FW: [SEEK-Taxon] looking into global unique identifier systems

[thau@learningsite.com]

Ok, here are some responses.

Handle System CONs
> > - speed should be solvable

Yeah, I'll work on that.  Shouldn't be too tough to deal with.

Lack of metadata
> > - this really depends on what data needs to be referenced
> > through the handle. If it is just the taxonomic concept
> > information there shouldn't be a problem, as it is the
> > metadata is identical for all concepts.

The metadata support in LSIDs lets you represent relationships between
LSIDs and gives standards on how to retrieve this information, and some
standards on how the information is formatted.  This is pretty nice
functionality, but since this is the function of the taxonomic name
server, maybe it's best left there.

> > - not so sure why you think it would be restricted to one
> > authority (your number 1883) although I would personally
> > prefer it this way

Yeah, you're right.  There could be more than one authority, and then it'd
be exactly like the situation with LSIDs.  You'd have to keep track of
which Handle System registrars you respect, and there'd be no way to
ensure that two registrars didn't issue guids for the same concept.  

What I was thinking about here was that there's no support for the
scenario where ITIS would have Handles like 1883.1, Species 2000 would
have 1888.2, etc. DOIs have this ability, giving greater control to the
data providers. The handle system doesn't have it natively.  We could
simulated it by assigning handles like 1883/1.3994 and 1883/2.4483, but
then we'd have to write some software that gave the data providers
limited access to modify information in our handle system.  I guess that's
a scoping issue to be addressed later.

> > I think from the users (people using GUIDs to resolve a
> > taxonomic concept) point of view the methods are equivalent.
> > It seems a pity though that only plug-ins for windows browser
> > exist in both cases, but this can change easily and for the
> > purpose of SEEK the lookups are probably hidden in specific
> > client software anyway.