The Power of Physical Interfaces

Way back in 1986, I discovered the joy of complex digital devices with custom physical interfaces. While I was extremely comfortable with computer keyboards and starting to find my way around with a mouse, I had no idea what I was doing with drum machines, sampling keyboards, or video editing tools. I could press keys, turn knobs, or even push dedicated buttons and cool things would happen, consistently. The patterns were easy to learn and I could quickly focus on the work rather than the interface.

I had, however, gone to that program for computer graphics, and to computers I would return, focusing on screens, keyboards, mice and other pointing devices, and eventually touchscreens. I could do everything with a computer, in one device, and often in a portable way, right?

Over time, I found myself sticking to tasks that worked better with a keyboard and a pointer. Writing is a lot of what I do, so that was fine - but editing photos always felt awkward, and anything to do with music, video, or sound felt like endless searching and clicking. Keyboard shortcuts help, of course, but when they're scattered across a keyboard I find myself looking away from the screen and back to the keyboard on a regular basis. Yes, I could create and edit video, but it felt awkward - in large part because I had experienced better ways to do it.

Almost a decade ago, battles over skeuomorphism, the imitation of real world interfaces on screens, annoyed me. Computer efforts to look like real-world devices and situations were always doomed to feel fake. There isn't real wood grain on a computer screen, and touchscreen buttons don't cast shadows. At the same time, though, removing fake shadows took out a lot of contrasts that made it easier to see divisions between parts of an interface, and the resulting interfaces weren't delightful either. Haptic feedback makes me slightly happier about typing on my phone's tiny screen, but, again, it isn't much.

In the end, I don't think those efforts to create a pure graphical interface were purely about style. They were trying hard to forget that graphical interfaces, even fancy multitouch screens, are often terrible experiences compared to physical interfaces. Range sliders have never been a great way to interact with a computer, though at least they provide a visual sense of state. It's possible to create interfaces that involve rotation on a touchscreen, but in my experience they're really hard to navigate with a mouse. Knobs and dials become visuals more than inputs, and often emblems of that supposedly dreadful skeuomorphic design.

When I can turn, press, or slide, and get instant feedback, I can get a lot more done. When I have a jog/shuttle dial, I can navigate video and sound smoothly and precisely . When I have a set of physical sliding faders, like the (upside-down) ones in the picture above, I can tell by feel or by look what's happening to multiple tracks in the music. I can work in the mostly dark, focused on the content, and still find my way.

When hardware manufacturers have the freedom to arrange interfaces, they can separate tasks and minimize confusion. I frequently have to stop and figure out what mistake I made when I hit the wrong key combination, because there isn't a clear visual indicator or obvious way to go back. We accept these problems because they are so familiar, but other interface approaches are worth exploring.

Designers will doubtless keep pushing the envelope of on-screen design, but for my own purposes I'm trying to let screens do what they do best: display information. Carefully-considered touchscreens can work wonderfully for some kinds of projects, as I learned long ago on a Xerox 5090 copier, but for most of my projects beyond basic text and numbers, I'm looking for physical interfaces that directly reflect the work I'm doing. They are still mostly digital underneath, but given that the work I'm doing is on digital devices, that's fine.

Our current software models create a lot of problems for this approach. We generally assume that every application has its own interface, with a few common approaches. Keyboards and some kinds of pointers are the only devices designers generally assume, and even the keyboards are optional. Application designers and operating system developers don't seem to spend a lot of time figuring out better ways to map physical devices to software interfaces. Unfortunately, even when devices are available, they often come with drivers that are proprietary, perhaps working with one family of software really well, but stuck with kludges for everything else.

One bright spot on this comes from music, where people have been connecting physical devices for decades. Even though so much audio equipment is devoutly proprietary, the nearly four decades of having MIDI as a common standard has opened doors. Allowing the control synthesizers from keyboards, drum pads, and sequencers was a huge step forward for music and for interface customization. Mapping MIDI signals to device behaviors often still requires effort, but a community of shared experience makes it easier. Now that MIDI often runs over USB cables, and many devices run as class-compatible without additional drivers, it's even easier.

As I've returned to the music and video projects I first explored in 1986, I'm glad to see that digital has made a lot of the underlying work easier. I don't miss the physical interface of audio tape, razor blades, and tape. At the same time, though, I definitely want the controllers I had then, rather than the touchscreen, keyboard, and pointer interfaces that dominate computing today. Can we build toward better controls?

(And can we even, perhaps, regularly connect Web interfaces to these tools?)