Re: Draining the Swamp: A Technical User's Experience

[Maciej Stachowiak <mjs noisehavoc org>]

On 07May2002 02:07PM (-0600), Richard Stallman wrote:
>     Here you seem to argue for providing convenient graphical preferences
>     for anything a user may wish to configure. Elsewhere you have stated
>     that you would like GNOME to strive to have as good a user interface
>     as the Macintosh. Unfortunately, those two goals are in conflict.
> 
> Practically speaking, there is not much conflict between these goals.
> Adding additional features to the GNOME configuration tool will not
> make GNOME as a whole much harder to use.

I'm going to have to disagree with you on that. Each visible feature
that is added amkes the tool as a whole harder to use.
 
>     Methods 1 and 2 above do not add any additional freedom - they merely
>     add convenience for some users. At the same time, they take away
>     convenience from other users, by making it harder for them to find the
>     preferences they truly care about. User testing under controlled
>     conditions bears this out.
> 
> This is a very surprising statement; could you tell us precisely
> what was tested in those studies, so we can see precisely what
> conclusions we can draw?

If you find this statement surprising, I recommend reading some books
about usability and human interface design. _The Design of Everyday
Things_ by Donald A. Norman and the _Macintosh Human Interface
Guidelines_ (or the newer _Aqua Human Interface Guidelines_) from
Apple are good starting points.

Sadly, I can't cite a specific study for this result off-hand, any
more than I can cite a paper that proves hash-tables have average-case
O(1) access time given a good hash function. But it is a well-known
result in the field.

Of course, ulitmately in the field of human interface results are
measured not by adherence to particular rules but by testing on human
subjects and observing the results. After all, the real test of a user
interface is how easy it is to use in practice.

Here is a simple user test you can try at home, or as a thought
experiment. Arrange 3 similar but not identical items on a flat
surface such as a table (for example CD-ROMs containing different
versions of the GNU/Linux operating system), in whatever way you find
logical. Now ask someone else to find one specific item from the
set. Most people can do that pretty quickly. Then arrange 20 items on
a flat surface and ask a passer-by to pick out a specific one. It will
probably take longer this time, and you could verify with a
stopwatch. Of course, with 20 items, spatial memory will help with
practice. But I bet that with 100 items, spatial memory won't help you
much without a *lot* of practice. And with 1000 items, it s pratcially
impossible to get any better, even if you added something like a
logical hierarchical organization using manila folders.

Now, this may seem like an extreme case. But you have to realize that
every configuration option that gets added to the interface takes you
further down the slippery slope, and that it's slipperier than you
think. Most users can choose between two items practically
instantly. Three may take a moment of thought. Seven may take a few
seconds of looking and thinking. Past seven, a normal person cannot
even count the items instantly by eye, let alone pick one out. (Some
rare people can apparently count accurately by eyeballing up to 100
items or so, but this is defintely exceptional, so if someone on this
list is one of those people, count yourself very fortunate).

And these hesitations of a few seconds at the sight of any large set
of things, or something unfamiliar, can add up pretty quickly.

> 
>     On the other hand, having the titlebar on the bottom of
>     the window rather than on top is something that very few people want
>     *and* which has no practical utility. So this preference is not
>     exposed in the UI, even though the underlying system makes it
>     technically possible.
> 
> I agree with that particular decision.  I'm not saying that GNOME
> should provide a GUI for every conceivable kind of customization--only
> for the ones that users often want to do.

The tricky word here is "often" - how many users have to need it and
how much, for it to be worth the cost? In reality this is often a
tricky tradeoff, not a relatively black and white issue like some of
my examples.
  
> It might be useful to look at a Macintosh and see which kinds
> of customizations it provides a GUI for.  That would make a good
> task list for GNOME customization features.

That's definitely a good starting point. GNOME does have a somewhat
different context in which to operate, however; it must support a much
wider range of hardware and underlying operating systems, has less
leverage in changing the lower layers to improve the user experience,
and has a somewhat different audience. So naturally the right
tradeoffs to make will be somewhat different.

Regards,

Maciej