[Sussex] Rantting

Geoffrey J. Teale gteale at cmedltd.com
Thu Apr 14 10:07:55 UTC 2005


Steve Dobson <steve at dobson.org> writes:

----- %< -------------------
> I am tolerance personified!  Doesn't my statement above on freedom prove that?
> I never said that it was invalid to use GUIs or that I was intolerant of those
> that use them.  I use a GUI desktop most of the time (but it runs more shells
> than anything else).  What I was saying was that GUIs put a limit on the
> way you interact with the computer - a point you appear to broadly agree
> with.
>
> This is, in fact, a well documented problem of GUIs.  By limiting the way
> you interact with a computer imposes a limit on the way a user thinks that
> a computer can be used.  

OK.. I fundamentally disagree.  The problem isn't with GUI's in
themselves, it's that we haven't evolved (or rather generally
accepted) input devices that match the potential of GUI and so we've
been restrained from developing GUI applications that utilise a more
complex mode of interaction than simple point and click.  

Text based control is more powerful only because we learn arcane
mechanisms for specifying filter conditions and forcing streams of
data through pipes.  There's no reason that GUI's cannot achieve this
level of power (that is they can achieve what you can achieve in the
command line without crossing that very fuzzy line into actual
"programming") and indeed they have, several times over in research
project done by Xerox, Sun, Microsoft and many others.  The keys are:

  - Replacing mice with 3 dimensional input devices.

  - A more complex representation of data (not just a 2d
    representation of a single level of a hierarchy of files).  Sun in
    particular have done a lot of work with 3d representations of
    carousels of data.

  - Visual, configurable representation of control and filtering
    tools. 

  - Visual chaining of tools (almost a visual representation of
    pipes).

That's all very interesting, but none of these research projects up on
our desks because almost invariably, when you've got a 30 minute sales
pitch people opt for a simpler, less powerful model that they can
understand easily.  No end of research showing that text-based input
systems are far more efficient for data-entry than GUI applications
has not made the slightest dent in the march towards GUI applications in most
data-entry driven industries, simply because people are now familiar
with WIMP and they think anything else is too complex to bother with.

I've been having trouble with my hands recently, a lifetime of playing
heavy strung guitars and a typing style akin to a jack-hammer on ice
have started to give me pains across the back of my left hand.  As a
result I will shortly be taking delivery of a Fingerworks Touchstream
LP "keyboard" to try and ease the impact on my fingers.  Now this
device replaces a lot of common key commands with gestures and indeed
acts as a mouse as well as a keyboard.  In doing all this it allows some
much more complex interaction with the GUI that makes some trivial
tasks a little more interesting.  Imagine a world with two mice
available to you to control GUI elements without moving your hands
from their typing position - I'm looking forward to playing with
things like this:

http://www.fingerworks.com/XWinder.html

.. a trivial example of what you can do with only a slight increase in
input device complexity.  Now with a little imagination consider the
wonders we might achieve if we designed GUI environments with this as
the base point instead of the model we use currently (as defined by
Xerox and Apple in the 60s, 70s and 80s).

-- 
Geoff Teale
CMed Technology            -   gteale at cmedresearch.com
Free Software Foundation   -   tealeg at member.fsf.org




More information about the Sussex mailing list