[4eyes] Engineering vs Research

[Cha Lee]

I"m sitting here listening to the panel at VR and it is both
compelling and frustrating.

Marc Erick Latoschik (organizer) just finished his introduction and
here were some of the points he addressed.

Why is there such a steep drop-off in the amount of engineering papers in VR?
		It is solved, hence not necessary anymore? This comment was made in
sarcasm I think.

		It is not research?
			Tell it to the OOPSLA and SPLASH communities.

		It is too expensive
			More expensive otherwise.

		It is not interesting
			No, this is the heart of CS.

Steve Feiner just finished and I missed half of it (bathroom) but the
gist was that it was "easier" and more "novel" back in the day to
submit a system paper. That each new system was indeed novel and that
now they are not so. Then he gave a weird definition of VR/AR but that
is something else.

Dieter Schmalstieg argues that a good system paper must be "more than
the sum of its parts." I get the feeling he is on the side of the
"research worth" argument. But he also acknowledges that good software
benefits the group but that it is hard to compromise between software
maintenance with the driving force of grad students completing the
PhD.

Carolina Cruz-Neira talked about her work on the CAVE. She argues for
a clear divide between software tool developers and application
developers. An argument for software research, is that the US
government has recently made a LARGE investment in encouraging
software engineering research. She didn't have any slides :/. Need
slides!

Questions:

Should there be a standard VR toolkit/library/language?
- Cruz-Neira: finite way to do all this re-engineering. We should make
an effort to make a standard. Arguments like "mine is better than
yours" hurts the community.
- Dieter: there is a distinction between free-ware and commercial
stuff. We should make libraries and not systems. Complexity must be
low.
- Feiner: GPL sucks. Use LGPL, BSD etc. This will make it more available.
- Marc: disagrees with Carolina and Dieter. A lot of
systems/programming language words arguing against. Gist is that there
is no "book" on how to develop VR systems b/c no one really knows how?
We just hack and put together packages. For example a thousand
different scene graph formats.

What would you say to people on how to submit a system paper? Since
its a topic of VR but a lot of VR system papers get bashed?
- Feiner: don't describe just what you did? No entitlement. Describe
what is different and why should we care? AT THE VERY BEGINNING OF THE
PAPER. Ridel's "How to Write a System Paper", 1983.
- Carolina: define the scope of your system. Systems do not
necessarily require A USER STUDY! (I totally agree!)
- Marc: agrees
- Dieter: disagrees, that a user study is very necessary sometimes!
- Guest: maybe this is not the right venue anymore. GDC is right now.
- Robert: rebutal, no this is not the reason why there isn't a lot of
systems papers. Only this year. Mostly we don't conflict with GDC.

Now we are discussing the difference between gaming and VR community. Tangent.

I think this topic is something we have talked about between us (the
grad students) and I am on both sides of this argument so I thought I
would blog on it while this was going on.


-- 
Regards,

Cha Lee
PhD Candidate
University of California, Santa Barbara