Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!linus!decvax!microsof!uw-beaver!tektronix!tekecs!orca!brucec
From: brucec@orca.UUCP
Newsgroups: net.works
Subject: RE: ICONS: Passing Fad or New Found Wisdom?
Message-ID: <1377@orca.UUCP>
Date: Thu, 7-Jul-83 12:08:11 EDT
Article-I.D.: orca.1377
Posted: Thu Jul  7 12:08:11 1983
Date-Received: Fri, 8-Jul-83 13:56:39 EDT
Lines: 107

IN-REPLY-TO: Vongehren@OFFICE-12 Wed Jun 22 00:50:00 1983
Subject: ICONS: Passing Fad or New Found Wisdom?

I meant to write this article and get it out to the net some time ago,
but got caught up in work, and completely forgot that I intended to do
it.  Since this is a subject about which I have strong feelings, and
since it relates to the work that I do, I decided "better late than
never," and went ahead to write this.

	A - Do you think that the current interest in the use of icons
	on terminals and computer displays is just a passing fad?

No.  The purpose of icons is to increase the bandwidth of the
information channel from a computer to a user.  It is a
well-established principle that people can recognize relatively
abstract geometric symbols faster than clusters of words in a language
they can read.  That's not to say that the recognition of an icon
can't be aided, especially for new users of a system, by one well
chosen caption word.  It seems to be the time in cmprehending multiple
words as phrases that slows down the recognition of written symbols.
The experienced user can easily ignore a caption, or even recognize it
as part of the abstract shape of the icon rather than a word (I doubt
you perceive the writing on a stop sign as a word after years of
driving).  I make no claim that icons can or should be "instantly
recognizable" in any sense without a period of training in recognition
of the icons in use on a system.  I do claim that once trained, a user
can operate faster using familiar icons than text.


	B - Aren't some of the current 'graphics' just a little too
	'cute' e.g. IBM upper-case 'lock'.  Is this just a sign of
	an immature field?  Will the marketplace tolerate this long
	enough for growth and maturity to occur?

I agree that there seems to be more marketing than cognitive
psychology in the design of a lot of icons.  Oh, well, it doesn't
really hurt anything but the feelings of people who want this field to
be taken seriously.  The real questions in the design of icons are
distinguishability between different icons on a given system, and size
and complexity of shape.  Even the cutest icon becomes just a shape to
be recognized after enough use (as an experiment, try writing the word
"thrash" down on paper and staring at it for a while.  Pretty soon, it
stops looking like a word.).


	C - What must happen for this field to mature?

I don't see anything that will stop it from maturing.  For the market
to take thsi form of interface seriously, it has to become easily
available (i.e. cheap), and easily usable (i.e. fast).  The economics
of the microcomuter industry guarantee that this will happen, as more
powerful hardware gets cheaper, and as more powerful operatin systems
become more common.  As long as there is a paradigm held in front of
the vendors in the marketplace, giving the less inovative of them a
model to copy, then progress will continue.  As flawed as I think the
Xerox Star and Appple Lisa are, they have provided such paradigms.
Enough vendors are working on copies or extensions of these models
that I think the field has the critical mass to keep going.


	D - Standardization of Signs and Symbols has occurred in other
	fields, e.g. Traffic.  Is there any effort to standardize within
	the computer field?  Should this be done?

Having been involved in the standards process in the computer field, I
can truthfully say that everyone wants a standard - his own.  The
field of iconic interfaces is too new for standards to be desirable,
since we don't yet know what we want.  It's probably best to leave
things alone until something like a de facto standard arises.


	E - Are there any obvious indicators for when it is inappropriate
	to use an icon in place of a word?

Clearly there is amaximum number of distinguishable icons in a system,
depending on the criteria used for desired speed of recognition.  The
less often an icon is used, the less recognizable it is.  So sometimes
it's better to use a word or two than an icon that the user has never
seen before.  Icons can be ambiguous in some circumstances, so for
communication requiring great precision (error messages from the
computer fro instance) icons are not desirable.  And I for one would
rather see a text string saying "the building is on fire" than a
little picture of a house with flames coming out.


	F - What would you offer as guiding principles for the use of
	icons in computer displays?  Will these differ for icon use on
	keyboards?

I can't offer any guiding principles since I haven't yet gone through
the cycle of designing and testing the effect of a set of icons.  I
will point out that the legends on a keyboard necessarily have a
different look from the icons on a display screen, due to the
difference betweem reflective and luminant shades, the texture of the
surfaces, and the lower resolution of the key legend.  This means that
key legends must be more cartoon-like (hard edges, high contrast,
etc.) than screen icons.  To date, most of the icons I've seen on
screens have been cartoon-like, possibly with some use of gray scale
half-toning.   I suspect that this is more a lack of imagination than a
basic limitation of the technique.


				Bruce Cohen

				UUCP:	...!teklabs!tekecs!brucec
				CSNET:	tekecs!brucec@tektronix
				ARPA:	tekecs!brucec.tektronix@rand-relay