonetheless believes that there's more common ground than separate territory.

"I don't know if they're that sharp," he says of cultural boundaries in play. "Play is universal, even more universal than humanity is. We've all seen puppies and kittens playing. But when I look at games, there are definitely interesting cultural markers. In the last few years I've spent a lot of time studying traditional board games. It's really interesting to see the Tæfl family of games from Scandinavia, which spread into the Celtic areas as well, you can see the cultural echoes. There is a clan, and a Jarl or King, the warband. That's baked into the game. Then when you look at Africa and you see the entire Mancala family - that's very clearly about communal survival and crop rotation and things like that.

" These are underlying problems which always manifest in life and they form the backbone of games."

"So you do see things which look like they echo culture, but you go a level deeper and you see resource allocation, projection of force, trajectory estimation, calculus - we see those things. So I actually think that this is a relatively short list, for one, and that they're universal. These are underlying problems which always manifest in life and they form the backbone of games. I think something like Tag is probably played everywhere, in some form. There are some games which are so elemental that they don't really need any explanation."

I wonder then, what he makes of the rise of metric design, of a core reliance on closely observed formulae and the hunt for perfect feedback loops, player manipulation and the psychological legerdemain of some free-to-play systems. Surely such formal systems are anathema to the universal language of delight which he describes? But Koster is far from opposed to metric analysis - as long as its accompanied by good design instincts.

"We were doing metrics before there was mobile, we were doing it in online games because we had to, to estimate things like server load," he explains - referencing his time at the helm of Ultima Online. "But we were doing much more detailed metrics on player behaviour too. I think metrics are best thought of as a really granular, high-speed method of playtesting, and game design has always used playtesting.

"The danger, when you get people who don't know game design but do know metrics, is that they conclude that they've 'solved game design' by reducing it to a certain set of metrics. That was actually something which was said to me by someone who was working on Facebook games: 'we're done, we've solved it. The metrics will tell us what to do, we don't need a designer any more.' It was ignorant, clueless. They met their deserving fate, eventually.

"Ideally, every designer would be skilled at reading data and asking the right questions, and every data person would be aware that they can only be as good as the questions which are asked"

"But a good game designer has always gone back to testing, to look at what it can teach them along the way, but they don't let it rule them because data is great for helping you refine, but it's not necessarily great for helping you to ideate. Because we're game designers we talk about it in terms of a hill-climbing algorithm, which is going to take us to a local maxima, the nearby hill and not to the new mountain over there.

"Figuring out where the new mountain is, how the mountain chain is oriented, even how to ask the right questions of the data, takes design chops. There's a balance to be struck. Ideally, every designer would be skilled at reading data and asking the right questions, and every data person would be aware that they can only be as good as the questions which are asked. A great example from the Facebook days is that the data metric suites themselves has assumptions built into them.

"All the talk, even to this day, is about DAU and ARPU. If you structure all of your databases around that question then it can be really hard to ask the question: how long does a player play. How long do they stay? We had this weird situation where, with the metric systems in MMOs, we knew exactly how much money we made from a player over a lifetime, because we knew what that lifetime was. In Facebook games we knew exactly how much money we made a day, but we didn't know how long we had the person.

"So there was this weird thing - one method is all about knowing how many people drive from New York to Boston, but not knowing what route they take. The other is all about knowing at any given moment how many cars are on a particular intersection. You have no idea where they've come from or where they're going. If you come in from the get go with those assumptions then the data is never going to give you the full story. You can query it as many times as you want, you'll never find out that they were heading to Boston for a wedding.

"It's a complicated subject and the real risk is absolutism. Design that ignores all data is going to be poor design, because design should be based on what people do and what they want. On the that ignores design is just a useless pile of numbers."

Related stories

Unity introduces Machine Learning Agents to help AI advancement

New feature goes into beta, primarily for researchers but engine firm keen to see what developers can do

By James Batchelor

Overwatch now mutes Xbox One players with bad reputation

Further efforts to stem toxicity after revelation that abusive players hold back development

By James Batchelor

Latest comments

Sign in to contribute

Need an account? Register now.