This document reports and analyzes the results from a survey
of 20 institutions that are actually developing grid applications.
Besides cataloguing what types of applicaitons are being
developed, which compilers and debuggers are in use, this
survey extracts some interesting rankings for desired properties
when designing large-scale systems and grid applcations, e.g.,
scalability, robustness, etc.  The exact numerical rankings
reported here are less important than the groupings that emerge.
For example, scalability was clearly a more important design
goal for large-scale systems than a recovery mechanism.
Surprisingly, single-login was pretty consistently ranked low
for grid applications. 

I suspect that these priorities would change as the commonly
deployed grid infrastructure changes.  That is, as more grid
infrastructure becomes more commonplace, what grid
developers view as important will change accordingly.  This
implies that it might be useful to resurvey periodically.