Biometrics: The Science Of Play

When Valve says something, the industry tends to listen. Transforming from respected fledgling developer to one of the biggest and most influential companies in gaming in a shade over 15 years leaves the entirely understandable impression that you have a sense for where things are heading, and, right now, one of Valve's destinations is Biometric data.

In essence, Biometric data is captured by measuring a subject's body as it goes through a process or experience: heart-rate, perspiration, skin temperature, brain waves, eye tracking, facial expressions, bodily movements, and so on. By combining this information with detailed one-to-one interviews, a rigorous picture of the subject's moment-to-moment experience is created; a more precise account than the subject could possibly give using only their own perspective.

Needless to say, those in the business of creating consumer products - where success and failure is dictated by expert reading of the market's tastes, whims and desires - have shown great interest in Biometrics, though Valve is one of the few game companies to have spoken about it publicly. Indeed, on an episode of the Steamcast last year, Valve's Gabe Newell indicated that this degree of knowledge about the player could be essential to pushing the medium into "more interesting and compelling" directions.

"There are a lot of high profile failures, and a lot of companies think they're good when, really, they need all the help they can get"

"We don't really think that that's in doubt," he said. "The question is really about when and in what forms that takes. Even very simple noisy proxies for player-state, like skin galvanic response or heart-rate, turn out to be super-useful, and they're very much at the beginning of the kinds of data that you can gather."

In the long-term, this could mean consoles with on-board Biometric sensors collating huge amounts of data about your personal tastes and habits, and, as a result, games that incorporate that data to create a more tailored experience. As Newell pointed out, the future of Biometrics is all a bit "science-fictiony", but for Graham McAllister, founder of the UK-based company Player Research, current technology and techniques can help the industry to make better games right now.

"There's nothing else like it. We had to create it," says McAllister, as GamesIndustry International's Dan Pearson is hooked up to the company's proprietary Biometric technology. "When we started four years ago it was a really tough sell for the industry. Do you want to make better games? 'No, no, we don't need it.' Well, your games are all getting 5 out of 10, so you probably do.

"There are a lot of high profile failures, and a lot of companies think they're good when, really, they need all the help they can get. They've relied on the strength of their designers; they think that's enough, but you can't be someone else. You can't pretend to be a ten year old girl playing a game. It's just not the same."

McAllister has been studying human/computer interaction, usability and user experience for about the same length of time that Valve has existed, and these subjects are at the core of what Player Research can offer the industry. "Our focus isn't on game design; it's on interaction," he explains. "It's a different way of looking at games."

For the purposes of our demonstration, Dan was monitored while playing Black Rock's overlooked racer Split/Second and Splash Damage's online shooter Brink. Sensors that measure "almost microscopic" sweat secretions were attached to his fingers, a special controller recorded his interactions with the control pad, and a camera observed his facial expressions and physical movements. The finger sensors produced a real-time graph that represented his emotional reactions to the experience, fluctuating between excitement, frustration, and all points in between. Each play session was followed by a detailed interview, comparing Dan's personal account with the gathered data.

It is a rudimentary example of Biometric play-testing - Player Research also uses eye-tracking and several other methods - but McAllister claims that a single session like this can suggest a host of changes that would benefit the game immediately. "This is all based on principles of cognition, usability, ergonomics," adds Sebastian Long, a user researcher at the company. "This stuff is very well grounded. There's no opinion in it, pretty much. It just is the way it is; someone just has to think about it."

From this fundamental process, there's no limit to how deep Player Research can go. The company has a growing database of subjects, each with a file detailing their specific gaming tastes and habits, along with more general demographic information. Both McAllister and Long have extensive experience in running unbiased experiments, and they regard finding appropriate subjects to test each game as one of the most important aspects of the process, and the one that most companies get wrong.

"Using friends, family, other game developers, you're wasting your time," says McAllister. "Don't do it."

To get the best results, Player Research would test multiple iterations of a game across a carefully chosen range of subjects. Of course, this scenario was all but unheard of in its early days, with many companies submitting discrete chunks of their game late in development, when significant changes are almost impossible to implement. To some degree, this was influenced by those companies underestimating or misunderstanding what Player Research can offer, but Long and McAllister believe that the situation is changing for the better.

"Every Valve game is in the Nineties on Metacritic. They do this every week - every week. When Half-Life 3 finally arrives, it's going to be awesome"

"For the majority of games that we see at the moment we're getting multiple iterations, which is great," he says. "We get the easy stuff out of the way first, then drop down as they refine the design... We've had it where they just want a pat on the head, 'Well done, you're game's very good,' we'll show it to however many players, and they'll ship it the next month. That doesn't help anyone. We present them with a list of issues, and they can't do anything about it."

"There are not many people who are trained to do this," adds McAllister. "They try to do it themselves, but get it all wrong. A developer could potentially do something in-house, but the question is, is what you're looking at real? Have you biased it in some way that renders it useless? The guy at Valve who does it has a PHD in experimental psychology, so he's used to running experiments and making sure they're fair. It's not something that somebody from QA can just have a go at."

Player Research now counts EA, Sony, Disney, Codemasters, Splash Damage, Natural Motion and the BBC among its clients, and the idea that a game's designers always know best is no longer such a hard sell. The fact that figures like Gabe Newell and Sony's Shuhei Yoshida have spoken publicly about the long-term potential for Biometrics has been a great help, of course, but as the industry becomes more hit-driven a deeper understanding of what motivates players will be a powerful tool.

And there are possible repercussions for emerging markets and business models, too. As games of all genres and sizes migrate towards the free-to-play business model, effective implementation of micro-transactions is one of the areas that Player Research is keen to explore.

"Probably from a yacht somewhere," Long adds. "If we can pin down the psychology of micro-transactions, we're sorted."

After four years in business, Player Research has been exposed to how long certain misconceptions and mistakes can remain part of a game's design. Some of the biggest flaws in a product can appear in its earliest stages, in something so fundamental it goes virtually unquestioned. McAllister believes that developers should question even their most basic assumptions, and making Biometrics a core part of the entire production process is the most effective way of doing so. For a studio making a racing game, a Biometric study of previous or competing racing games could save enormous effort down the road.

"There's things we can do before the code's written," he says. "We worked on Brink as well. It was a classic example of too late... Paul Wedgewood told me that one the biggest mistakes on Brink was not using us enough. We came in at the end and presented them with a load of stuff, but they couldn't change it. But if they'd have known a year before that..."

Of course, defining and addressing problems is one thing, but McAllister also believes that Biometrics can be a tremendous aid to finesse and polish. The precise ebb and flow of a game like Uncharted 2: Among Thieves, for example, is made possible through a combination of time, money and raw talent, but not every studio can draw upon such reserves. The sort of technology that Player Research uses is particularly good at mapping spikes in the user's interest across linear levels - a short-cut to Naughty Dog's degree of mastery.

"Naughty Dog is good at it, but what if you're competing with them, and you aren't quite there yet?" McAllister says. "Naughty Dog will polish the hell out of Uncharted, but who else has the resources to do that? Very few, and there are a lot of games like Uncharted that are nowhere near as good.

"We're going to get people to play your game, and we're going to tell you where the fun is... Every Valve game is in the Nineties on Metacritic. They do this every week - every week. When Half-Life 3 finally arrives, it's going to be awesome."