Re: [Usability]Re: Short article on OSS usability

[Seth Nickell <snickell stanford edu>]

> Or is the author saying that *other* beneficial types of usability 
> discussion are conspcuously absent from our mailing lists?  If so, what 
> kind of discussion would be more useful?

A design process. 

Start with user goals, add context, extrapolate to tasks, perform a task
analysis, prioritize tasks (frequency, importance, urgency, etc),
prototype an interface around these tasks, test interface (flip-chart,
mockup, simple prototype implementation, etc), refine interface,
implement, usability test, refine interface, implement 1.x, rinse
repeat.

Right now open source development leads to, at best:

Duplicate existing application/interface/feature in another system | Do
something that seems really cool *and* is fairly simple to hack out,
implement, post early version/screenshot, people argue based on their
personal taste, take some ideas into consideration, finish
implementation, release, usability test, tweak some things, implement
1.x, rinse repeat.

Problems with the current process:
1) Good usability testing is relatively rare
2) Usability testing, especially non-comprehensive usability testing,
isn't super good at finding fundamental interaction design problems
3) Programmers are rarely willing to throw away lots of code out if such
a deep problem is found, if that is necessary to fix the problem
4) Personal taste rarely reflects any sort of deep thinking, and also
samples from a very limited population. This means that fundamental
interaction issues are usually ignored, and even the top level interface
is tuned to enthusisiasts (at best).
5) This process really doesn't include a good mechanism for seeing the
forest rather than the trees
6) Big important usability issues that aren't easy or super-cool (undo,
for example) *tend* not to get done

The net effect is that even the best usability work within an open
source context is focused on very visible but comparatively minor
details rather than looking at the big picture. If you look at GNOME
relative to, say, MacOS, this is pretty obvious.

-Seth