> ...the scientific community has a credibility issue which it is now
attempting to address.
Take heart - the issue is nowhere near as big a problem as one might fear:
http://community.lsoft.com/scripts/wa-LSOFTDONATIONS.exe?A2=ind1601&L=BEE-L&
F=&S=&P=202898
Only about 14% of medical/psychological study results could truly be called
"irreproducible", which is not that bad, given simply issues of
recordkeeping, and the massive amount of metadata required to understand
actual techniques, as opposed to the very cursory methodology descriptions
most common in papers.
When humans and animals are the subjects (like medicine and psychology), and
there are restraints on what is permitted to be done to the subjects, due to
those pesky institutional review boards and ethics panels. But outside
medical work, "reproducibility" is most often a side-effect of improving on
the last attempt, eliminating more uncertainty in prior measurements when
shinier toys appear. The most recent example is the latest go-round to stop
the French from selfishly keeping the only real kilogram to themselves:
http://phys.org/news/2015-07-precise-avogadro-redefine-kilogram.html
http://tinyurl.com/nu33u62
A better measurement of Avogadro's number gives us a better cross-check
value for Planck's constant, which lets us define a kilogram without needing
a single physical "sample" to act as the planetwide standard, which was
never a good idea, as the last one (a platinum/iridium cylinder) lost 50
micrograms of mass over about 110 years. (Note that when everyone wants
reproducibility, they very carefully keep the toys - the black squarish
thing in the photo in the link above is the business end of the X-ray
interferometer used to get the atom distances in the new standard sphere
that represents the Kilogram.)
The biggest problem with "reproducibility" is mere funding in my experience
- there is funding for research, but not funding to preserve all the notes,
data, photos, digital data, the specialized kluges made up for that one
experiment, all the detritus left over when experiments end. It is
impossible to know which items might be the crucial ones - no one does a
forensic analysis of WHY something worked the way it did, and I'll testify
from my own experience that often, the (prototype) apparatus is a barely
working kludge, and often needs tweaking, twiddling, and frobnicating by the
person who made it work to make it work again.
The whole issue of "reproducibility" is a very theoretical discussion. No
one wakes up in the morning thinking that they can enhance their careers or
attract new funding by verifying that someone else's "breakthrough" is in
fact, a breakthrough ('cept astronomy, where verifying someone else's
sighting is considered neighborly. But the marginal cost of doing so and
time to do so is usually trivial.)
The essentially adversarial nature of the process assures that people will
attempt to disprove things rather than support things, and if unsuccessful,
will not bother to publish their failures to disprove, as Journals rarely
publish negative results. But then they don't often publish mere
verifications, either, so any time you do anything but honestly original
work, you end up painting yourself into the same corner as the USDA did when
the EPA demanded a "published" study on Tylan residues, but the journals
were all sick and tired of publishing bee-medication residue studies, so
the paper was published in ABJ, and everyone held their breath and tiptoed
around so as to not awaken the EPA.
There is always the Journal of Irreproducible Results (http://www.jir.com ),
but even it is starting to become derivative and consistently more obscure.
***********************************************
The BEE-L mailing list is powered by L-Soft's renowned
LISTSERV(R) list management software. For more information, go to:
http://www.lsoft.com/LISTSERV-powered.html
|
