Since beginning my foray into Industrial Design and Interaction design, there are a few common threads that keep popping up in my projects and annoying me. Today I want to write about one that came up in a recent project where I was designing a ubiquitous computing/ambient weather forecasting device. As is usually the case, we were required to plan a heuristic evaluation.. for a physical device.
This shouldn’t be a big deal, but the problem is, everyone immediately jumps to Neilsen for the heuristics they’ll be using. This is all well and good; they’re a pretty good set, but they are intended for 2D screen based interfaces, especially web, where it’s automatically assumed that the interface will have the user’s full attention; ie: staring at a computer. I’ve googled “heuristic evaluation” +”physical device” about a hundred times over the past two years and always get the same sad results. Usually, the only things remotely worthwhile are various archives of this IxDA forum thread from 2008. The gist is that a guy is looking for heuristics for physical device testing, and the IxD folk tell him to look up ergonomics.
In my own practice, I take Neilsen’s 10, ditch the ones that really don’t apply, alter some, or grab some from other limited sources. One of these other sources that I’ve used before is a list from a consulting company called Tristream. I used these in the evaluation of a piece home/industrial automation back-channel management software. It wasn’t totally ideal, but it got the job done.
Back to my ambient weather forecasting project.. I was lucky enough to find a great paper titled “Heuristic Evaluation of Ambient Displays” from the ’03 CHI conference. In it, some folks from UC, Berkeley and Intel Research went through the heuristic selection/creation process I described above, in this case, especially for ambient displays. The paper discussed their test of these heuristics and found them to be significantly more useful in identifying major issues than the original set from Neilsen. In the test, “a single evaluator will only find about 13% of major issues” but “a single evaluator using the ambient heuristics finds 22% of known major problems on average”. (Markoff et al, 2003, p. 175)
Where am I going with this? I really think there should be a published set of heuristics for physical devices. I understand that the type of device greatly impacts the heuristics that should be used, but it would be nice to have a basic starting point that made more sense for 4D interaction than just the stock set from Neilsen. I’m hopping to draft such a paper over the summer. It’d be of tremendous use for myself, but I think the Industrial design community as a whole could benefit. If any of the designers out there have suggestions to this end, please let me know! I’d appreciate the input.