Re: Some Questions About VRML

Mark Waks (justin@dsd.camb.inmet.com)
Thu, 23 Mar 95 15:13:49 EST


Len asks:
>1. What is VRML for? "bringing VR to the masses" isn't too
>specific. DOOM, 3-D Dinosaurs, Myst, et al. do that at the level
>the masses care about such things.

Far as I'm concerned, the answer is straightforward: cyberspace. My
interest grows out of the simple observation (which I think I and a
zillion other people made simultaneously a year or so ago) that the
Web + VR = cyberspace. Of course, it probably won't come out
resembling the common media version much, but I still think it has
some *very* intriguing possibilities for changing the way we view the
Net (and, indeed, all communications-oriented media).

No, I don't have a clear idea of where it's going to go. On the other
hand, nobody had the *foggiest* notion of where HTML was going to
take us, either. I mean, you write:

>Tim Berners-Lee had a task and a user
>base in mind: researchers communicating across the Internet.

And while that was the real user base, HTML was a nice little
curiosity. Then came Mosaic, and *boom* -- it revolutionized the
Net (and I mean that *quite* literally; the Net is a very different
place than it was even a year ago, for better and worse).

This strikes me as another likely "steam-engine time" technology,
one where we're only going to really understand the potential when
we can get it out onto peoples' desks and let them play. I *suspect*
that it's going to be pretty revolutionary as well (although probably
slower to catch on, since it's more resource-intensive in every way
than HTML). We'll see.

Anyway, this is why I care particularly about getting prototypes out
there *now*; we're really not going to understand what we need until
people start screwing with it. Yes, this opens us up to the sort of
developmental chaos that HTML has gone through (and continues to go
through), but I'd rather a little chaos than being stifled by
attempts to get it perfect upfront...

>2. Who will use VRML? Programmers or VRAuthors?

Ultimately, scene builders. I think it unlikely that, after about the
second generation, anyone but serious hackers are going to program in
raw VRML. It's useful to have it be comprehensible, so that hackers
*can* work in the guts, but I take it as an article of faith that 95%
of all developers are going to work in higer-level tools pretty quickly...

>3. How complex can a VRML simulation become? If the PC
>environment is your target, will VRML be too simple for
>the industrial designer and too difficult for Joe Homepage?
>Where is its niche in the scale of systems?

I expect this to evolve upwards; I'd expect us to start with simple
simulations, and gradually increase in directions as the "market"
demands...

>4. What will VRML apps be used for? Touring a museum
>is great. Looking around the space shuttle is great. Using
>VR as an interface to find a file ain't so great. Like the ancient
>debate over GUIs and command line prompts, it was always
>easy to show that for efficiency, entering whereis and
>cd was a lot better than click_on_icon_that_might_be_it and
>explore_next_set_of_icons. Hieroglyphs and ideograms
>are poor ways to express complex concepts over time.
>That is how text came into being.

Yes and no. There is considerable potential in VR worlds for
making things simple, by relating them ever-more-accurately to
the real world. This, I think, is where the media version of
VR is basically rubbish. While I suspect there will be some
cool innovations, I'd say that it's a good bet that most real
VR will look a *lot* like the real world.

Comparing it to old GUIs is misleading. The early GUIs weren't used
much because they *were* far harder to use than simple text. But
that's gradually been evolving away, as designers start to
understand the "ergonomics" of a good GUI, and develop systems
that are more and more intuitive to the average user. The DWIM
factor of the average GUI (that is, the level to which is just
does the right thing when you click somewhere) has been increasing
quite steadily. I've been finding, over the past few years, that
the current GUIs *are* awfully nice for many tasks. (This from a
long-time keyboard snob.)

Similarly, I expect the early VR systems to have a relatively low
utility, and be basically toys, like the early Macintoshi. But
that *will* evolve away, as people learn how to use the content-rich
environment to make tasks intuitive. I really believe, deep in my
gut, that an interface that is closer to the real world will be
faster to grasp and use for most people, once we understand which
real-world limits deserve to be overcome...

Ultimately, there's an element of faith here. I suspect that what
we're putting together here is going to be nothing more than a cool
toy for a year or two, as we internalize the lessons we're learning.
Sooner or later, though, I suspect we're going to start finding
applications for which it really works nicely...

-- Justin

Random Quote du Jour:

"(cosmologists, after all, undergo years of training so that they can say
things like `When the universe was the size of a grapefruit' without
blushing or laughing)"
-- The Economist