[Taxacom] LSID versus names

[Richard Pyle]

Yikes!

I'm (reluctantly) breaking my relatively long-standing (though imperfect),
self-imposed abstention from Taxacom in recent months.  The bulk of this
discussion is either silly or grossly misdirected, so I'll ignore that bit
(suffice it to say that Doug Yanega and Neal Evenhuis represent my
perspective the best).  However, having just jumped into a
distantly-relevant topic on another email list, I can't let some of this
stuff go unchecked.

First of all, it would be helpful (to me, at least) if we used a few terms
correctly.  An LSID (Life Science Identifier) is a particular *kind* of GUID
(Globally Unique Identfier).  Another *kind* of GUID is a UUID (Universally
Unique Identfier).  There are other *kinds* of GUIDs (e.g., DOI).  To put it
into terms more familiar to this audience, it might be helpful to think of
the "GUID" as the genus, and "LSID", "UUID", and "DOI" as member species of
that genus.

No wait....on second thought, that will probably not be helpful.  Nevermind.

In any case, this is *not* an LSID:
B26AB2A6-972F-4A18-9D1D-486A980CF80F

It is a UUID; and in this case, it was generated in the context of ZooBank.
For complicated reasons that I touched on in my post sent to the other email
list (but which are too technical to delve into here), ZooBank represents
the UUIDs it generates in the form of an LSID.  In this case:
urn:lsid:zoobank.org:act:B26AB2A6-972F-4A18-9D1D-486A980CF80F

*That* is an LSID.  You can tell because it starts out with the prefix
"urn:lsid:...".  As a side note, you can also tell from the LSID (although
you're technically not supposed to) that it was issued by ZooBank (the
"zoobank.org" part), and that it represents a nomenclatural act (the "act"
part).

But I digress.

So, if you think of it in terms of the genus/species analogy I mentioned
(but then retracted) above, it might be helpful to think of ZooBank as a
taxonomist who is treating "UUID" as a subspecies of "GUID LSID", to yield
"GUID LSID UUID".

No wait....on second thought, that will probably not be helpful.  Nevermind.
Again.

So that's the techno-nerd part of my commentary; and as pedantic as it may
seem, this is as much to my mind as annoying as fingernails on a chalkboard,
as regular participants of Taxacom would find the conflation of taxonomic
names with taxonomic concepts.  Eeek!

Getting more into the taxonomy side of things:  I have been playing with
scientific names and computer databases since about 1988 or so; and if there
is one thing that has been ABUNDANTLY clear from my experience, it is that
text-string scientific names are WONDERFUL as (mostly) unique identifiers
for use by human brains, and HORRIBLE as unique identifiers for use by
computers.  There is a reason why ITIS did not follow the NODC convention
for generating identifiers (am I dating myself here?)

Here's just one of MANY examples why text-strings are not good as unique
identifiers for names:
http://www.google.com/search?q=Drοsοphila+melanogaster
(if you get more than one result, you cheated)

Conversely, the various flavors of GUIDs (LSIDs, UUIDs, DOIs, etc.) are
WONDERFUL (to varying degrees) as unique identifiers for use by computers;
and HORRIBLE (to varying degrees) as unique identifiers for use by human
beings.

The answer is not rocket science:  the answer is for humans to continue
using scientific names as (mostly) unique identifiers, and for computer
databases to use proper GUIDs (of one flavor or another) as unique
identifiers.  This is why I take great exception to Chris' earlier
insinuation that GUIDs that are optimized for use by computer systems are
somehow going to replace scientific names:

> The point is that a couple of small changes would have adapted the
> traditional ICZN to the modern computer world. BUT NO, the traditionalists
> refused to change. So, instead we are going to get ZooBANK with
>
> B26AB2A6-972F-4A18-9D1D-486A980CF80F E9541A64-EC44-4856-B2AB-
> B4E8400358F8 &
> 1FDB5781-C8A0-4088-8D18-DCD5BB01C548 for our unique name! [or in the
> old fashion system, Rhopalopsole exigupspira Du & Qian]

So.... a couple things:

1) UUIDs, LSIDs, and any other GUID used to represent a taxon name are not,
themselves, intended as "names".  Borrowing from Neal's example, the
following are identifiers that are used by various informatics systems
(i.e., computer databases):
8C466CBE-3F7D-4DC9-8CBD-26DD3F57E212
BD084528
08320
063463
2682
H00634762
446007073
219-68-2664
http://hbs.bishopmuseum.org/staff/pylerichard.html
http://www.linkedin.com/pub/richard-pyle/5/30a/234
...etc...
[note: some of those numbers were adjusted slightly, to protect the guilty]

As effective as all of these identifiers are for use by computer systems to
distinguish me from all my homonymous brethren on this planet, I've no plans
to abandon the moniker "Richard Pyle" when I introduce myself to new people.
My *name* is "Richard Pyle", no matter that different computer systems use
different GUIDs to represent me.  Likewise, the *name* of the new species
described by Du & Qian in their 2011 article in ZooKeys is "Rhopalopsole
exiguspina", and even if later taxonomists combine it the species epithet
with a different genus, nobody will ever think of its name as
"B26AB2A6-972F-4A18-9D1D-486A980CF80F".

2) I'll give Chris the benefit of the doubt and assume that the reason he
included three UUIDs (instead of one) in his example was not so much to make
it look silly (as Neal suggested), but rather as a way of representing the
"Du & Qian" of the name+authorship "Rhopalopsole exigupspira Du & Qian".  As
Neal pointed out, this belies a fundamental misunderstanding of how these
identifiers are used.  In the case of the text-string name, you *NEED* to
add the "Du & Qian" for uniqueness, because there might be (now, or in the
future), a homonymous "Rhopalopsole exigupspira" under a different
authorship.  Let's ignore the historical reality that authors alone do not
disambiguate all homonyms (indeed, Author+Year+Page appended to the name
doesn't even eliminate all duplicates); the point is that UUIDs are (for any
practical purpose) truly globally unique (again, as Neal already explained).
Hence, it is silly to insinuate that you need the UUIDs for the authors when
you have the UUID for the name.  This is taking a weakness of the
name-string-as-identifier approach that Chris seems to advocate, and falsely
transferring it to the computer-based identifier system.

I had some more points to add here, but this message has gone on long
enough.

Aloha,
Rich



This message is only intended for the addressee named above.  Its contents may be privileged or otherwise protected.  Any unauthorized use, disclosure or copying of this message or its contents is prohibited.  If you have received this message by mistake, please notify us immediately by reply mail or by collect telephone call.  Any personal opinions expressed in this message do not necessarily represent the views of the Bishop Museum.