“When in doubt, leave it out”

“When in doubt, leave it out” was Howard Simons’s instruction to Bob Woodward and Carl Bernstein when uncertainty remained about a story. It set a high bar for publication: doubt could mean postponing a report or pulling it at the last moment, whether the uncertainty concerned a single word or an entire series. The rule mattered not as a promise that errors would never occur, but as a standard against which the newsroom’s decisions—and their consequences—could be measured.

The rule and its editor

Simons, The Washington Post’s managing editor and its number-two editor, urged Bernstein to pursue a broader explanation for Watergate while remaining cautious about what the newspaper ultimately printed. He was willing to reason from fragments and sketchy information during reporting, but he distinguished that exploratory work from publication. His instruction named the possible scale of restraint explicitly: a word, phrase, sentence, paragraph, story, or series could be delayed or withheld if doubts persisted.

The rule therefore did not require reporters to stop investigating whenever evidence was incomplete. It placed the restraint at the point where a theory or lead might become a public assertion. That distinction allowed the reporters and editors to keep pursuing uncertain leads without treating uncertainty itself as proof.

How restraint worked in reporting

Woodward and Bernstein developed a working practice in which either reporter could act as a brake on the other. If one had doubts about the strength of their reporting, they told the editors they agreed to hold off, even when they privately disagreed; sometimes they held off without telling the editors anything. Before publishing their report connecting H. R. Haldeman to control of the campaign’s political-espionage funds, they reviewed the underlying evidence again. The story’s central new fact—that the President’s chief of staff was the fifth person to have controlled those funds—carried implications larger than the single factual addition, so they tested whether their largely invisible evidentiary foundation could bear the weight of the published claim.

The same discipline shaped choices about how to approach vulnerable sources. When the reporters considered contacting grand jurors, the newsroom had private misgivings about the legality and propriety of the plan. Their proposed procedure was limited: identify themselves, explain that an anonymous mutual acquaintance had suggested the juror might know something about Watergate, ask whether the person was willing to talk, and leave unless the juror volunteered information without prompting. The plan’s constraints show an effort to keep reporting from becoming pressure or inducement, even as the reporters pursued information that might otherwise remain inaccessible.

The need for caution was sharpened by the conditions inside the campaign organization. During evening visits, one employee trembled and pleaded with Bernstein to leave before others saw him; another said she wanted to help and burst into tears. The Bookkeeper later told Bernstein that people had already lied to the grand jury before it knew what questions to ask. Such accounts did not resolve what had happened, but they showed why a reporter could not treat silence, fear, or a source’s accusation as self-verifying evidence.

The cost of getting it wrong

The clearest counterexample was The Washington Post’s October 6 report about Alfred Baldwin’s alleged memo recipients. Bernstein and Woodward believed Baldwin had named Robert Odle and Timmons and had identified Sedam’s name; a Justice Department source confirmed details, and the newspaper published the account as an advance on the Los Angeles Times report. Weeks later, the reporters learned that the initial FBI report did not clearly establish whether the memos Baldwin had seen concerned wiretap conversations or routine security matters. They came to believe they were routine memos unrelated to wiretapping. The report had been rushed into print and was incorrect.

The mistake was not an abstraction about journalistic fallibility. Three men were wrongly accused on the front page of the newspaper read by their families, neighbors, and friends. Odle was nearly in tears when he complained to prosecutors, struggled to find work, and was later dismissed from a Department of Agriculture consulting position when his name continued to surface in the investigation. Timmons was dejected, and his wife urged him to leave the White House staff; he stayed only after a long conversation with the President. The episode gives the rule its practical force: when verification does not settle what a piece of evidence means, publishing the stronger interpretation can impose lasting costs on people the evidence does not justify accusing.

Limits and retrospective meaning

The reporting also exposed a boundary between caution and the pressures that could erode it. Hugh Sloan said his lawyers had advised him against public statements because they might be used against him in civil litigation. He had not, he said, been told outright to remain silent, but the message to keep the organization together had been clear. His account, alongside the Bookkeeper’s report that witnesses had already lied, presented a reporting environment in which the truth was obscured by both legal self-protection and organizational pressure. Simons’s rule could govern the newspaper’s handling of uncertain claims; it could not make a source free to speak or undo deception that had already shaped the evidence.

Months after the trial, Simons described his unease with The Washington Post’s position as a gradual process: the paper had entered the situation increment by increment, without the early legal consultation that had accompanied its Pentagon Papers decision. His reflection complicates any reading of the rule as a complete safeguard. Restraint had to be repeatedly exercised amid deadlines, investigative ambition, ethical uncertainty, and the accumulation of small decisions. “When in doubt, leave it out” remained a clear editorial test, but the reporting history showed why applying it required vigilance—and why a failure to pause could not be repaired simply by recognizing the mistake later.