[Taxacom] Digitizing Science Collections

[Doug Yanega]

John Grehan wrote:

>  > At any rate, the point of all this was about whether the uncritical
>>  synthesis of all label data for a taxon is a useful thing to have ... I
>>  guess it depends on what you want to use it for, but probably not that
>>  useful for displaying on maps that are supposed to indicate known
>>  distributions ...
>>
>
>I think in practice it just means that one records localities as they are
>until one has reason to do otherwise.

After considerable experience with all this, I've concluded that a 
different approach is required: I effectively consider all label data 
to be guilty (flawed in some manner) until proven innocent. In 
practice, one should not serve records to data users that have not 
been confirmed, and our practice reflects this; nothing is trusted 
until it has been scrutinized, and our database contains only the 
cleaned data. I've been chastised for this, but if we've gone through 
the trouble of correcting omissions and errors, there are two 
significant points to make: (1) there is little utility in showing 
the pre-corrected verbatim data; admittedly, it has a few potential 
uses, but since every specimen being databased has a GUID assigned, 
tracking which specimen is which is NOT one of those potential uses. 
(2) Funded databasing projects are ultimately under severe time/labor 
constraints. If the verbatim label data are not *absolutely 
necessary*, then the amount of time needed to enter them (even if 
it's just by taking photos) cannot be justified, as every minute 
spent is a minute taken away from *necessary* work. As such, I 
consider many such projects to have their priorities backwards; that 
is, it is evidently common to invest in verbatim data capture first, 
and then only worry about quality control and data cleanup and 
georeferencing if there is time and money left over to do it. This 
approach looks great on paper, since verbatim capture takes (assuming 
labels are legible and often-duplicated) about 1/10th the time 
required to do thorough georeferencing and data checking, and 
verbatim capture can be accomplished with cheaper and less-trained 
technicians. The alternative is to check every label, and instead of 
entering it verbatim, enter only the cleaned, corrected, and 
georeferenced data; if time and money are left over at the end of the 
project, THEN go back and enter the verbatim data (or take photos).

In essence, it comes down to this: which is preferrable - spending X 
dollars and getting 200,000 verbatim records entered/photographed, 
but not knowing for certain whether *any* of the data are trustworthy 
(and many, if not most, NOT accurately georeferenced), versus 
spending X dollars and getting 20,000 records entered, every single 
one of which is confirmed clean, trustworthy, and accurately 
georeferenced?

This isn't a hypothetical choice, nor purely fabricated figures. It's 
a real choice, and most people who have made this choice went for the 
first option. It seems that funding agencies prefer that approach, 
but I'm personally not happy with the trade-off, and not happy that 
people assume that data cleaning and georeferencing are trivia that 
can be dealt with AFTER we have everything digitized and online. 
Online data that are full of invisible errors awaiting correction are 
like a fruit basket in which one out of every ten fruit is toxic.

Sincerely,
-- 

Doug Yanega        Dept. of Entomology         Entomology Research Museum
Univ. of California, Riverside, CA 92521-0314        skype: dyanega
phone: (951) 827-4315 (standard disclaimer: opinions are mine, not UCR's)
              http://cache.ucr.edu/~heraty/yanega.html
   "There are some enterprises in which a careful disorderliness
         is the true method" - Herman Melville, Moby Dick, Chap. 82