Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!crdgw1!crd.ge.com
From: meltsner@crd.ge.com (Kenneth J Meltsner)
Newsgroups: comp.graphics.visualization
Subject: Re: Out of the lab, into the classroom
Keywords: reuse fabulous
Message-ID: <18252@crdgw1.crd.ge.com>
Date: 4 Apr 91 21:26:22 GMT
References: <1991Apr2.202227.13796@agate.berkeley.edu> <1991Apr3.172156.3960@nas.nasa.gov> <1991Apr3.193914.4569@agate.berkeley.edu> <1991Apr3.215031.7763@nas.nasa.gov> <1151@durer.cme.nist.gov> <18226@crdgw1.crd.ge.com> <1991Apr4.195401.22442@eagle.lerc.nasa.gov>
Sender: news@crdgw1.crd.ge.com
Reply-To: meltsner@crd.ge.com
Organization: GE Research and Development Center
Lines: 125

|>From: tttron@escher.lerc.nasa.gov (William Krauss)

[stuff deleted]
|>
|>I really can't say what "type" I am so I'll refrain from comment 8~)
|>(Though I get the impression you're a "disgruntled visualization customer - 
|>perhaps you haven't heard of our money back guarantee!")

I'm actually a disgruntled expert systems customer, but I've been
seeing behavior in the visualization types similar to that exhibited
by the expert systems types in their heyday.  I was pretty dismayed,
for example, by the Visualization '90 Proceedings and the number of
content-free papers.  It's hard to quantify, but I think you're
definitely going need to shake out some of the under-sized apples (if
not actually rotten) before the field can be appraised properly.  I
understand this is all pretty new, but I'm tired of the hype.  And I'd
certainly be cautious about doing my graduate studies in an
ill-defined discipline like "visualization."  Or "knowledge
engineering," for that matter.


|>>"We've done the hard
|>>part: here's a renderer, some data display tools, etc.  Now go and do
|>>good science with your data."  I'm not sure they realize the hard part
|>>is just beginning, and that the tools they've given the scientists
|>>don't help a lot.  Visualization can help reduce the information
|>>overload, but it doesn't help with the insight or analysis to make the
|>>conceptual leap from the video to understanding the real world.
|>
|>I can give a person a hammer, but that does not a house build. I can
|>also give him/her ALL the TOOLS, supplies, manuals, instructions, 
|>(ad infinitum) but it will never BUILD the house. I beg to differ on the
|>"conceptual leap" which you claim is lacking. Visual feedback IS a
|>conceptual BRIDGE - it may not be the best one, it may not be the ONLY
|>one - but it IS a bridge nonetheless (I don't believe too many folks
|>can deny this). Granted its use is the researcher's preference. You
|>personally may not find them useful, but there are
|>MANY who do.

The visualization image or movie is used to make the subject clear
enough that the *user* can make the conceptual leap.  Without that, he or
she is simply putting familiar results in prettier packages.

|>
|>>...automatically construct graphs from
|>>data, graphs that would automatically display the correct variables in
|>>a pleasing and informative way.  Too bad nobody has any programs to do
|>>that sort of thing on the market.
|>
|>Your concept of "pleasing" may not necessarily be someone else's. ("De gustibus
|>non disputandum est" - one cannot dispute taste). Ahh, one might claim that
|>"scientific visualization" is not ART - that could be an interesting thread!!

Fine.  Let the user specify some preferences and allow the program to
work within those.  I'm flexible (but the software I use rarely is...).

|>>Visualization people will only be useful when they understand that
|>>they are actually one of the least important parts of the whole
|>>process.  Critical, yes, but not the part that actually contributes to
|>>the scientific world or towards a solution of an engineering problem.
|>
|>I am an engineer by "trade" who happens to enjoy being a "liason/interpreter" 
|>between the "engineer types" and the "visualization types." Different people
|>*SOLVE* problems in a variety of ways. In addition, people perceive (see, 
|>interpret, etc) in many ways. I don't think it's reasonable to assume that
|>there will be a definitive "visualization tool" that will allow EVERY
|>researcher complete control of his/her environment/results. If one thinks of 
|>"visualization" (whatever that nebulous word means) as a type of
|>microscope its meaning becomes a bit more focussed (pun intended). Graphical
|>feedback into one's simulation (or what have you) allows one NOT ONLY to 
|>"reduce the information overload" BUT to place it into a more "digestible,"
|>understandable form. That form (and more importantly its USEFULLNESS), 
|>however, is ultimately up to the investigator. That places the importance
|>and responsibility on the engineer/scientist's shoulders - exactly where
|>it should be. The "visualization types" provide some of the tools (some of
|>us "engineer types" happen to be "visualization types" too!!) and the 
|>scientist uses/abuses the tool to meet his/her own requirements (or develops 
|>his/her own). 
|> 

I don't want a tool that allows me complete control over everything.
I'd like a magic tool that'd extract the interesting bits and show
them to me, that would show me where models diverge from experiment,
or where experimental results appear to be inconsistent.

|>I guess you could say that "visualization types" are "Computer
Optometrists" |>- helping folks see what THEY'RE doing (not to SEE for
them) and suppyling |>"corrective lenses" for their "visual acuity."
|> |>Another thing to remember is that some of us "visualization
types" also |>work DIRECTLY in conjunction with the investigator -
essentially becoming |>an EXTENSION of his/her perspective by
developing these tools (whatever they |>may be). There's much less of
an "us versus them" attitude than one might |>think. What about
developing VALID models? Is that not an INTEGRAL portion |>of the
design/analysis process? So then "visualization types" can be an
|>INTEGRAL part of the "picture" 8`).

I'm sure visualization personnel are valuable parts of their teams,
but I'd simply like software that didn't require a specialized staff
person's time to run and feed.  I can buy good statistics programs,
and I can plan out a statistical strategy with a statistician if I'm
doing something out of the ordinary, and then use the programs.  I
don't need another middleman to run the program.  Why can't
visualization be the same sort of thing?  What value does a
specially-trained visualization expert contribute to the process?  I'm
not saying "visualizers" don't contribute any value, but I need this
sort of question answered before I'd plan a project to use one.

By the way, I have the same questions about "knowledge engineers" and
their value in scientific and engineering activities.

I think visualization as a tool may have a lot to offer.  I can't
argue with methods that show me the results from my simulations more
clearly, or bring out details I'd miss otherwise.  But is this a
science or an engineering discipline?  Is this something we should
have doctoral programs for, or should we encourage properly trained
computer graphics people to specialize in this "bridging" function?  I
(honestly) don't claim to have an answer to these questions.

===============================================================================
Ken Meltsner                        | meltsner@crd.ge.com (518) 387-6391
GE Research and Development Center  | Fax:  (518) 387-7495
P.O. Box 8, Room K1/MB207	    | Nothing I say should be attributed
Schenectady, NY 12301               | to my employer, and probably vice-versa
=================Dep't of Materials Science, ACME Looniversity=================