How Valve built the controller of the future

See the many mutants that died for gaming glory

Earlier this month, we visited Valve’s headquarters in Bellevue, Washington for an exclusive look at the development of the mould-breaking Steam controller. In interview with Anna Sweet and Greg Coomer, we discussed the exhaustive trial-and-error-and-trial-again it takes to come up with something revolutionary.


Whether or not you’re convinced by the Steam controller, there’s no denying that it is definitely a New Idea. With flat trackpads instead of sticks and buttons in unexpected places, it makes even the PS4’s touchpad-toting controller look unimaginative. But why did Valve make one in the first place?

“For a few years now”, say Greg, “our customers have been asking us to solve a problem, which was that they had to give up all their games when they go to play in the living room. To do that you need three things: hardware, like a Steam Machine, input and an operating system.

At first, our goal for our input device wasn’t all that clear. We were doing some experiments on wearable computing and virtual reality, but for the living room, we needed an input device that was able to play the whole Steam catalogue, without requiring somebody to bring a mouse and keyboard onto the sofa with it. So we looked at what PC gaming uses by default, and of course that’s a mouse and keyboard. And we realised we had to make the games think they’re being played by those traditional devices, because there’s no way we can update all those titles to have some new input method - they were never built for a game controller, and every development team has since disbanded. You can’t just crack a few thousand games open to magically support something new. So that was our goal: to translate between the expectation of a mouse and keyboard, and the actual buttons and surfaces on the controller.”

READ MORE: Valve Steam Machine preview


If Valve was going to trick every game into its 3000-plus catalogue into thinking they were being played by traditional means, it would need a new device. Possibly quite a weird one.

“With a mouse and keyboard, you’ve got 104 keys in one hand, and a super-high-resolution 2D pointing device in the other. We started building controllers that had tons of buttons on them, and controllers that all kinds of different methods of doing that high-resolution 2D pointing. One of the early front-runners was a trackball.

We built a lot of prototypes that had a trackball on the right-hand side. We had small trackballs, we tried cueball-sized trackballs that punched all the way through to the back of the controller so that the ball’s being held around its middle, and you can hold the whole thing and have really precise input. That was great, because you could spin it and stop it with your fingers at the back of the pad. Trackballs are actually really, really good in terms of 2D pointing. It’s precise and high-resolution, and it gives you a lot of physical control. But it also had some downsides: it’s asymmetrical, for one thing, so right-handed only; different people liked different configurations, so it’s not a one-size-fits-all solution; and a trackball has a lot of moving parts, which means you’re going to get a lot of Cheeto dust stuck in there." [For readers not familiar with Cheetos, they are a cheese puff snack, larger than a UK standard Wotsit but not as nice].


Stumped by the Cheeto Dust Problem, Valve began investigating other means of control.

“When we started trying trackpads, they were way behind trackballs for a long time. It wasn’t until we started changing the software that began to see its potential. There wasn’t a ton of physical differentiation between the prototypes, but in the firmware, there was a tremendous amount of work that we were doing to get the trackpad to behave differently for different kinds of games. We started to think, this is more promising, because it gives us more power as software developers to adapt to the requirements of a different games, or to games in the future that we haven’t thought of yet.

For a while, we thought touch input was so strong that we even went off past trackpads and investigated touch input across an whole flat surface. It was useful to try that, but it taught us that wasn’t where we wanted to go - often we ended up needing to have a divided experience, where you’re looking down at your hands a lot and then up at the screen, and the tactile feeling of a controller was absent.

It also taught us how important haptic feedback is for any touch device. That channel of information from the touch input device to the user’s hands is vital. It can carry a number of different kinds of information on what’s happening in the game, but on a nuts and bolts level it tells you what part of the touch surface you’re using. We subdivide those surfaces into slices, or concentric rings, and those boundaries are impossible for the player to detect unless you have haptic feedback.”