sean@utoday.com (04/26/91)
Responses to net.views question posed to this group.
On Fri Mar 29 14:59:57 EST 1991 we posed the following question
to readers of this and other newsgroups to generate user response
for a regular column in UNIX Today!. Included below is a summary
of all responses received.
Thank you for your comments.
------------------------------------------------------------------------
QUESTION No. 2: Is a single GUI standard really necessary?
------------------------------------------------------------------------
The following is a summary (not a complete collection of the responces
we received):
No. Innovation, however, is really necessary, and premature or
ad-hoc standardization works against that.
-------------------------------------------------------------------
No, just as a standard dashboard layout for automobiles is really
necessary. It helps considerably if a manufacturer is internally
consistent in their GUI from product to product, however. One of
the benefits of upgrading old equipment is the lack of retraining
needed. We've found that going from a SparcStation-1 to a SparcStation-2
is essentially transparent to the user, while going from [OpenLook,
Motif, SunWindows, MacIntosh, Amiga] to any of the others is not
nearly as easy.
-----------------------------------------------
I don't think so, and if bloody idiots like Apple and Lotus keep
suing the pants off of lookalikes there will be such a lack of
standardization that users will become enourmously dissatisfied with
retraining time and gratuitous differences.
back on track... I think that the diversity of user interfaces is
good. Each one implements some features of a perfect window system
and each one has a failed experiment or two. Hopefully GUI designers
will continue to learn from previous mistakes, and the continuous
experimentation will function as evolution. Good mutations survive
and bad ones die (unless originated by the Big Blue monolith).
I doubt that the industry could ever be forced into one truly
identical GUI. Look at UNIX for a supporting analogy. It's all UNIX,
but a sysadmin on a Sun will be tearing his hair out when he tries to
configure an HP9k or an RS/6000 (personal experience).
There will always be enough differences to cause confusion in users
and ire in sysadmins.
Normal users would probably like the total standardization of the
market. It would mean they would only have to allocate a small
partition of their brains to understanding the computer and spend the
rest getting their jobs done.
I like diversity however. The availability of Tk, Xaw, Interviews
and the pseudo-availability of Motif and NextStep is nice and provides
incentive to evolve and improve.
Competition and diversity are vital to the advance of the art.
-------------------------------------------------------------------
Hello Gents:
YES, it would be great if there was only ONE GUI for the world to follow.
One government, one religion, and, why not, even one sex.
The problem with this thinking is IT IS LATE. Let's say Open Look was
adopted as THE GUI for all. What does that do to all of the people who
have started Motif development? Much economical loss. This is what
lawyers just love about technologists.
The inverse is true if Motif were to become THE standard. In fact, given
the shear quantity of available Open Look based applications, despite all
of the perceived de facto standard press for Motif, this would be even a
greater aggregate economical loss for developers.
Before there was a Motif, and before there was an Open Look, standards
bodies might have looked into this. Unfortunately, many standards bodies
are reactionary to what drives the real world, and that is sales volume.
Do you suppose magazines like Computer Systems News started writing about
Unix before every single computer company in the world endorsed Unix? No.
They're sheep. So also is it with many standards bodies.
So, what to do? There are several developments unbderway to provide a
"veneer" software layer that will allow an Open Look or Motif based
application to share a common API. Think of it as a GUI detente'.
SO THE REAL ANSWER IS "NO."
-------------------------------------------------------------------
A single GUI is no more necessary than a single keyboard layout.
It is desirable, but not necessary.
The important thing is that the efforts concentrate on one or two GUIs
sothe quality, performance and look is professional enough for an
end user to trust and feel comfortable with a GUI-based application.
He/she shall be able to move from one application to another (even
between platforms and OSs) and still feel "at home".
-------------------------------------------------------------------
From the application developer's position a standard GUI is neccessary
to allow a single development effort to cover an effective market. Without
a standard, applications have to be developed for EACH GUI available, this
makes for fewer applications, hurting the users.
From the user's point of view, a standard GUI is valuable not only for the
reason stated above, but also for the "transference" of application skills
from platform to platform, GUI to GUI, and application to application.
Warning: SOAP BOX
What has been left out of the GUI standards efforts (by everyone other than
Big Blue) is any addressing of the current installed base of conventional
terminals, connected to time sharing systems. This is a signifigant
road block to the wide adoption of the GUI concept by commercial software
developers and users.
The benefits of a GUI are NOT based on the graphical appearance (at least
not mostly), rather these benefits derive from the BEHAVIOR of these
applications. "It is not that it looks like a duck, its that when I poke
it, it QUACKS!".
Soooo...
I believe that both OSF and UI are missing an important segment of users
that can benifit from the GUI movement. It would be unfortunate if only
users of networks and workstations continue to be the only benificiaries
this major software technology.
-------------------------------------------------------------------
Organization: Sun Microsystems, Inc. Rochester, NY
Let the users choose, rather than marketing types, the standardization
will be a selection of supperior product at the moment. But consider the
fact standardizing an unknown is not the best choice. Maybe no one has
thought of the best GUI, so perhaps two or three GUI's provides healthy
competition and incentive to inovate.
-------------------------------------------------------------------
Necessary? No.
Desirable? Doubtful.
>From what I've seen, we're not even close to point where a single
standard would even be beneficial. Standards tend to gravitate
towards a lowest common denominator, otherwise the market gets
confused and either ignores the standard or "refines" the standard.
GUI needs vary. GUI's tend to hog resources. What I need and what
I want for a CAD application can be quite different from what I need
and what I want for text editing. A "GUI standard" is premature. The
market needs to bite, kick and claw some more before it'll be ready to
settle down and adopt a single standard. The breakeven point in
resource tradeoff just isn't here yet.
-------------------------------------------------------------------
No, a single GUI isn't needed. Multiple GUI's fosters competition and thus
progressively better interfaces. Look at how the Mac interface has stagnated.
However, it would be nice if the user could pick the interface regardless of the
application. One of the prime considerations that should go into developing a
UI should be giving the user a choice.
Maybe you could contact Bill Rouse, the president of my company. Our company
does quite a bit of research in the area of UI in general.
-------------------------------------------------------------------
I don't believe that a single standard is necessary. After observing people
of various levels of computer "competence"(receptionists to programmers) at
work I firmly believe that switching from one GUI to another is trivial task
that anyone can handle.
Of course if having one standard GUI meant not having to deal with MS Windows
then I'm all for it!
-------------------------------------------------------------------
Not only no, but hell no. A single GUI would be convienient - no need
to learn new paradigms from environment to environment. But the
implications of a single GUI standard are frightening.
Who would own this standard? Afterall, Apple is trying to own the
concept of a desktop, Lotus owns a specific type of menu for
spreadsheets. If one company bent on making profits were to own the
"standard GUI", everyone would lose.
Even if a standard GUI were to be made that was free, the inherent
creativity-stifling element of standards is dangerous. GUIs have not
existed long enough to know exactly what the right one looks like. If
we arbitrarily decide on some specific GUI then it will make it much
more difficult to successfully develop alternative, potentially better
interfaces.
As it is, the profusion of different GUIs is not too bad. Some of the
important fundamentals - scroll bars, radio buttons, and the like have
more or less standardized themselves (unless someone decides they own
the idea of a scrollbar, in which case we are in trouble). Also, part
of the notion of a GUI is that it be intuitive; if it is truly
successful then even if you have never seen the GUI before, you can
sit down and use it - why bother standardizing it? The dangers of
enforcing a potentially ill-chosen interfaces outweigh the slight
gains in ease of use that would come of a standardization effort.
-------------------------------------------------------------------
YES A GUI standard is necessary.
There are two issues here. The first is a GUI's look and feel, and the associated standards, (ie Openlook vs OSF motif). The second is the issue of software compatablity. I'll talk about both in relation to the company I work for,
but do rememmber that my comments do not necessiarly reflect the views
of my employer !!
1/ Look and feel.
The company that I work for has just released a report that reccomends
Microsoft windows and the standard "Interface" for PC's. The
main reason for this is that it will cause a large reduction in
personal training costs inside the orginisation because of
the similarity of operation of software packages written in this \
environment. (ie pull down menus operate in a similar manner etc)
And of course MOTIF is also very similar to windows.
The bottom line is that buy having a standard look and feel it is
felt that the company will save a LOT of money.
2/ Software Compatibility.
The section I work for is into X Terminals. A typicamenus of
applications have a selection of items that run on various
hardware platforms, from minicomputers to large IBM mainframes.
Because of the standard ( X ) it means we can choose different
platforms to run differemt applications (Typically, getting the
platform that runs the software application best) and then offering
our X Terminal users a consistant interfact. The typical user does not
even know what hosts a application he is using runs on.
From the above to comments you can see that for a large orginisation GUI
standards make very good sense.
-------------------------------------------------------------------
That depends. A single API for all GUIs is vital. A single user interface
(OpenLook, Motif, Presentation Manager, Windows, MacOS, etc... take your
pick) is a complete waste of time... different people have different needs
in a user interface.
The whole business of GUI wars between Open Look and Motif is insane. Not
being able to compile an X windows program on a Sun-4 because you don't
have the Motif toolkit would be funny if it wasn't so expensive. A simple,
high level API that lets you select whatever UI style guide you like with
no more effort than recompiling... on anything from a Mac to an HP Snake...
has to come about before window programming becomes the great resource
sink of the next century.
The "UNIX" of window systems has yet to emerge. We're still at the OS/360
stage.
*************************
* END OF SUMMARY *
*************************