Home' Technology Review : November December 2008 Contents 52 YEARS AGO
TECHNOLOGY REVIEW NOVEMBER/DECEMBER
On page 84, Simson L. Garfinkel
explores Wikipedia's episte-
mology and discovers that, far
from being the free-for-all its detractors
portray it as, the world's most popular ref-
erence is decidedly rigid. In its e ort to
ensure accuracy, Wikipedia relies entirely
on "verifiability," requiring that all factual
claims include a citation to another pub-
lished source (preferably online, prefera-
bly in English). As a result, Garfinkel
argues, "on Wikipedia, truth is received
truth: the consensus view of a subject."
There are, of course, times when the
consensus view and the truth align per-
fectly. The problem is how to determine
when this is the case. In an essay for TR
published in April 1956, the former head
of the MIT mathematics department,
H. B. Phillips, described one method for
doing so. Since Phillips was a mathemati-
cian, it's little wonder he appealed to the
laws of probability for a solution.
Here ... is the objective criterion determin-
ing whether we know or do not know. When
nearly all agree who claim to know, it is rea-
sonable to assume that the majority view is
correct. The answer may still be wrong, but if
a decision is necessary the probability of error
in such a case does not justify hesitation.
Although Phillips referred to his cri-
terion as "objective," it in fact takes an
agnostic attitude toward the truth. It
doesn't matter whether we have fully
established the truth of any given state-
ment, because if the relevant experts
are unanimous in their opinion, we can
proceed as if we had. However, Phillips
realized that unanimity on anything is
di cult to achieve, and that we are often
left to evaluate conflicting claims.
The problem is then what to do when
agreement is not practically unanimous. This
problem has been handled in several ways.
One method is to leave the decision to a dic-
tator. In primitive societies that was probably
not a bad solution, but one that is now com-
pletely obsolete. Another method is to leave
the decision to the intellectually superior.
When the experts are in substantial agree-
ment, as in science and engineering, that is
certainly the correct solution. But when there
is considerable di erence of opinion, there is
no evidence that the intellectuals supply any
better answers than ordinary people. ...
These methods all have a common defect,
namely, that they lead to a single solution,
and when the experts do not agree any single
solution is a matter of chance and therefore
probably wrong. Some would say there is as
much chance that such a decision would be
right as wrong. But this is not correct. The
choice is not one out of two, but one out of
many. It is as if one should say, "I don't know
how much two times three is, so I'll take a
chance and say it is seven." Such guesses are
almost certain to be wrong.
Phillips once again advocated a proba-
bilistic solution, reasoning in this case
that the more freedom people have to
answer a question or propose a solution
to a problem, the greater the
chances that any one of them
will have success.
If any single solution is proba-
bly wrong, the only way of increas-
ing the chance of success is to try
simultaneously a large num-
ber of solutions. The probability
of including a correct solution
will increase with the number of
choices, and will be greatest if each
individual makes his own choice. ...
When the proper course is
known, action can be directed by
rule or law. But when the proper
course is not known, each indi-
vidual should be free to go his own way to
provide the greatest diversity of action and
therefore the greatest probability that some-
body will be right.
This diversity of thought and action
is what Wikipedia has tried to harness
in building its vast and ever-expanding
knowledge base. By letting anyone con-
tribute, regardless of his or her creden-
tials, it runs the risk that absurdities,
inconsistencies, and misinformation will
flourish. But a free society, as Phillips
argued, must allow each of its members
"the privilege of being wrong."
The Privilege of Being Wrong
THEN AND NOW, WE FACE THE PROBLEM OF
DETERMINING WHAT IS TRUE.
By MATT MAHONEY
Technology Review (ISSN 1099-274X), Reg. U.S. Patent Office, is published bimonthly by the Massachusetts Institute of Technology. Entire contents ©2008. The editors seek diverse views, and authors opinions do not represent the official
policies of their institutions or those of MIT. Printed by Brown Printing Company, Waseca, MN. Periodicals postage paid at Boston, MA, and additional mailing offices. Postmaster: send address changes to Technology Review, Subscriber
Services Dept., PO Box 16327, North Hollywood, CA 91615, or via the Internet at www.technologyreview.com/customerservice. Basic subscription rates: $34 per year within the United States; in all other countries, US$47. Publication Mail
Agreement Number 40621028. Send undeliverable Canadian copies to PO Box 1051 Fort Erie, ON L2A 6C7. Printed in U.S.A.
COURTESY OF THE MIT MUSEUM
52 YEARS AGO IN TR
H. B. PHILLIPS: "It is presumptuous for any
one to take his own feeling of mental certainty
as final evidence that he is right."
Links Archive January February 2009 September October 2008 Navigation Previous Page Next Page