policy issues

[Tom McGlynn]

I think the emphasis of this document is slightly different
than what I'd like to see.  The issues that are brought up
are fine: what concerns me is the 'we'.  Rather than
trying to mandate some policy for all centers and registries,
it seems to me that what we need is to provide the model
and structure through which registries indicate what their
policies are rather than dictating some IVOA-wide policy.
My quick reaction to the policy concerns is given after each paragram.


	Tom

> 
> ·         Do we need to control or restrict in any way the ability of 
> groups or individuals to register resources and services?  Thus far we 
> have tried to be as open and inclusive as possible, though we then have 
> the risk of people registering poor quality or inappropriate services, 
> or of permitting some sort of malicious corruption to the registries.
> 

We should impose a policy, we should allow registries to
describe their policies. A registry should be able to indicate
that it is open to all entries or requires some screening
process and should indicate what that service entry or screening process is.

> ·         Should we develop some sort of VO endorsement for registered 
> resources and services, i.e., a “seal of approval”?  Such endorsements 
> could be managed by the various national projects, with a national 
> endorsement serving as an implicit IVOA endorsement for IVOA member 
> organizations.  What process should we use to review and grant such 
> endorsements?
> 

There should be a mechanism for indicating that a registry
entry has been endorsed by some validating authority,
but we should neither mandate nor forbid registries to
use such authorities, nor should the registry
be particularly concerned how or when the endorsement
is given.  The registry may need to worry about
being able to authenticate the endorsement if we
actually want to care about security for these things.
That may be overkill for astronomy data.

Registries should be able to choose which endorsements they use.
There may be many endorsing authorities.   This can
get recursive since there could be registries of
endorsing authorities, and authorities that endorse
endorsing authorities.

While this sounds kind of Alice-in-Wonderland'ish, I think
that actually what Bob may have in mind.  E.g., there would
be national endorsing authorities, which are endorsed by
some single IVOA authority.  A user asking for only IVOA-endorsed
data can look for services endorsed directly by the IVOA or by any of
the endorsing services that it has approved.

Note also, that endorsments mechanism may need to be conditional
and/or quantitative.  E.g., data may be endorsed for astrometric
purposes but not photometric, or the endorsement may indicate
that the astrometry is good to 0.1" and the photomtry to 0.1 mag.
The boundary between endorsements and descriptive metadata is
potentially quite thin.

However, I don't think this really needs to impact the registry
protocols much.  Registries just need to be able to include
endorsements as part of the descriptive metadata for a service
and allow users to query against it in much the same way.

The process to review and grant endorsements is completely
divorced from the registry to my mind.


> ·         We will need to have some editorial control over the 
> registries, e.g., to remove inappropriate and inactive resources and 
> services, and to correct erroneous entries.  Who should have this 
> responsibility, and how is editorial access controlled?

As for editorial control: that is up to the registry
provider.  I don't see this as being some master registry
that provide strict control over.  The policies at,
say, the CDS' registry may be very different from
one at MAST.

> 
> ·         Should we, and if so, how do we, control the number and 
> quality of registry servers/ services themselves?  How do we assure that 
> multiple registry services are properly synchronized?  Do we care if 
> different registries focus on different types of resources and services, 
> or should all registries be 100% inclusive?

Who is the 'we' here? If the IVOA or one of the national projects
wishes to build a registry ourselves through some mechanism
it should feel free to do so.  However I do not feel that we should be
getting into policing.  There is no reason why all registries should
be the same.  There may be site registries which contain only
a set of local services.  These may be harvested (by OAI or some other
mechanism) to get entries for more general registries.  There may
be registries of only SIA services or data of interest in ISM studies
or whatever.  Multiple registries  enable serving diverse communities.

So long as we put a requirement on registries such that they
are harvestable, it's inevitable that someone will build a registry
that combines data from other extent registries.

One specialized registry that will likely be needed is a registry of registries.
This might indeed be something that the IVOA wants to build itself,
but even here, I don't see that this needs to be exclusive -- others
can and probably will build such registry registries.