mcg@shark.UUCP (Steven McGeady) (08/20/83)
On the subject of icons, and "de-mystifying" computers and software
systems, I would like to point out the following:
There is an excellent paper entitled "Analogy Considered Harmful",
by Thomas Moran (Xerox), and Frank Halasz (Stanford), printed in
the Proceedings of the First Conference on Human Factors in Computer
Systems, March 1982. The entire Proceedings of this conference
address many (if not all) of the questions thus far raised in this
newsgroup.
In this paper, the authors mention that it is tempting, but dangerous,
to explain or model the workings of an automatic system by means of
analogies, as these analogies are inevitably imperfect, and the flaws
in them lead at best to confusion, and at worst to fundamental
misunderstandings of the operation of systems.
The authors take a stance in direct opposition to many others, in
particular the (in)famous Donald Norman.
Consider the typical model presented that a filesystem is just like
a filing cabinet: the files are folders, which have labels, (their
names), etc. Directories are filing cabinet. One may move a folder,
or relable it. Up to this point the analogy works. But consider
file protection: how does this fit into the model - one can extend
the analogy by saying that there is a combination lock on the filing
cabinet, and individual locks on each file. Hmmm - somewhat unlike
most filing cabinets. What about sub-directories? Well, says the
analogist, its just like having a file cabinet inside another
file cabinet. What about file-to-file links? What about undeleting
previously deleted files. One may continue to extend the analogy,
but what you end up with is something remarkably *unlike* a filing
cabinet.
The same things can be said about the "what-you-see-is-what-you-get"
editor vs. typewriter analogies. ("It's just like 'X'ing something out.")
The authors point out that it is far more effective to try to
teach the actual conceptual model behind a system, rather create an
imperfect analogy to the "real" world.
The point here is not that icons and de-mystification are bad,
but rather that a computer designer should strive for an internally
consistent model of his system, rather than one which is imperfectly
analagous to something assumed to be familiar.
S. McGeady
Tektronix, Inc.