Paleonet: Scientific death-knell of databases?

[David Campbell]

WoRMS has been quick to respond to corrections when I have found them.  In
that it is somewhat unusual - often there is no way to submit a correction
to a database, or it takes much searching to find a way to submit
something, only to get a reply that passes the responsibility off on
someone else.  When I discovered that a relatively automated database had a
photo of a completely different snail with the specimen entry that I was
looking up, I got a reply that I had not provided the correct
identification number for the entry in their database.  The number I had
submitted was that used as an identifier in a linked database; neither the
reply nor the entry indicated which of several numbers was the key one.
Some time after I replied including all of the reference numbers I could
find in the entry, it did get fixed.  WoRMS also has a number of taxonomic
experts closely involved, again unlike many other projects, though
something seems to be amiss for some of the brachiopods.

Paleobiology Database has some taxa that have been carefully checked and
some that are a mess.  In particular, people seem to confuse homonyms (and
near-homonyms with common misspellings) more often than properly sorting
them out.  And many taxa have been entered straight from a publication
without any evident familiarity with the taxon (for example, many genera
from Sepkoski's compilation that were entered floating as unclassified
names in whatever level of classification he gave).  Uncritical entry of
wastebasket taxa has made many bivalve families have highly inaccurate
temporal ranges, for example.  In turn, uncritical use of the dates from
PBDB is not uncommon as a substitute for research when calibrating
molecular "clocks".  Also, PBDB is growing well, but is still far from
complete in its coverage.  What biases that may cause are hard to know.

Many of the larger online taxonomic databases mostly mine each other for
information and have little original content or error checking.  Locating
the original source of an error is thus very difficult.  Often, the
philosophy seems to be to build a database, taxonomic experts volunteer
their time and expertise to provide all the information, and someone else
can then take credit for analyzing the data.  It overlooks the fact that
the taxonomic experts need some form of employment and credit.  There are
too many grand synthesis studies out there (not just in paleontology) that
simply assume reliability of data grabbed from a database, without
verifying it against original sources.

Some of the errors are the product of automation rather than any editor.  A
database incorrectly combining homonyms blamed a source database, but the
source had it correct - it was their synthesizing of different records that
was at fault.  A typo turning Pectinidens into Ens led to Ens appearing in
several databases as a genus.  Similar, Biodiversity Heritage Library is
extremely valuable as a source of publications, but the name search has a
very high rate of both missed real names and spurious hits.


On Thu, Nov 12, 2015 at 2:15 AM, Christian Emig <brachnet at aliceadsl.fr>
wrote:

> Bonjour,
>
> A freshly issued paper:
> Emig C.C., Bitner M.A. & Álvarez F. (2015). Scientific death-knell against
> databases? Errors induced by database manipulations and its consequences. *Carnets
> Geol.*,15 (16), 231-238.
> http://paleopolis.rediris.es/cg/1516/index.html
>
> As authors of the *World Brachiopoda Database* hosted by WoRMS, we
> decided to publish what may appear as the visible part of an iceberg (!)
> concerning WoRMS as well as many other online databases.
>
> Best regards,
>
> Christian C. Emig, Aleksandra Bitner, Fernando Alvarez
>
> _______________________________________________
> Paleonet mailing list
> Paleonet at nhm.ac.uk
> http://mailman.nhm.ac.uk/mailman/listinfo/paleonet
>
>


-- 
Dr. David Campbell
Assistant Professor, Geology
Department of Natural Sciences
Box 7270
Gardner-Webb University
Boiling Springs NC 28017
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.paleonet.org/pipermail/paleonet/attachments/20151112/2069ad54/attachment.htm>