Re: Prototyping the next generation panel?

[Karl Lattimer <karl qdh org uk>]

On Tue, 2008-10-28 at 15:25 -0400, Colin Walters wrote:
> Hi Calum,
> 
> On Tue, Oct 28, 2008 at 2:30 PM, Calum Benson <Calum Benson sun com> wrote:
> 
> > How about we start as we hopefully mean to go on, by usability testing
> > mockups of some alternative shell/panel designs first, to figure out if
> > we're heading off in roughly the right direction before we even think about
> > committing anything to code?  (If those initial investigations have already
> > been done, can we see the results somewhere?)
> 
> See:
> http://live.gnome.org/Boston2008/GUIHackfest/WindowManagementAndMore
> 
> Our goal in the near future is to do some early prototyping and finish
> work on underlying infrastructure, hopefully get to the point where we
> can iterate and get "functioning mockups" in front of designers and
> usability experts.

Putting things in front of designers and usability experts is almost
useless, sure they can pick out the retarded mistakes that developers
make however I've also seen usability experts make unforgivable mistakes
in usability... The only way to be sure about this stuff is to put in in
front of people who don't know any better and watch them fail, how they
fail and then determine why they fail.

A usability expert can follow the rules of interaction design but can
they spot the problems that will be faced by someone of lesser
experience? The simple answer to that is no...

That's why big companies pay for focus groups and usability tests with
multiple camera angles and click logging etc... The usability experts
and interaction designers can set developers on the right path, but
cannot ensure that all obstacles are out of the way along the road.

Any usability expert that refutes what I've said here is arrogant at the
very least, one of the big problems we have in GNOME is a belief we know
what's right and no data to back it up... For instance, this bug sticks
in my craw a bit http://bugzilla.gnome.org/show_bug.cgi?id=361958 it is
indicative of the arrogance of developers in interaction design...
Breaking one of the golden rules, discoverability... 

usability, human computer interaction and user experience are the holy
trinity and you can't make a stab at what's right, you start with an
idea and develop it until its right, not by augmentation by a panel of
experts because all experts make mistakes, but by watching people make
the mistakes you're too clued up on to make yourself.

That's why I feel its necessary to put something out there on the web
and log as much of the data back as we can. HTML/JS can easily be
knocked together into interaction tests, and moreover its a cheap way of
getting at least _some_ data, sure we won't have the benefit of watching
people's expressions or how they're moving the mouse, and where their
eyes are drawn to on the screen but we'll have the clicks, and the
clicks help.

Put the current design mockup on a website, add an success event to the
"add workspace" button and a fail event to everything else, then ask a
user to add a workspace and watch the sprawling number of fails that
occur, until almost randomly the tester hits the right button. Next
modify the image, and add the text to the black unused space at the
bottom of the screen "Add workspace -->" and watch how many successes
you get, first time!

People don't make up the rules of human computer interaction, they
determine them from people... Don't take the human out of HCI...

BR,
 K