Xref: utzoo comp.sys.next:1131 comp.sys.mac:24657 comp.cog-eng:810
Path: utzoo!utgpu!watmath!watcgl!lrbartram
From: lrbartram@watcgl.waterloo.edu (lyn bartram)
Newsgroups: comp.sys.next,comp.sys.mac,comp.cog-eng
Subject: Re: replacing the desktop metaphor (Why any metaphor?)
Message-ID: <7552@watcgl.waterloo.edu>
Date: 4 Jan 89 19:28:22 GMT
References: <850@mtfmi.att.com> <673@cogsci.ucsd.EDU> <1489@umbc3.UMD.EDU> <22616@pbhya.PacBell.COM> <66401@ti-csl.CSNET> <4510@xenna.Encore.COM> <624@hindmost.gtisqr.UUCP>
Reply-To: lrbartram@watcgl.waterloo.edu (lyn bartram)
Organization: U. of Waterloo, Ontario
Lines: 57

In article <624@hindmost.gtisqr.UUCP> kevin@hindmost.UUCP (Kevin Bagley) writes:
>In article <4510@xenna.Encore.COM> bzs@Encore.COM (Barry Shein) writes:
>>The issue is not which is more "intuitive" (whatever that means)
> [stuff deleted]
>>Or is taking a little time to learn how to use a tool a dirty word?
> For myself and my co-workers, 'NO', but for my sister-in-law running
> a tanning-parlor whose least desire in the world is to learn programming,
> 'YES!' emphatically. Especially when the power and complexity would
> be a total waste of time.

	There is another issue here.  It is simplistic to assume that
ease of use is only related to the primary learning curve.  An interface which 
is cryptic, inconsistent and incongruent with the user's exisiting knowledge in
other areas is not only difficult to learn - it remains difficult to remember
and more importantly to predict.  The ability to infer further ways of using 
the system from its currently known procedures directly affects how easy the
system is to learn on a continual basis and to use.  This current discussion 
has centred around word
processing/text editing as the canonical interface, and while this is a well-
known case, i think there are some serious drawbacks with using it as THE
classic case.  For one, use is non-critical.  By this I mean that an error is
unlikely to be fatal (dependent on temperament, i suppose) and there is a
flexible time leeway to investigate and determine decisions on how to accomplish
the task at hand.  What about interactive systems where time (or some other
variable) is critical?  The air traffic controller may be able to afford a
long learning curve at the beginning.  HOwever, (s)he cannot afford a heavy
cognitive load on memory, prediction and interpretation when using the system,
and all decisions must be made within a fixedtime limit.  This is where
"intuitive" stops being a trendy buzzword and becomes a crucial feature.  
Studies have shown that the inconsistent interface (such as the command line 
interface of UNIX) and poor feedback (such as the lack of mode reflection
in vi) are not only related to a high learning curve but a continued error rate.
Moreover, as Draper has pointed out, the difference between the expert and the
novice in a system like UNIX is that the expert knows where to look to find out
things and the novice cannot fathom the complex paths for looking.  Prediction
 - or more properly, inference - is difficult in such a system.  The user must
infer sercondary procedures - ie, how to find out about the actual operation
desired - rather than primary procedures - how to actually do it.

>>There seems to be a fascination in this field with catering to some
>>mythical person with a two-digit IQ, total fear of computers, and not
>>enough technical sense to operate a push-button phone.
> I think that creating a smoother and more 'intuituve' interface is
> *not* catering to a lesser IQ, but rather to a person who may have
> a higher IQ than you, but who has different priorities and responsibilities.

Agreed!
What is the point of an "intuitive" interface?  To enable the user to operate
from a set of models and guidelines that do not require conscious and rational
deduction at each step.  The air traffic controller probably does not have a
two-digit IQ nor an irrational fear of computers - just a lack of 
contemplative and predictive time.  The above comment (delineated by the >> 
brackets!) is a typical example of comments from those who think that the
main need for user interfaces lies in the domain of text editing.  Actually,
much of what we can learn from these applications is more crucial in other
types of systems.  So think twice before you sneer at supposed "ease of
use" and studies in "intuitive" interface design.