Eric Hellman wrote:

> We need good global metadata catalog/registries. Which of today's
> catalog functions will require a local institutional catalog tomorrow?

I think this is an interesting question.

My opinion is that the libraries of tomorrow will have a distributed catalogue: some of it local, some of it non-local.

It makes sense to me that libraries invest in describing resources which are produced locally (i.e. as publishers, or institutional repositories), as well as in cataloguing resources which are locally appreciated in a distinct way. Institutions might just want to deal with a few facets of particular local interest, i.e. tagging resources according to some local vocabulary, and otherwise rely on other catalogues for their metadata.

As well as their MARC records, each library of the future will collect a growing variety of metadata about their holdings, lending histories, reviews contributed by users, clusters harvested from usage patterns, or from full-text transcriptions, etc, etc, all of which they will want to make use of in conjunction with other catalogue data. Some of this data may be of general utility, but other data will be local in scope (and privacy laws may prohibit some data exchange in any case).

These distinct information systems have to be easy to federate, efficient, and work transparently to users, so that institutions can confidently "walk on two legs"; relying on outside services for some data, while concentrating their own efforts on the areas where they can add most value to their users.