Anonymous Sources, Who Gets Them and What a Reader Should Demand

"A person familiar with the matter." Read that phrase aloud and notice how much work it asks of you. You are being invited to believe a claim from someone you cannot see, whose motive you cannot judge, whose access you cannot test, and who has, by the terms of the deal, nothing to lose if they are wrong. And yet some of the most important stories in American history rest on exactly that arrangement.

So we have a paradox. Anonymous sourcing is both indispensable and dangerous, both the whistleblower's shield and the leaker's alibi. The task is not to ban it or to swallow it whole. The task is to know when to trust it. That requires definitions, a little history and a short list of questions. Let us take them in order.

Definitions first

Journalists use several terms, and they are not interchangeable.

• On the record: the speaker is named and can be quoted directly.
• On background: the information may be used, but the source is described only generally, for example as "a senior official."
• On deep background: the information may be used, but not attributed to anyone at all.
• Off the record: the information may not be published, though it may guide the reporter's questions.

Different newsrooms define these terms in slightly different ways, which is already a problem. The reader sees only the final label, "a source close to the investigation," and has no idea which of these four deals produced it.

The claim I will defend has three parts. One, protecting a source's identity is sometimes morally required and is a legitimate tool of free-press journalism. Two, anonymity is also routinely used for reasons that have nothing to do with protecting anyone from harm: to push an agenda, to settle a score, to float a trial balloon. Three, a reader is entitled to judge the practice by the standards that the major newsrooms have published for themselves, and by five questions of the reader's own.

The steelman: why anonymity exists at all

Let me state the best case for the practice, because it is a strong one.

Some truths exist only inside institutions that punish people for revealing them. An employee who reports fraud at a company, a civil servant who reports waste at an agency, a soldier who reports misconduct in a unit, all of them may lose a career, or worse, if named. If the only acceptable story is one with a name attached, then the institution with the most power to retaliate also has the most power to suppress. Anonymity is the counterweight.

The American tradition of anonymity is old, and it is not shameful. The Federalist Papers were published under the pseudonym "Publius." The authors, Alexander Hamilton, James Madison and John Jay, argued the case for the Constitution in the New York newspapers in 1787 and 1788 without signing their names. The pamphleteers of the Revolution used pen names. In 1735 the printer John Peter Zenger was prosecuted in New York for publishing a newspaper critical of the royal governor, and he was acquitted by a jury; see John Peter Zenger and the Jury That Defended American Press Freedom in 1735 for the story. In 1995 the Supreme Court, in McIntyre v. Ohio Elections Commission, struck down an Ohio ban on anonymous campaign literature and described anonymity as an honorable tradition of advocacy and dissent. So the principle that a person may speak without giving a name is not a loophole. It is part of the inheritance.

That is the steelman. Now to the matter of how it goes wrong.

Mark Felt and the Watergate model

The most famous anonymous source in American history is the one nicknamed "Deep Throat." Between 1972 and 1974, Bob Woodward and Carl Bernstein of the Washington Post reported the story of the Watergate break-in and the cover-up that followed. Woodward's confidential source guided him through the investigation. For over thirty years nobody outside a small circle knew who it was. In 2005 the source disclosed himself: W. Mark Felt, who had been the number two official at the FBI during the Watergate years. The disclosure came in a Vanity Fair article by his attorney in May 2005, and Woodward, Bernstein and the Post confirmed it.

Notice what made the Watergate arrangement defensible. The source was real, was in a position to know, and was talking about misconduct inside the highest office in the land. Notice also what made it work as journalism. The Post did not run the story on his word alone. The paper's editors, as the Watergate story is usually told, pressed the reporters on their claims. The anonymous source was a signpost that pointed them to documents and witnesses. He was not the story. He was the lead.

And notice the complication that historians point out. Felt was not a pure whistleblower. He was a senior official with his own interests inside the Bureau, and the motives of anyone who leaks are rarely single. That does not make what he said false. It means the reader of any anonymous account should ask why the person is talking, a point we will return to.

Judith Miller, Valerie Plame and the case where anonymity went wrong twice

If Watergate is the best-case story, the episode of 2003 to 2005 is the cautionary one, and it is useful precisely because it cuts in two directions.

In July 2003, the columnist Robert Novak published a column that identified Valerie Plame as a CIA employee. Her husband, former ambassador Joseph Wilson, had publicly criticized the administration's case for the Iraq war. A criminal investigation followed to find who had disclosed her identity. Reporters were subpoenaed. Judith Miller of the New York Times, who had gathered material but never wrote a story about Plame, refused to identify her source and was jailed for contempt in 2005, serving 85 days before her source released her to testify. The source turned out to be I. Lewis "Scooter" Libby, the Vice President's chief of staff, who was convicted in 2007 of perjury and obstruction. Years later it emerged that Novak's own original source had been Richard Armitage, a State Department official, a fact Armitage acknowledged in 2006.

Two lessons come out of this. First, there is a genuine constitutional question about whether reporters can be forced to give up sources. In Branzburg v. Hayes (1972) the Supreme Court held, by a vote of five to four, that the First Amendment does not excuse a reporter from answering a grand jury's questions. There is no federal shield law, though many states have passed their own. The legal protection for confidential sources is therefore thinner than most readers assume, which is one reason newsrooms treat promises of confidentiality with such gravity.

Second, and this is the harder lesson, the same reporter had been a prominent channel for anonymous claims about Iraqi weapons of mass destruction before the war. In May 2004 the New York Times published an editors' note acknowledging that some of its coverage had not been as rigorous as it should have been, and that information from anonymous sources, some of it from Iraqi exiles with their own agendas, had not been adequately challenged. That is a newspaper confessing, in its own pages, that its anonymous sources had been used to push a case for war. Think about that. The same tool that shielded a whistleblower in Watergate carried an official line into print in 2002 and 2003. The tool did not change. The judgment applied to it did.

This is the point I would press hardest. Anonymity is not a virtue. It is a risk transferred from the source to the reader. The reporter takes the source's word. The reader takes the reporter's word about the source. At every link, accountability thins. That is acceptable only when the story is important enough, the source is well-placed enough and the verification is strong enough to justify it.

What the newsrooms promise themselves

Here is something many readers do not know. Major American news organizations have published written standards for anonymous sourcing, and the standards are remarkably similar in outline. I will describe them as a category, because the exact wording varies and changes over time, and you can read the current text on each outlet's own site.

The common elements, in general terms, are these. Anonymity is meant to be a last resort, used only when the information is important and cannot be obtained on the record. The reporter is expected to know the source's identity and, in most newsrooms, an editor must know it too. The source must have direct knowledge, not merely have heard something. The story should explain, as far as it can without revealing the person, why the source is being granted anonymity and what position the person holds. And the paper should not use anonymity to let sources make personal attacks or take cheap shots at named people.

Whatever the details, the architecture of the standards implies a test: the standards themselves tell you what to look for. If a story grants anonymity for an anodyne comment that anybody could have made on the record, the newsroom has violated its own rule. If the description of the source is so vague that it could cover a thousand people, the newsroom has not met its own duty to tell the reader why this person is credible. And if the anonymous voice is used to attack a named individual, the newsroom has ignored the rule most clearly written. A reader who holds the outlet to its own standards is not being unreasonable. That is the same standard the outlet claims.

Five questions to ask whenever a story rests on "people familiar with the matter"

Here is the numbered method. Put these on an index card.

1. How close is the source? "A person who attended the meeting" is a firsthand witness. "A person briefed on the meeting" is secondhand. "A person familiar with the matter" could be anyone from a participant to someone who read a memo to someone who heard a rumor on the elevator. Demand to know which, and treat vague descriptions as weak evidence.
2. Why is the source talking, and why anonymously? Whistleblowers risk their livelihoods, and their anonymity has an obvious reason. A senior official who wants to shape a story, discredit a rival or test public reaction has a different reason. If the story does not say why the person needed to be hidden, assume the reason may be convenience.
3. Is there anyone on the record who confirms it? The strongest anonymous story has named confirmation somewhere, a document, a filing, a recorded statement or a named witness. If the whole structure rests on unnamed people, the whole structure is as strong as its weakest unnamed person.
4. Does the story give the other side a chance to answer? A story that makes a damaging claim from anonymous sources and offers the target a one-line "did not respond to a request for comment" has not tested its claim. Look for a real answer from the accused, and look for whether the story engaged it.
5. What happens if it turns out wrong? Check whether the outlet has a record of correcting anonymous-source stories openly. The best test of a newsroom's seriousness about anonymity is what it does after an anonymous claim fails. A paper that quietly edits and moves on has told you something. A paper that publishes a prominent correction has told you something else. The corrections page is where this shows; see The Corrections Page Is the Most Honest Part of the Paper, So Read It.

Those five questions will not tell you whether a story is true. They will tell you how much weight to place on it before you know, which is the useful skill in a world where most stories arrive before the verification does.

The objection from the whistleblower

The strongest objection to everything above deserves a fair hearing. It goes like this: you are asking readers to discount anonymous reporting, but anonymity is the only thing standing between the truth and retaliation. Every time a reader says "show me a name," a powerful institution smiles. If the public demands names, the powerful will make sure that nobody gives them.

That is a serious point and I do not wave it away. A culture that distrusts all anonymous claims will end up hearing only what the powerful allow. The Watergate story, the many accounts of waste and abuse that came out through confidential channels and stood up in court, all of these would have died in a world of mandatory names.

But the objection answers a question I did not ask. I am not proposing that anonymity be banned or that reporters be forced to reveal sources. I would defend a reporter's right to protect a source against a prosecutor's demand, and I think Congress and the states are right to write shield laws. The question is how a reader should weigh the result. A reader is entitled to be skeptical and still support the practice. Weighing is not banning. The whistleblower's protection depends on the system being credible, and the system is credible only if the public can tell the verified leak from the planted one. A press that wants trust has to help the reader make that distinction, and the five questions are the reader's side of that bargain.

There is a second reply. The same institutions that benefit from loose anonymity are not only the powerful bureaucracies the whistleblower fears. They include the newsrooms themselves, which can use "a source" to float opinions that would not survive a name. Skepticism about anonymity protects the public from both.

The conservative stake in this

Conservatives have a particular reason to care. We have watched the same arrangement serve different masters at different times. Anonymous sources have carried stories that damaged public figures on the right, some accurate and some not, and have carried official lines on the left and right alike when administrations wanted to be heard without being held to account. The principle is not partisan. It is that power should answer for its claims.

It also connects to a Founding idea. The case for the Constitution was made by "Publius," but the argument was open to inspection, signed by its logic, and answerable by anyone who read it. That is the model: an anonymous speaker attached to a transparent argument. The danger arises when an anonymous claim of fact arrives with no way to check it, because then the reader has no logic to inspect, only a trust to extend. The Federalist's author hid his name but not his reasoning. The anonymous leaker hides both.

For the broader pattern of how coverage can lean without lying, read Seven Words Reporters Use to Tilt a Story Without Telling a Single Lie and How to Read a News Story From the Bottom Up and Catch What the Lede Hides. For what happens when the whole structure collapses, see The Night a Network Bet Its Reputation on Memos Typed Like a Word File and The Laptop Story Twitter Blocked and the Letter That Gave It Cover.

A worked example, clearly hypothetical

Here is an invented case, labelled as such, so you can practice. Picture a headline: "Agency head plans to resign, sources say." The story cites "two people familiar with the matter," offers no named confirmation, says the agency "did not immediately respond" and runs on a Friday evening.

Apply the questions. Source closeness: unknown. Motive: possibly a rival, possibly a staffer, possibly someone wanting a reaction. Named confirmation: none. Other side's answer: none yet. Track record: unknown. Verdict: interesting but unconfirmed, and you should wait for Monday. That is not cynicism. It is the correct amount of belief for the evidence given.

Now imagine the same story with one change: the second paragraph quotes a named board member, who confirms a meeting on the subject. The weight shifts at once. The anonymous sources are now corroboration rather than foundation. You can see how the reader's confidence is built, brick by brick, from things you can check.

The case in brief

1. Anonymous sourcing is a legitimate and sometimes necessary tool, rooted in the American tradition from Publius to McIntyre.
2. It transfers risk from the source to the reader, so it should be a last resort and should be corroborated.
3. History shows both sides: Mark Felt and Watergate on one hand, the Iraq-era weapons reporting and the Plame episode on the other.
4. Newsrooms publish standards for anonymity, and a reader can hold them to their own rules.
5. Five questions, covering closeness, motive, corroboration, the other side's answer and the correction record, tell you how much weight to give a story.

What you can do: the next time a headline rests on unnamed sources, write down the five questions and answer them before you share the story.