When you work with a computer a lot, there are some very valuable
characteristics you want to have. You want your computer to respond
so quickly that you never have to wonder if it got that keystroke or
click or other gesture. You want it not to demand more work than is
necessary to get the job done. And you want it to stay out of the way.
That last part is really easy to get wrong. A modal dialog box is
anathema. It demands your attention, distracting you away from your
task, to deal with the computer's mechanism. Modes, where a given
gesture has a different meaning also get in the way. Separate
applications are major and dangerous modes, despite the fact that
almost all software is delivered in separate applications of which
you are expected to use several. Bad news.
When you don't have that problem, you can easily achieve
automaticity, that blissful state where your fingers know what to do
so your attention can remain on your task. That achievement,
sometimes known as becoming a power user, facilitates getting a lot
of work done smoothly and easily. Good news.
Pie menus are one class of controls which encourage automaticity, so
much that you can often proceed faster than the screen can display
the image. On the other hand, using rollover/hover to see a larger
version of some image or display, usually requires a significant
delay before the gesture takes effect, and clicking on a link almost
always takes an annoyingly long delay, even when the linked page is
local. Since exploring a large data space is a frequent computer
activity, this matters.
Ease of learning an interface is another issue of importance for any
new interface mechanisms. If you're old enough, you remember that it
took many minutes for most folks to become comfortable and competent
with the mouse or track ball. You may even remember the Star Trek
scene where a character encounters a Macintosh and picks up the mouse
and speaks into it.
Jef Raskin included a hospital information system of his design in
his seminal work, "The Humane Interface". His novice users were
already comfortable with the mouse, but using the mouse buttons for
zoom in and zoom out was new to them. Nevertheless, because the zoom
world was geographically laid out without overlaps, the system was so
easy to learn that people were comfortable and competent in a single
minute. Of course, computer experts took a little longer.
This ease of learning was so striking in comparison to other hospital
information systems, where you take a two week course before they let
you at a real keyboard, that I want to make use of it. I claim that
tens of millions of years of history when our ancestors made it back
to the nest have installed geographic navigation talent in our very
bones. OK, in our genes.
However, the zoom keys themselves are not wonderful. They constitute
a velocity based navigation tool, and thus have the same drawback as
joysticks in comparison to mice and track balls as displacement based
tools. This leads me to want automatic zooming to just the right
amount to read the text which was writ small in the place I'm going.
Two methods of initiating the zoom occur to me. It seems safe, and
pretty easy, to mouse into an area containing small stuff and click
to zoom in. On the other hand, it would be even easier to just mouse
there and let rollover initiate the zoom by itself. Rolling back
across the border zooms right back out to the containing page. With
fonc, I believe that making the test of and between those ideas would
be quite easy. Is that right?
I believe that with a zoom world like this, anyone who has the skill
to arrange shelf space in a store could build a web site that was at
least that easy to navigate. And if one wanted to include a data
sheet or instruction booklet for some item on a shelf, it would not
take much room to put it next to the item no matter how much data it
contained.
And perhaps you've heard of the program which substitutes for a
police sketch artist. It puts up a panel of twenty random cartoonish
faces, described by about twenty five variables. The victim who is
trying to identify his attacker doesn't have to deal with the
variables, though; he just picks one or two faces which remind him,
slightly, of the attacker. The computer then generates another random
twenty faces based on the chosen one(s). Four cycles of this is
usually enough to elicit "That's him!" The method is much cheaper,
faster, and more reliable than using the human sketch artist.
I am told that PhotoShop already has some image manipulation
functions which work in the same explore-and-pick fashion. Since in
many data rich environments, we do a lot of exploring and rather
little picking, this style of interaction may have a variety of uses.
Imagine descending through a series of portals labeled with possible
choices in designing a new object. If backing out of those choices
undid them, then the exploring would feel safe and easy. The only
serious decision to make is whether to keep one of the constructs,
and even that is easily discarded as well.
Is fonc the right way to explore notions like this?
Is fonc ready for an aging boy-programmer who learned fifteen or
twenty languages, but not Smalltalk, to play with this year or next?
Dr. Parnas thought I understood the module part of OO programming,
and inheritance seems not too hard. Incidentally, the biologists have
just realized that our cells and genes exhibit the reality of
multiple inheritance, containing material from ancient viruses as
well as the symbiont mitochondria.
Richard Karpinski, Nitpicker [EMAIL PROTECTED]
148 Sequoia Circle, Santa Rosa, CA 95401
Home +1 707-546-6760 Cell +1 707-228-9716
http://cfcl.com/twiki/bin/view/Friends/Karpinski/WebHome
ps Put (or leave) "nitpicker" in the subject line to get past my spam
filters.
_______________________________________________
fonc mailing list
[email protected]
http://vpri.org/mailman/listinfo/fonc
