From: Jeff Pierce [jpierce@cs.cmu.edu]
Sent: Wednesday, August 04, 1999 12:46 PM
To: 3d-ui@hitl.washington.edu
Subject: Re: Idea for discussion: violating assumptions

Rudy raises an excellent point.  I certainly don't intend to imply that
realism is always bad; for certain VR applications it's essential.  And
I've already added a 4th category for assumptions: those that we should
never violate because it would cause too much confusion.

In fact, I think the strongest new techniques I've come up with so far play
with violating the assumption that appearance and reality are different.
Optical illusions, photos, windows, maps, mirrors: if we adopt a slightly
more "naive" interpretation of these devices we can actually make them more
powerful.  Does anyone remember the scene in Dances with Wolves (I think)
where Costner hands his telescope to the Indian next to him to look at some
buffalo, and the Indian tries to reach to grab them because it looks like
they're within reach?  With VR that's the kind of thing we can actually do,
and as Rudy points out it leverages devices and ideas that are already
grounded in the real world.

Jeff

At 11:12 PM 8/3/99, Rudy Darken wrote:
>Consider all the possible techniques imaginable on a continuum with
>"perfectly real" on one extreme where we exactly duplicate the real world,
>and "perfectly abstract" on the other end where the opposite is true. Jeff
>argues that perfectly real is not necessarily a worthwhile goal (I tend to
>agree). However, keep in mind that for lots of training applications realism
>tends to be of greater importance. But also be aware that perfectly abstract
>is an equally bad idea for the simple reason that by its definition, a
>perfectly abstract interaction technique has no basis in the real world and
>is therefore unusable (or unintelligible to the user). All interfaces we
>know of have some grounding in the real world (e.g. metaphors) which is what
>makes them useful to us. I think Jeff is arguing for something in the
>middle. It would seem to me that the thesis is hiding in finding just where
>in the middle is the optimal place to be. I would suggest that "optimal" in
>this case has to be quantified in terms of usability characteristics (i.e.
>learnability, efficiency, etc.)
>