Roy has given a very useful structure to the debate. As both a data user and a data provider, I think that the present lack of even the most basic information for many data sets, stops them being used by anyone who is not already familiar with the data. In other words, the main worry should not be that we are asking for a little more from data providers, but that we are asking them to waste their time if they go to the trouble of installing the software needed to serve their data to the VO, but don't describe it, because in all honesty it will mostly be used by their existing clientele who would probably use the data by some other route anyway.
When looking for data myself, or working with science users in AstroGrid/Euro-VO/RadioNet workshops, we are asked time and again for selection by position, time and waveband - even within rough limits. Resolution and sensitivity are also mentioned. It ain't rocket science... we spent a lot of time developing the Registry DM. We are now developing other models which are even more complex and maybe one could argue that some models are running ahead of forseeable implimentations, but that is _not_ true for the Registry.
In particular:
> We have come a long way from Simple Cone Search. If a developer wants to
> expose a catalog by IVOA service, it turns out to be pretty complicated, with
> all the MUSTs and SHOULDs
That's why we keep MUSTs to a minimum!
> Questions
> (1) Can the IVOA support this level of complexity?
As long as the coarse, Must and Should levels are clear, then frankly sometimes if is quicker to leave Mays in place until the first levels are got to work, then they can be refined.
> (2) Where is the robust software that leads the service developer
> through the maze?
That is the most important question. One thing which may help is that many data providers already have metadata in a database and we should be concentrating more on providing mappings for this (e.g. observing log) than trying to squeeze it out of FITS headers.
> (3) Can the IVOA Recommend any of the above standards in the absence of that
> implementation software?
Yes. a) We can't develop software without standards
b) There are prototypes - I've seen NVO forms and we have several
tools under development in Europe
> (4) If somebody makes a service that has fabulous science data but NONE OF
> THE ABOVE, should the IVOA reject it?
No, there are standards to describe the level of comliance with standards, too! But there would be a good argument for a VO strongly offering assistance to describe the data....
all the best
Anita