bestkungfu weblog

Fixing open-source user interfaces

Filed in: design, Linux, Thu, Jul 29 2004 18:40 PT

A couple weeks ago, I took the leap. I bought a new Athlon 64 processor, hard drive and CPU, and set it up to run only Linux. It’s my first Linux-only desktop since about 1993.

Now, since then, I’ve had plenty of time to think over my decision. Some of it has been fantastic: MythTV, for example, is doing a great job in its beta test of replacing my TiVo.

But in other areas, well, there is still a way to go. Some things, like build errors and other installation conflicts, remain, and the nature of Linux development is that they are periodically resolved. But most of the trouble, in my experience is in the realm of the user interface. I wrote about this in a magazine just over five years ago, and while the situation has improved a bit, it seems that without a change of ideology in open-source software development, open-source software will be limited to knocking off new applications like iTunes. That’s not where you want to be: even the best copycat designer isn’t going to do much to advance the practice of user interface, just like the best copycat painter isn’t going to advance the world of art.

Many KDE and Gnome apps suffer from widget-itis. For example, I downloaded KBear, a graphical FTP client which was highly recommended by users, and found that it was completely unusable. Whatever I clicked, it did something I didn’t expect. And this is from a client representing a protocol that I know backwards and forwards. The interface was packed with KDE widgets, to the point that the simple use case of downloading a few files was damn near impossible. Interfaces need to be built around the user they’re designed for, not the widgets a windowing toolkit can supply.

All those years ago, I said that developers needed to learn some UI skills, instead of slapping together interfaces around the technical stuff that was already written. But now, given the scope of many projects, it’s probably more viable now to just get UI specialists in to work on them.

I don’t know why every user interface designer in school or looking for work right now isn’t working on the front end of an open-source project. There are more than enough to go around. And if Linux or any other open-source product is going to take hold of the market, they’re going to need to compete on equal footing with commercial software. That means doing things like usability testing — on real, non-geeky people — and solid user-centered design practices as a matter of course. Some of the stuff coming out of commercial software houses is great, but it’s not rocket science. It can be matched, or exceeded — and it must, if Linux on the desktop is to succeed. Being free is no longer an excuse.

Update: (30 July) No sooner do I post this, than do I find the best HCI-themed comic on the Web, OK/Cancel, has a comic and article on the topic today. Seems like I’m not alone here.

Comments are closed.

Powered by WordPress (RSS 2.0, Atom)