I personally have a particular universal resource concept that I
was able to discuss with a few people privately, but which could be
the logical outcome of a workshop such as this if enough people
started to understand the idea.  What I would like to do is to open up
a discussion for a major visualization initiative that my work with
Priscilla Frisch, combined with partiview and Stuart Levy etc., could
lay foundations for:  here is the idea I propose - e.g. to set up as a
national repository (managed by the Kavli or some similar resource - I
don't think the Hayden is quite right):
   -> Following Szalay's general data access framework (or something
like that), put together a pixel quadtree of the celestial sphere,
with a mechanism for locating, in principle, a cumulative spectral
plot of all the spectral data at each pixel, or averaged pixel, in the
quadtree.  I mean EVERYTHING - all spectral data; the quadtree handles
the variation in resolution throughout the sky.
   -> Assemble COMPETING graphics models for the material and structure
lying along the radial line for each pixel.   The idea is that anyone
in the community can contribute a model, and they would compete
Darwinianly for survival of the fittest - the community would pick
and choose alternative models.
   -> HOW to use these models and make them compete is the final, and
key idea: a public set of processing and computer graphics rendering
tools, with the partiview paradigm, would be maintained on the site to
COMPARE the actual computer graphics rendered spectral data in a
SIMULATED pixel (image = block of pixels on the celestial sphere of
course) to the REAL DATA from the database of Sloan-style-organized
data that corresponds to the measure spectra at the pixel.  The models
then compete with the rendering process to add up all the
contributions from all the material along a particular ray to create a
simulated projection into whichever real spectral image data the model
is supposed represent.  The rendered set of models on any ray that
best describes and matches the real data wins until someone submits a
model that has a better explanation.
   This basically provides a permanent context for the submission,
visualization, and evaluation of astrophysical measurements, models,
and simulations for every new block of instrumentation for the
foreseeable future, and integrates the entire future of astronomical
visualization into one public, sharable, competitive framework.