The New York Times  [Printer-friendly version]
January 3, 2006

THE NITPICKING OF THE MASSES VS. THE AUTHORITY OF THE EXPERTS

[Rachel's introduction: "It seems natural that over time, thousands,
then millions of inexpert Wikipedians -- even with an occasional
saboteur in their midst -- can produce a better product than a far
smaller number of isolated experts ever could."]

By George Johnson

Uneasily sharing space on the top ledge of my computer browser are two
buttons I click almost daily for an information fix: Encyclopaedia
Britannica, as old and steadfast as the ligature in its name, and a
mercurial upstart called Wikipedia, in which almost anyone anywhere
can fiddle with the prose.

It may seem foolish to trust Wikipedia knowing I could jump right in
and change the order of the planets or give the electron a positive
charge. But with a worldwide web of readers looking over my shoulder,
the error would quickly be corrected. Like the swarms of proofreading
enzymes that monitor DNA for mutations, some tens of thousands of
regular Wikipedians constantly revise and polish the growing
repository of information.

Sometimes there are abuses. An uproar last month over a prank article
implicating a distinguished journalist in the Kennedy assassinations
caused Wikipedia to tighten up the rules a bit. But for the most part,
the utopian experiment has been a surprising success.

Wikipedia's rough-edged entries on science are often more detailed and
current than the ones in Britannica, which still credits Hwang Woo
Suk, the South Korean stem cell researcher accused of fraud, with
successfully cloning human embryos. But can I really be sure, as
Wikipedia tells me, that Dr. Hwang was born Dec. 15, 1952, when
Britannica insists on Dec. 15, 1953? The question is whether to trust
an encyclopedia that evolves like an organism or one that was designed
like a machine.

A study last month in Nature showed that the decision is far from
clear-cut. Calling on experts to compare 42 competing entries, the
journal counted an average of four errors per article in Wikipedia --
and three in Britannica. That is not much of a difference, and a look
at the details only adds to the anxiety. A fact is surely a fact, but
what constitutes an error can be as hard to pin down as a bead of
mercury.

A high school student looking for information on Dmitri Mendeleyev
(also spelled Mendeleev), the Russian chemist renowned for the
periodic table of the elements, would have learned from Wikipedia that
he was the 14th child in his family instead of the 13th surviving
child of 17 -- what Nature's reviewer, Michael Gordin, a Princeton
University science historian, said was one of 19 mistakes in the
article.

But it wouldn't have helped to defer to the competition: Dr. Gordin
gave Britannica a demerit for describing the chemist simply as the
17th child. It is an imprecision one might easily commit. Dr. Gordin
was surprised when I told him, in an e-mail message, that his own
book, "A Well-Ordered Thing: Dmitrii Mendeleev and the Shadow of the
Periodic Table," uses the same number. "That's curious," he said.
"I believe that is a typographical error in my book. Mendeleyev was
the final child, that is certain, and the number the reliable sources
have is 13."

These, he said, are in Russian, and they apparently were not consulted
by "The Norton History of Chemistry," by William H. Brock, which
like Wikipedia says there were 14 children, or "The Development of
Modern Chemistry," in which Aaron J. Ihde goes with 17. In his book
"Galileo's Finger: The 10 Great Ideas of Science," Peter Atkins, an
Oxford University chemist, says that the number, "according to one's
source," is 11, 14 or 17.

Wikipedia seems determined to try them all. Scrolling through the
various versions of its article -- more than 300 at last count
extending back to July 5, 2002 -- one can watch as the number
oscillates between 14 and 17, stopping briefly at 15 (with the
explanation from an anonymous editor that "a child was recentely
[sic] discovered to exist") then to 16 before returning to 14 again.

For several minutes on Nov. 10, before the vandalism was quietly
corrected, Mendeleyev was "the oldest of five hundred million
children," and in October some numbskull scrawled at the top of the
page, "IM COOL: IM DOING A REPORT ON DMITRI MENDELEEV AND YEA IM COOL
AND HES COOL." Three days later the graffiti was swabbed away.

After the Nature report, Wiki's entry, like the others deemed to have
flaws, was flagged at the top with a warning label ("This article has
been identified as possibly containing errors") and retreated
temporarily into the safety of imprecision -- Mendeleyev is "one of
many children of Ivan Pavlovich Mendeleyev and Maria Dmitrievna
Mendeleyeva (nee Kornilieva)" -- before adopting, in an act of faith,
Dr. Gordin's number, 13.

Britannica clings to 17, as it has apparently done since the online
article was reproduced from the 15th Edition, first printed in 1974.

Misstating the size of a 19th-century scientist's family is not
exactly a howler, but what about the other mistakes Nature enumerated?
Some were unambiguous -- Britannica's writing that the theory of the
strong nuclear force, called quantum chromodynamics, was formulated in
1977 instead of 1973, or Wikipedia's noting that the deadline for
receiving proposals for the Nobel Prize is Feb. 1 instead of Jan. 31.
(Again, the Wiki error was quickly fixed.)

But many of the purported blunders seem open to debate. Wikipedia was
wrong, one reviewer decided, when it said the embittering agent used
to denature ethanol is denatonium, instead of identifying it more
precisely as denatonium benzoate. But all a reader had to do was click
on the word to call up an entire article on the substance, which noted
that it also comes in the form of denatonium saccharide.

Britannica, on the other hand, was taken to task for writing that
"Croton (now Crotone, Italy)" was the home of the ancient Greek
mathematician Pythagoras and his number-worshiping cult. "The Italian
town is Crotona, not Crotone," a reviewer objected. But not so fast.

The name has appeared in history as Crotona and, for that matter,
Kroton (when it was part of Greece), but Crotene is the modern name.
In an ideal world the Britannica editors might have included these
etymological details. But at worst, this is an imperfection, and when
you start looking for those there is no end.

Just as forgivable are some of the sins of omission. Should an error
really have been scored against Britannica because its entry on Agent
Orange does not mention that there were also Agents Purple, Pink and
Green? There is always more you can put in an article, and part of the
editorial art is deciding what to leave out.

Whatever their shortcomings, neither encyclopedia appears to be as
error-prone as one might have inferred from Nature, and if Britannica
has an edge in accuracy, Wikipedia seems bound to catch up.

The idea that perfection can be achieved solely through deliberate
effort and centralized control has been given the lie in biology with
the success of Darwin and in economics with the failure of Marx.

It seems natural that over time, thousands, then millions of inexpert
Wikipedians -- even with an occasional saboteur in their midst -- can
produce a better product than a far smaller number of isolated experts
ever could.

Meanwhile the competition has some catching up to do. While Wikipedia
includes a good, balanced article on the history of Britannica,
Britannica has not a word to say about Wikipedia, as it rapidly
becomes one of the most significant phenomena on the Net.