Late last week, Sun gave a practical present to the Linux community: the findings of its GNOME usability study.

"Usability," like the word "design," found voguish status among vaguely mod information age sages. Everyone knows, for instance, that things should be "usable," and it's a sign of deep knowing to make remarks about the usability or unusability of a given thing; it's also a sign of knowing to talk about "good design" or "bad design." In either case, sometimes what you ought to be hearing is "I didn't like it," or "I did like it," or "I think this sucks but can't tell you why."

Some people spend a lot of time thinking about usability, and they've developed robust methodologies and bodies of work around figuring out what helps people do useful things with the tools in their lives. Many, many more people aren't quite so sure what makes something usable as they are that they want more things to be that way, whatever it may mean.

The danger of an over-used concept is that it becomes trite and worn out, its meaning lost in the obfuscatory wilderness of knowing party banter and guru-for-a-day proclamations. Some people I've spoken to, for instance, use the word "unusable" in everyday assessments of fairly routine things, and not as a wry way to describe a shopping cart with a straying wheel. They like it because it's a shinier way to say "this thing sucks" or "I didn't know you had to push down before turning." Others use the word "usable" to mean "dumbed down" or "simplistic," and base their understanding of usability on some vague notion of what works well with stupid people.

Although there are many, many talented people at work on building the better desktop, entrusting them with characterizing what makes a piece of software "usable" or not is a bad idea. Not because they're flawed or less than talented, but because they live with the software they're creating day in and day out. Anyone who's been around GNOME for a while has probably become used to its idiom on a level that's blinding when it comes to anticipating what less proficient and capable users need to succesfully navigate an interface. It never occured to me, for instance, that anyone would have a beef with "moon and stars" icon meaning "exit the desktop." At least not until I remembered that I'd once passed the mouse over it tentatively the first time I encountered it to make sure it did what I thought. That isn't to say that there haven't been some nice touches added. Ximian has made GNOME a better (more usable) experience for a lot of people, for instance. But there are a few more yards to go in this push.

I'm not even a hacker, and I know that I'm no more fit to figure out what that oft-cited, sometimes-seen end users need than a belligerent bash zealot: I'm used to the underlying structure of GNOME in particular and the overall family of Linux GUI applications in general. The impatience I feel when a guest user on my machine becomes ensnared in the inexplicable differences in dialogs between, say, Motif-based Netscape, gtk-based Abiword and "alien file system" StarOffice tells me I've made my peace with some terrible interface shear that others simply can't be expected to embrace as good or normal, at least until they learn to create symlinks.

So the gift Sun has given is one of outside perspective. They did more than any amount of buttonholing friends and spouses in front of a glade session, or mailing list feedback could ever hope to produce: they provided end users, methodology, and a testing environment, and they got useful feedback from people with nothing at stake in the process... they got usable (pardon me) data from which decisions can be made, personal preferences can be definitively laid to rest, and heretofore problematic blindspots exposed.

There are some cranky types out there who don't want any part of this, believing that a more usable (end-user friendly) Linux is representative of a "dumbing down." They'll hassle anyone who takes the time to listen or read to their comments over "bloat" and "cruft." My own experience has been that even the most heavily engineered distributions (I'm reserving judgement on whether Mandrake or SuSE gets the nod on that) strip down admirably well if one so desires, leaving all the "bloat" back on the installation CD, where it does nothing to weigh down the desk and dent the floor. The general vibe, though, is that more friendliness (what people used to say before "usability" came along) is a good thing. While the Linux desktop may be pretty good (it's good enough for me as a platform to work from, even if I do wish I could just go into the store and buy more than $9.99 Loki games for it), it can't hurt to get a little better, and enjoy a little more polish.

Despite GNOME being the target of this particular study, it's also a great opportunity for every other desktop project's developers to at least ponder where GNOME has gone wrong and look for similar weaknesses in their own framework. Nothing in any of this dictates a particular practice, but reading the study provides a way to see where interface designers have fallen off the path and learn how they might have arrived at some questionable decisions. After all: GNOME and KDE hackers are all, at heart, hackers, even if their more restless acolytes in the end user community would like to pretend there are genetic differences.

Next: How to Help Someone Use a Computer »