Ricardo,

Regarding LSIDs specifically, our biggest challange is to find an

appropriate way to define stable byte-identical data for our objects. We have therefore been thinking mostly of turning everything to metadata. We would be very interested in knowing what your thoughts are regarding that issue.

Some comments on metadata and a few observations on the process. Hoping it lands before you break.

I did not find persistence listed as a use case for GUIDs. But persistence of both GUID and data object should be one of our most important goals. LSIDs biggest advantage lies in the fact that the data should not change. This means that Donald's requirement for equivalence of ID and data object can be satisfied as long as we, even if much of it is duplicated as RDF metadata, also preserve the record (TCS, ABCD, DC,...) as data. LSID metadata will be more than an objects data values - extending to relationships between objects and links to alternative representations etc - and rarely complete for any one record. Rarely persistent. It is hard to imagine the benefit of a full ABCD, TCS or UBIF record as RDF.

GUID metadata != GUID but GUID == Data object. Data object == object metadata but GUID != object.

With persistence, facts based on data behind GUIDs may be compared and evaluated. Persistence enables the establishment of protocols for consolidating after market validation and correction of distributed data sets. The mapping of duplicates. Our data use agreements can specify that the GUID remains with an object at all times and that update or correction of that object will require a new version number, registration and appropriate documentation using metadata, preferably for both new and superseded versions.

We can distribute maintenance of our primary sources to obtain global benefit from the value added after extraction.

Without it ... we already have data extracted, altered, used for analysis then dumped, or worse ... lauded as superior, re-distributed ... languishing...re-used years after it has passed its use-by-date.

If this process, as I perceive it after reading through these 240 posts, is going to work, then the custodians of object/concept/name metadata among us will have to give freely of these data. APNI, IPNI, AVH, BPH, etc will simply have to accept that they cannot participate, or even survive, if they are not prepared to freely distribute any or all of their data holdings to those that ask. But distributed with the above data use caveat conserving GUID and data object equivalence: GUIDs in this scenario may function as digital signatures establishing custodial rights wherever a record is found. Always pointing back to whence it came. Easy to check, update or communicate correction. Returning value to the source. Perhaps even making it worth the while and expense. Giving us time to complete the task.

greg -- Greg Whitbread ghw@anbg.gov.au ANBG/CPBR/ANH