The Times has no official anonymous sourcing rules—no "two source rule" or anything like that&mdashp;but they do often pass around internal documents to serve as guidelines. The paper’s public editor, Clark Hoyt, recently recruited some Columbia J-school students and had them take some sample papers from 2004—when Al Siegal and Bill Keller last sent out an anonymous source guideline sheet—and compare them to 2007, and they found that anonymous sources had basically dropped by half at the paper.
But reporters shouldn’t feel afraid to use them! Mr. Keller sent out a "refresher course" memo last night. There are helpful tips like: "There is, on the face of it, something ludicrous about a government or corporate "spokesman" insisting on not being identified by name; we should push such sources to speak for attribution."
Mr. Keller acknowledges the paper’s occassionally awkward insistence that they explain why some sources are not named (for instance! "Mr. Smith couldn’t speak today out of fear of offending those he works for"), though doesn’t offer a suggestion how to work around it. "It is sometimes helpful to inform the reader why we have allowed the source to be anonymous. Sometimes the reason is so obvious that stating it feels silly, but sometimes the answer is interesting and may help the reader appraise the source’s motivation. We should do it deftly, without falling into formulaic expressions that don’t mean much."
As part of his ongoing campaign to hold us to our own standards,
Clark Hoyt recently enlisted friends at the Columbia J-School to gather
data on our use of anonymous sources. The results present an excellent
opportunity to remind ourselves that unnamed sources are not to be used
lightly, and to reiterate the rules governing their use.
For this project a group of grad students under the tutelage of
Richard Wald compared newspapers on a sampling of days before and after
February 2004, when Al Siegal and I distributed guidelines on the use of
anonymous sources. The students read every article in every section, noted
and categorized every piece of information that was not attributed, and
then attempted to draw some conclusions.
By their count, the use of anonymous sources in The Times dropped
significantly — by about half — between 2004 and 2007, when they took
their samples. They also found many instances where our use of unnamed
sources fell short of our own standards, or rather their interpretation of
our standards. My own view is that, like any attempt to measure what is
essentially a matter of judgment, the report is more valuable for focusing
attention on an important subject than for its metrics. (The study, for
example, invents an "adequacy" test for description of anonymous sources
that is significantly different from my own.) When the Columbia data has
been posted on the CJR website, we will link to it, spread sheets and all,
so that you and Times readers can explore. Clark plans to make it the
subject of his column Sunday. After that, Jill Abramson has volunteered to
field questions on the subject from readers, for whom the use of anonymous
sources is a matter of keen interest.
I’d like to thank the Columbia students for what sounds like
amazingly tedious labor, and for provoking a useful update on an important
What follows is a quick review of our standards. The original memo is
available on the Newsroom Navigator page.
— The ability to offer protection to a source is an essential of our
craft. We cannot bring readers the information they want and need to know
without sometimes protecting sources who risk reprisals, firing, legal
action or, in some parts of the world, their lives when they confide in us.
You will occasionally hear an editor suggest that news organizations should
foreswear anonymous sources altogether. This is high-minded foolishness.
Without the option of protecting sources, with recourse only to an
increasingly redacted public record, the coverage of government and other
powerful institutions would tend more and more toward press-conference
stenography. Many important stories would not be reported at all. To pick
one current example: the humanitarian crisis in Myanmar, where a zealously
secretive regime requires us to withhold the identity not only of our
sources but of our reporters in the field.
— But, as a general rule, stories based on unnamed sources
generally are less convincing than those based on named sources and
documents. And the cumulative reliance on unnamed sources can erode our
credibility. Anonymity should not be granted casually. It should not be an
automatic or assumed transaction between reporter and source. We expect
reporters to press sources to stand by their statements, on the record.
— When we agree to protect the identity of a source, we owe readers
as much information as we can supply to help them judge whether the source
is credible. That means we want to tell readers how the source was in a
position to know. (Was he in the room? Did she read the document?) And it
means we want to inform readers whether the source had a bias or an ax to
grind. Sometimes a source’s affiliation tells you most of what you need to
know. There is, on the face of it, something ludicrous about a government
or corporate "spokesman" insisting on not being identified by name; we
should push such sources to speak for attribution. But at least in the case
of an unnamed "spokesman" the reader is left in little doubt about the
person’s point of view. We can assume the person is passing along the
viewpoint of the agency/company/institution he or she is paid to represent.
— We avoid letting sources hide behind anonymity for the purpose of
partisan or personal attack. If such opinions are worth reporting and
cannot be specifically attributed, they may be paraphrased or described
after consultation with an editor. But, as the 2004 memo put it, "The vivid
language of direct quotation confers an unfair advantage on a speaker or
writer who hides behind the newspaper…" This is not to say that we
proscribe all opinions expressed by anonymous sources. One of the critical
elements of our reporting about the NSA’s warrantless eavesdropping program
was the opinion of lawyers inside the government that the program might be
— Quantity is not the same as quality, which is why we do not have a
"two source rule" or a "three source rule." One actual participant in an
event may be better than three people who heard about it third-hand, or
from one another. One neutral witness may be more valuable than a crowd of
— It is sometimes helpful to inform the reader why we have allowed
the source to be anonymous. Sometimes the reason is so obvious that stating
it feels silly, but sometimes the answer is interesting and may help the
reader appraise the source’s motivation. We should do it deftly, without
falling into formulaic expressions that don’t mean much.
— In a case where an article depends heavily on anonymous sources
who have a common reason to insist on anonymity, a well-crafted sourcing
paragraph — conveying the number and quality of sources and the reason
they refused to be identified — may be more helpful to readers than
repeating the same sourcing
boilerplate over and over.
— At least one editor must know the identify of a source. With that
knowledge comes the responsibility to enforce the rules listed above — to
question whether the anonymity is justified, to assure that we’ve told the
reader all we can about the credibility of a source. By the way, Craig
Whitney, who does regular spot checks, said he has rarely come across an
instance where an unnamed source was not known to an editor.