ESR posted a rant about unusable software a while back. I stumbled across a counter-rant by John Gruber. He posted some clarifications and corrections later.
I certainly agree with ESR's sentiment that ‘desktop’ interfaces shouldn't require us to know all the little details about how stuff operates under the hood. This sort of thing pisses me off all the time. Indeed, the printer in the office at work cannot be printed to networkily from the Windows box, because I tried and was left baffled. I'm busy, so I can't sink infinite time into learning and fiddling with Samba.
ESR uses an “archetypal nontechnical user” called Aunt Tillie as his test for whether software is usable.
Gruber's basic points are:
- Free Software people shouldn't be worrying about Aunt Tillie until the software is at least usable to us hackers. Agreed. I find it hard to care about whether Aunt Tillie can print her knitting patterns over a network when I have to struggle to get a web browser (any web browser) to print a web page (any web page).
- Creating good user interfaces is really hard. Hackers should respect people who can do it, rather than just respecting people who can work out how to use cryptic software. I'd add that a lot of programmers seem to suffer from ‘usability blindness’, writing things that everyone struggles to use, and not noticing. They even don't notice that the software other people have written is hard for them to use. It seems to be taken for granted that even good software should entail weeks of hard work to install. People who create hard-to-use software should be spanked until they know better.
- It's practically impossible for us to fix this without sacrificing either ‘cheapness’ (and therefore freedom) or rate of development. We should give up and buy Macintoshes.
I want to disagree with the last point, although I suspect that the open source model does give us bad software quickly, and (when that software is popular, and if we're lucky) really good software eventually. That's really just the ‘worse is better’ argument. Hack something up without worrying about things like usability, and if people like it we'll fix the worst problems in the next version.
It works like that sometimes, but all too often people just aren't aware that the quick hack is defective. It's fine to come out with something hard to use, but you have to know and accept that. You're making it hard to use so that you can get something out there quickly for people to try. It's a deliberate sacrifice of usability, RAD focused on basic functionality rather than user experience.
There's also the concern that we need to maintain flexibility, for all those people who want KDE rather than Gnome, or Sawfish or Ion or twm. I use Sawfish, with my own choice of key bindings, but I'd give that up if there was one ‘standard’ desktop that actually made everything easy. Assuming I could get to enough things from the keyboard, of course. But Gruber thinks there are many of us who don't just use the standard thing who wouldn't go for it, and I fear he's right:
The worst part is that if anyone succeeds at putting together a usable desktop for Linux, these anti-usability Linux advocates will piss all over it.
Well, I can hope for the best. I have to, since I'm one of those people who cares about usability. I care because I have to use the software, not because I want to sell Linux to billions of numpty desktop PC users. And I'm not going to be buying a Macintosh. As a programmer, the availability of source code is the one innate usability advantage that Free Software has over the competition.