Designing for a physically diverse audience
A few things that seem kind of related to me:
- HP's face-tracking webcam apparently doesn't detect black faces as faces. (Kind of a fun video; the black man and white woman demonstrating the problem seem to have a sense of humor about it, or at least that's how I read their tone.) HP responds: "We are working with our partners to learn more. [...] We believe that the camera might have difficulty 'seeing' contrast in conditions where there is insufficient foreground lighting." Sounds quite plausible, but why didn't they test their system with dark-skinned faces in medium light levels before shipping the camera?
- Nikon camera's blink detection always detects a blink in some Asian faces. (Commenters on other sites also noted that the blink detection often detected blinks in nonblinking white women, and in fact in shots containing no people at all.)
- Microsoft's Project Natal initially had trouble tracking darker-skinned people, but MS says the problems will be fixed before launch.
- The iPhone has a lovely touch-sensitive screen—that can be nigh-unusable for people with long fingernails. (Most of whom, of course, are women.) Fortunately for such people—and for those who wear gloves, and for others who have trouble with the touch-screen interface for whatever reason—there are third-party iPhone styluses available (not pointed; they have a padded round tip, and are designed to be detected fingers) (but you can't do pinch-to-zoom with only one stylus, of course). But it always seemed odd to me that Apple didn't acknowledge that this was an issue from the start (by, for example, providing a stylus as an official accessory). It made me wonder if their testing group didn't include anyone with long fingernails.
- I was delighted a while back to see that iPhoto can now do face detection and recognition. But I was annoyed to discover that it was pretty much incapable of detecting my face. My working theory is that it has a hard time detecting faces with beards. I gather that the latest update to the software fixes some face-detection problems, so this may be resolved now; haven't gone back to check yet.
Of course, that last item is a little different from the others: men with beards haven't historically been discriminated against (the '60s notwithstanding), and we're well-represented in the tech world, so it seems unlikely that nobody at Apple tried the feature on a member of that group, whereas in the other cases it seems reasonably plausible to me that the test group wasn't diverse enough.
Regardless, I think that all of the above examples (and many many more) highlight a couple of things about technology design as practiced today:
- Designing to accommodate many different kinds of people is often a hard technical problem even if you're trying to do so.
- It's easy for designers to focus on particular sorts of people and to forget that their customer base will include lots of other kinds of people.
Of course, other kinds of diversity complicate things even more. Most tech is not designed for use by people with various kinds of disabilities. A lot of the world is not designed for large or small people. Most tech is not designed for use by people with different cultural assumptions from the designers/creators. A lot of tech relies on users speaking the same language as the designers. (One trivial example: many computer icons are images of puns that don't make sense except in English.) In our heavily gendered world, there's plenty of tech that doesn't take gender diversity into account (in various ways). Some tech even assumes that the users share the designers' religious or political beliefs. Or sexual orientations.
All of these are tough problems. I sometimes have a hard time even remembering that some readers of my blog live in other countries, or, say, other hemispheres; designing technology that's useful for everyone in its target audience is orders of magnitude harder.
But it's a problem that's worth trying to solve.