The Corrections Page Is the Most Honest Part of the Paper, So Read It

Pop quiz. Name the part of a newspaper that is written by people who are admitting they were wrong, in public, in plain English, with their own bylines on the line. It is not the opinion page, where certainty is the whole business. It is not the front page, where confidence is the house style. It is a small box tucked away near the back, and nobody reads it. I am here to tell you that it is the single most informative square inch in the building.

Here is my claim for this post, and I will give it to you straight. A news outlet's habits about its own errors tell you more about whether to trust it than any rating, any chart of left and right, and any slogan about fairness. Every outlet makes mistakes. That is not the question. The question is what it does next. By the end of this post you will know the vocabulary, the warning signs, a few famous examples and a method you can use in ten minutes.

Four words that mean four different things

Newsrooms use a handful of terms, and most readers lump them together. Do not. Each one tells you something different about what happened.

A correction. The outlet states that something it published was factually wrong and gives the right information. A name was misspelled, a date was wrong, a number was off by a factor of ten, a person was quoted as saying something he did not say. A correction is the most ordinary admission. It says: we got a fact wrong, here is the fact.

A clarification. The original was not exactly false, but it was ambiguous or misleading, and the outlet is making it clearer. A sentence that could be read two ways. A figure that was accurate but lacked a base. A clarification is a signal that the writer's meaning and the reader's reading did not match.

An editor's note. This is an addition of context that is not strictly a fix of an error. It might disclose a relationship that should have been mentioned, explain that a story was changed after publication, or tell readers that the outlet no longer stands behind part of the reporting. Editor's notes vary enormously. Some are brief and honest. Some are long, defensive essays. Their length and tone are themselves information.

A retraction. The strongest word. The outlet withdraws the story, or a major part of it, because the foundation turned out to be unreliable. A retraction means that the thing should not have been published, not merely that one detail was wrong. Retractions are rare and, when handled well, remarkably candid.

Put them on a ladder. A correction repairs a detail. A clarification repairs a meaning. An editor's note repairs the context. A retraction removes the whole thing. A reader who sees the same outlet issuing only the first and second kinds should ask a harder question: is it really that precise, or is it that it does not apply the stronger words when it should?

Why a corrections page tells you about an outlet

There are three reasons the corrections page is worth reading. I will count them.

1. It is a record of the outlet's actual error rate and error types. If an outlet's corrections are mostly typos and misspellings, that is a good sign: the errors that get through are small. If they are about the substance of its claims, such as who did what, what a law says or what a number means, that is a different outlet.
2. It shows what the outlet considers worth fixing. Some outlets correct a wrong initial and never correct a misleading framing, because a framing is not a "fact." Others will correct an omission. The categories of error that an outlet acknowledges mark the edges of its standards.
3. It shows the speed and the prominence. A correction published within hours, with a visible note, is a different animal from one appended at the bottom six weeks later. The Society of Professional Journalists' code of ethics asks journalists to acknowledge mistakes and correct them promptly and prominently. Promptly and prominently: two adverbs, and you can test an outlet against both.

I want you to notice that this is a standard that applies to every outlet, from the oldest national paper to the newest conservative website. It is one of the few tests of trust that is not partisan. A site that is on your side and never corrects anything is not more trustworthy than one you dislike that corrects often. It is less.

The silent edit

Now we come to the part that should concern you more than any correction: the change nobody told you about.

On paper, a mistake in a printed paper stays in the printed paper, and the correction appears the next day. The record exists in a thousand attics. On the web, an article is a living thing. The writer or the editor can change a headline, a paragraph or a number at any time, and unless the outlet adds a note, the reader who comes later sees only the new version. The old version evaporates. Journalists sometimes call this a "stealth edit" or a "silent edit."

To be fair, not every silent edit is sinister. Fixing a typo does not need a note, and a headline may be rewritten for style. Most outlets have policies that distinguish a trivial change from a substantive one and add a note for the second kind. The problem is that readers cannot tell which kind they are looking at. A change in the number of people killed, the name of the suspect, the claim about what a law does, or the identity of a source, is substantive, and it deserves a note. When it does not get one, the outlet has altered the record without telling the public.

Here is the practical danger. A story is posted with a dramatic headline and shared a hundred thousand times. Two hours later the headline is quietly softened because the body did not support it. The hundred thousand shares remain with the original. The correction, if there is one, reaches a fraction of the audience. This is how a false impression survives its own refutation.

How the web remembers

Fortunately, the web has a memory, and anyone can use it. The Internet Archive's Wayback Machine, a nonprofit project that has been saving copies of web pages since the mid-1990s and offering them to the public since 2001, stores snapshots of millions of pages. If you know a page's address, you can often see how it looked at different times and compare versions side by side. Other archiving services, which let a user save a page on demand, work in a similar way. If you are about to share a story that makes a serious claim, saving a copy first takes a few seconds and protects you if the story changes.

There have also been volunteer projects built on this idea. Some tracked the changes to articles at major outlets automatically, showing readers what had been rewritten between versions. Those projects came and went with the technology, and I will not pretend that any of them is a standing service you can rely on. The point is that the method exists, and that you can do a version of it by hand. When you are reading a claim that matters, compare the current page with an archived copy, and look at the headline first.

Some famous retractions, read for what they teach

History offers some instructive examples, and the instructive thing is how the outlet handled it.

The Washington Post, 1981. After the paper learned that a Pulitzer-winning feature was fabricated, it returned the prize and published an account of how it happened. That is the model of a full public retraction: it names the failure, takes back the honour, and shows the work. The story is told in Fabricated From the Byline Up, Four Scandals That Shamed Elite Newsrooms.

The New York Times, May 2004. In a note from its editors, the paper looked back at its coverage of the run-up to the Iraq war and said that some of the reporting had not been as rigorous as it should have been, and that it wished it had been more aggressive in re-examining claims as new evidence emerged, or failed to emerge. That is an editor's note, not a retraction of any single article. It is worth reading because it is a rare case of a paper reviewing its own pattern of coverage, not just one story. It also shows the limit of the form: the note came after the war had begun.

Newsweek, May 2005. The magazine published a short item reporting, on the word of an unnamed official, that U.S. interrogators had desecrated a Koran at Guantanamo, and after violent protests abroad and a challenge from the government, the magazine retracted the story a week later, saying that, based on what it now knew, it was retracting its report that an internal military investigation had found Koran abuse. A retraction under pressure, and an illustration of how one anonymous source can carry a whole story. See Anonymous Sources, Who Gets Them and What a Reader Should Demand.

Rolling Stone, 2015. The magazine published a long account of an alleged assault at a university fraternity in November 2014. After other outlets raised questions, the magazine asked the Columbia Graduate School of Journalism to review the work. The school's report in April 2015 found a failure of basic reporting practices, and the magazine retracted the story. An outside review with a published report is the gold standard of retraction, whatever the subject of the story.

CBS, 2004. The network that aired memos about a presidential candidate's military service later said it could not vouch for them, and commissioned an independent panel. The Night a Network Bet Its Reputation on Memos Typed Like a Word File tells that story.

Look at what these have in common. In each case the outlet eventually did the right thing, publicly. In each case it took outside pressure, and in each case the delay was part of the damage. An outlet that retracts promptly on its own initiative is rarer than any of these, and when you see one, you have seen a good outlet.

The steelman, and the answer

Here is the best objection to everything I have said. Corrections pages are marketing. They are small, curated, and often written by the same people who made the error. An outlet can run a spotless corrections page by simply never admitting a mistake, and it can look scrupulous by correcting trivia while leaving the framing untouched. Some of the most biased coverage in the country corrects spelling errors with great diligence. So a clean record proves little.

I accept all of that, and I accept it partly because it is true. A corrections page is not a measure of fairness. It cannot catch what the outlet chooses not to see: the story it did not cover, the word it chose, the source it never called. The tools for that are different. See Seven Words Reporters Use to Tilt a Story Without Telling a Single Lie and Why the Fact-Check Became Its Own Genre and How to Check the Checkers.

But the objection proves less than it seems. A corrections page is a floor, not a ceiling. An outlet that cannot meet the floor, that does not correct its own factual errors promptly and visibly, has told you what its standards are. An outlet that clears the floor has not proved itself fair, but it has passed the first test. And there is one more thing the page can show. The pattern of what an outlet corrects reveals what it believes it owes the reader. When corrections cluster around one subject or one side of the political line, that clustering is data. Read ten of them and you will see it.

A ten-minute method

Here is what you can do with this.

1. Find the page. Most major outlets have a corrections page or a standards page, often linked at the bottom of the site. If you cannot find one, that is information.
2. Read the last twenty entries. Sort them: typos, numbers, names, claims. How many involve substance?
3. Check the timing. Does it say when the original ran and when the fix was made? Gaps of days or weeks on a serious claim are a bad sign.
4. Look at the placement. Is the fix at the top of the article, where readers will see it, or at the bottom where they will not?
5. Test one story. Take a story you saw shared widely, look up its archived version, and compare it with the current page. Did anything change without a note?
6. Repeat for an outlet on your own side. A tool you apply only to your opponents is a weapon, not a tool.

The media watch page compares how outlets across the spectrum headline the same story, which gives you a starting point: when the headlines on one story diverge, go to each outlet's page and see whether the differences survive. For a related skill, see How to Read a News Story From the Bottom Up and Catch What the Lede Hides.

The Founding yardstick

The Founders built a government on the premise that rulers will make mistakes, and that the remedy is not to hope for angels but to build ways of correcting error. James Madison wrote in Federalist No. 51, "If men were angels, no government would be necessary." The sentence is usually quoted about government, but it applies to the press with equal force. If reporters were angels, no corrections would be needed. They are not, and neither are we. What we can ask of any institution, public or private, is that it admits error openly and fixes it, and that it has a structure that makes the admission routine.

Seen this way, the corrections box is a small piece of constitutional machinery. It is the press's own system of checks. A newsroom with a good one is practicing self-government in miniature. One without it is asking for the kind of trust that free people should not give to any institution that will not say it was wrong.

I will add one joke, since you have been patient. The next time you hear a politician say that his critics are "relentlessly negative," ask what his corrections page looks like. Public servants do not publish one. If they did, it would be the most-read page in Washington, mostly by people checking whether the second entry says the opposite of the first.

The case in brief

1. A correction fixes a fact, a clarification fixes a meaning, an editor's note adds context and a retraction withdraws a story, and the ladder tells you how serious the failure was.
2. An outlet's handling of its errors, promptly and prominently, is a non-partisan test of trust that every outlet can be held to.
3. Silent edits to online stories change the record without telling readers, and web archives let you compare versions to catch them.
4. Famous retractions, from the 1981 Pulitzer to the 2005 magazine item to the 2015 outside review, show that the best outlets correct in public and the worst correct only under pressure.
5. A corrections page is a floor, not a ceiling; it cannot catch what an outlet never covers, so use it alongside the other tools.

What you can do: open the corrections page of the outlet you read most, read its last twenty entries, and note how many concern substance and how quickly each was fixed.